3470

IN THE UNITED STATES DISTRICT COURT

FOR THE DISTRICT OF COLUMBIA

UNITED STATES OF AMERICA, ET AL., )

)
Plaintiffs, )
) CV No. 20-3010
vs. ) Washington, D.C.
) October 2, 2023
GOOGLE LLC, ) 9:15 a.m.
)
Defendant. ) Day 14
___________________________________) Morning Session

TRANSCRIPT OF BENCH TRIAL PROCEEDINGS

BEFORE THE HONORABLE AMIT P. MEHTA
UNITED STATES DISTRICT JUDGE
 3471

APPEARANCES:

For DOJ Plaintiffs: Kenneth M. Dintzer

U.S. DEPARTMENT OF JUSTICE
1100 L Street, NW
Washington, D.C.
(202) 307-0340
Email:
[email protected]

Adam T. Severt
DOJ-ATR
Antitrust Division
450 Fifth Street, NW
Suite 7100
Washington, D.C. 20530
(202) 307-6158
Email: [email protected]

For Plaintiff
State of Colorado: Jonathan Bruce Sallet
COLORADO DEPARTMENT OF LAW
Consumer Protection Section,
Antitrust Unit
Ralph L. Carr
Colorado Judicial Center
1300 Broadway
Suite 7th Floor
Denver, CO 80203
(720) 508-6000
Email: [email protected]

William F. Cavanaugh, Jr.

PATTERSON BELKNAP
WEBB & TYLER LLP
1133 Avenue of the Americas
Suite 2200
New York, NY 10036-6710
(212) 335-2793
Email: [email protected]
 3472

APPEARANCES CONTINUED:

For Plaintiff
State of Nebraska: Matthew Keith McKinley
NEBRASKA
DEPARTMENT OF JUSTICE,
OFFICE OF ATTORNEY GENERAL
Consumer Protection Bureau
2115 State Capitol Building
Lincoln, NE 68509
(402) 471-2683
Email:
[email protected]

For Defendant Google: John E. Schmidtlein

Benjamin M. Greenblum
WILLIAMS & CONNOLLY LLP
680 Maine Avenue, SW
Washington, D.C. 20024
(202) 434-5000
Email: [email protected]
Email: [email protected]

For Third Party Defendant

Microsoft Corporation: Matthew Lloyd Larrabee
Dechert, LLP
1095 Avenue of The Americas
27th Floor Mailroom
New York, NY 10036
(212) 698-3671
Email:
[email protected]

Court Reporter: William P. Zaremba

Registered Merit Reporter
Certified Realtime Reporter
Official Court Reporter
E. Barrett Prettyman CH
333 Constitution Avenue, NW
Washington, D.C. 20001
(202) 354-3249

Proceedings recorded by mechanical stenography; transcript

produced by computer-aided transcription
 3473

- - -

WITNESS INDEX

- - -

WITNESSES DIRECT CROSS REDIRECT RECROSS

PLAINTIFF's:

SATYA NADELLA 3486 3539

SATYA NADELLA 3533
- - -

INDEX OF EXHIBITS

- - -

PLAINTIFF'S ADMITTED

UPX0736 3527

- - -

INDEX OF EXHIBITS

- - -

DEFENDANT'S ADMITTED

DX424 3552

DX423 3554

DX500 3562

DX500A 3562

DX451 3590

DX456 3594
 3474

1 P R O C E E D I N G S

2 COURTROOM DEPUTY: All rise. The Honorable

3 Amit P. Mehta presiding.

4 THE COURT: Good morning. Please be seated,

5 everyone.

6 COURTROOM DEPUTY: Good morning, Your Honor.

7 This is Civil Action 20-3010, United States of America,

8 et al., versus Google LLC.

9 Kenneth Dintzer for the DOJ Plaintiffs.

10 Jonathan Sallet and William Cavanaugh on behalf of

11 Plaintiff States.

12 John Schmidtlein on behalf of Google.

13 THE COURT: All right, Counsel, good morning,

14 everyone. I hope everybody had a nice weekend.

15 All right. So we've got some scheduling issues to

16 address this morning. Mr. Dintzer, do you want to start

17 things off.

18 MR. DINTZER: Yes, sir.

19 Thank you, Your Honor. If I could just take a

20 moment to hand up some paper that might help us.

21 THE COURT: Sure.

22 MR. DINTZER: Thank you, Your Honor. May it

23 please the Court.

24 So there are -- we had two significant witnesses

25 become unavailable for this week, and so I want to make this

 3475

1 clear. This is not a time problem, we have enough time for

2 our witnesses, this is a logistics problem to make sure we

3 get the witnesses in front of the Court when we have some

4 time this week.

5 So what we have decided to do is move up part of

6 our expert presentation, Professor Whinston, for the issues

7 of market definition and monopoly power.

8 THE COURT: I'm sorry, what's the second one?

9 MR. DINTZER: Monopoly power.

10 THE COURT: Okay.

11 MR. DINTZER: This exactly tracks -- we had one

12 witness take care of all the economics. Google had two,

13 Dr. Israel is doing the same two issues, definition and

14 power, and as well as some small things, and then Professor

15 Murphy is doing the rest for them. So what we're doing

16 simply is asking to break up Professor Whinston so he does

17 basically the Israel portion this week and then he'll do the

18 rest of it in two weeks when we had it scheduled on our

19 calendar.

20 We don't see any prejudice. Google said, just

21 last Friday, that they were expecting us to complete our

22 case this week, which would have meant Professor Whinston

23 would have had to do his entire thing this week. So asking

24 them to do two pieces of it, they clearly were prepared to

25 do more, so we can't see any prejudice there.

 3476

1 And quite honestly, so this makes sure that we

2 don't have downtime this week, or at least if we do, it's

3 just a very small amount and also, quite honestly

4 Your Honor, it's a bit easier on the Court's ears to not

5 have to sit through the whole chunk of Professor Whinston as

6 opposed to an opportunity to break it up.

7 THE COURT: I love to hear from economists.

8 MR. DINTZER: We project about five hours of

9 Professor Whinston's testimony. So this allows us to do it

10 in probably about a two-and-a-half-hour chunk. And so

11 that's what we're asking.

12 To be able to do it, I mean, we believe Google

13 should cross. And significantly, we need to be able to talk

14 to him, after he's finished his direct and cross and

15 redirect so that we can prepare him for the rest of it to

16 come in two weeks.

17 And we were surprised that Google found any of

18 this problematic. They actually were the ones arguing for,

19 in situations where a witness' testimony is broken up, that

20 there by able to be communications, but now they're -- if

21 the Court will remember just a few weeks ago. And so now

22 they're saying we shouldn't be able to do that, which

23 doesn't really make any sense to us, so that's it. We want

24 to move --

25 So if you look at the two items I just handed out,

 3477

1 the first thing just shows you where we've been, because

2 Google likes to suggest we haven't been making any progress,

3 and actually we've been ripping through our case and getting

4 the witnesses that the Court needs to see and we'll do that

5 again today.

6 The second thing shows what the calendar would

7 look like with Professor Whinston fitting in for the time

8 this week, and we have it -- it looks about three hours for

9 us and then Google wouldn't tell us how long they need for

10 cross, but it looks to be about two hours.

11 Without bringing Professor Whinston this week,

12 there will be some downtime this week, and that may not be a

13 problem to the Court, the Court may appreciate that, but we

14 just -- we don't want to take it upon ourselves to create

15 that kind of a bubble.

16 THE COURT: Sure. Okay.

17 All right. Mr. Schmidtlein, let me just hear from

18 you, and then we can figure this out and then move forward.

19 MR. SCHMIDTLEIN: Thank you, Your Honor.

20 I guess a couple of observations. One of the --

21 over the weekend, Mr. Cavanaugh had asked us to bring

22 forward two witnesses who I think were scheduled for last

23 week.

24 THE COURT: Right.

25 MR. SCHMIDTLEIN: We have agreed to do that.

 3478

1 On Mr. Dintzer's calendar here, I only see one of

2 them being brought forward, not the second one, which

3 I thought we had agreement that we were going to bring

4 forward the second one that would help ameliorate some of

5 these concerns, Mr. Varian, who you see listed on the week

6 of October 10 through 12 that the States were going to call.

7 So we're trying to do that.

8 We do think Professor Whinston should testify

9 together in one chunk. That is, I think, the way we've

10 always been presuming the way this case was going to go.

11 And, again, we're the ones who are a little bit surprised to

12 find out that now, because of the delays that have been

13 occasioned, we can cast aspersions back and forth, but the

14 fact of the matter is, they are behind, they haven't ripped

15 through all the witnesses that they were supposed to rip

16 through, and, you know, now we're going to let

17 Professor Whinston basically spread out his final exam over

18 three weeks, and we don't think that's appropriate. And we

19 don't think they should be allowed to continue talking with

20 him while he is "still on the stand," which is what he would

21 be.

22 I suggested, if we got laid over on a weekend and

23 I had a fact witness who we were going to be doing a major

24 direct on different topics, they opposed it and I think

25 Your Honor said, listen, that's not how we're going to roll
 3479

1 here, that's not the usual rule. So if that's the rules

2 we're playing under, that's the rules we should play under.

3 And as I said, we have really worked to try to

4 reschedule, to try to ameliorate these problems and we think

5 Professor Whinston should go all at once.

6 THE COURT: Let me ask you this.

7 MR. SCHMIDTLEIN: And if they don't, they can put

8 on his direct and we'll cross him all at once.

9 THE COURT: That's what I was going to ask you.

10 Give me your honest assessment of whether you would be

11 prepared to cross him this week despite not expecting his

12 testimony this week?

13 MR. SCHMIDTLEIN: I would prefer to cross him all

14 together because, although the issues are somewhat separate,

15 they do bleed over a little bit, and particularly the way in

16 which he's addressing them, they bleed over a little bit.

17 And, you know, we found he's going to take a

18 while. If his performance during his deposition is

19 indicative of how he's going to perform at trial, it's going

20 to take a while for him, and he shouldn't be broken up over

21 two sessions.

22 THE COURT: Okay.

23 And -- all right. Okay.

24 So let's do this. Let's get -- Mr. Cavanaugh, did

25 you want to be heard about anything?

 3480

1 MR. CAVANAUGH: Very quickly, Your Honor.

2 We did ask to make sure we filled time this week,

3 we did ask for Krueger on Wednesday. It's fine.

4 Varian on Friday also is fine. He is a short

5 witness. We had originally projected an hour. He's

6 probably closer to half an hour, Your Honor. I just don't

7 think he'll be a long witness.

8 And if we get Whinston in and it doesn't make

9 sense for Varian, we can do him at some later point.

10 THE COURT: So are you, Mr. Cavanaugh, I see that

11 there's no time allocated for you for Professor Whinston.

12 Is that correct?

13 MR. CAVANAUGH: That's correct.

14 As well as a number of other witnesses. We don't

15 anticipate questioning Mr. Chang, we don't anticipate

16 questioning Mr. Judah. So there will be a number of

17 witnesses this week, I don't anticipate us questioning,

18 Your Honor.

19 THE COURT: All right.

20 All right. So let's get Professor Whinston on the

21 stand this week on his direct examination.

22 I will defer the cross so that Google can do it

23 all at once. So let's see where we are. It sounds like he

24 should be able to fill a good amount of the week those days

25 that he'll be on, and that'll also give us a little give in

 3481

1 the joints in case certain witnesses run late.

2 MR. DINTZER: Your Honor, we're comfortable with

3 that provided that we understand that we can -- that we can

4 continue to prep him after he's -- I mean, it's two weeks,

5 and to be perfectly blunt, we haven't finished prepping him

6 for the other part of this case. So if we can't prep him,

7 then we can't call him. So we would prefer to call him this

8 week. We believe it's for the benefit of the Court and for

9 the case.

10 THE COURT: If you're telling me you really can't

11 get him fully prepped, then I'm prepared to give you a

12 little slack, but that's not how we've been usually

13 operating here.

14 MR. DINTZER: If it would be the Court's

15 preference, then we'll keep him where he is in two weeks.

16 THE COURT: That I don't want to do. That is not

17 my preference. I want to make sure we're filling the time.

18 MR. DINTZER: No, we appreciate that. That's why

19 we did this in the first place.

20 As long as we understand that we can -- he'll

21 testify on these two portions that we've discussed, we can

22 continue to prep him, then we're comfortable with that,

23 Your Honor.

24 THE COURT: And to be clear, if I allow that, you

25 know, I will permit Google to object to any sort of direct

 3482

1 examination that bleeds into the two areas that you've

2 already covered; in other words, if there's going to be a

3 third area that you've prepped him for at a later date, it

4 will be limited just to that topic.

5 MR. DINTZER: And to be clear, Your Honor --

6 THE COURT: I'm not talking about redirect,

7 obviously.

8 MR. DINTZER: Right.

9 THE COURT: Because redirect will be for the full

10 three topics because Google wants to do the cross all at

11 once.

12 MR. DINTZER: I want to make sure the Court

13 understands, it's not one topic, that we're basically

14 bifurcating his testimony. There's a good chunk left, a

15 range of additional issues that he'll be addressing, the

16 procompetitive justifications, basically the rest of the

17 case. The only part that he's going to be testifying on is

18 the market definitions and the monopoly power.

19 So as long as the Court understands that and we

20 have -- and the Court -- what the Court held when we

21 discussed this several weeks ago is with good cause, the

22 Court would allow the parties to do that, and we believe

23 that, to make this work -- so it will be -- and --

24 I'm sorry, I'm sorry, I misspoke. The procompetitive

25 justifications will come up in rebuttal when we address --

 3483

1 when he appears in our rebuttal case. But the other

2 elements that he has to address, the anti-competitive

3 conduct and the like, I misspoke.

4 THE COURT: Okay.

5 MR. DINTZER: But we believe we've shown good

6 cause and I believe the Court is holding that we have. So

7 as long as we have that, we will be presenting

8 Professor Whinston beginning Wednesday or Thursday.

9 MR. SCHMIDTLEIN: Okay. And we'll confer with

10 Mr. Cavanaugh about making sure that, I think, if we can get

11 Mr. Varia done this week, we were happy to do that.

12 THE COURT: Given that he's short, ideally, if we

13 can get him, and given Professor Whinston's, we can be

14 flexible. And if Mr. Varia needs to, we need to finish

15 Professor Whinston early -- not early but pause, we can

16 slide a fact witness in, we don't have to --

17 MR. SCHMIDTLEIN: The only thing I don't want, if

18 we can avoid it, if for him come to out here, be ready to

19 testify on Friday and then he has to stay over a long

20 weekend.

21 THE COURT: My point is just the opposite. If

22 he's going to be here, we know he's short, it seems like the

23 only other witness that would prevent him from going on is

24 Professor Whinston, and we can put a pause on his testimony

25 if it's still ongoing.

 3484

1 MR. SCHMIDTLEIN: Thank you, Your Honor.

2 MR. DINTZER: If Google is not crossing

3 Professor Whinston, the time that we had scheduled -- the

4 reason Varia didn't show up is because we didn't know how

5 long they'd take crossing. If they're not crossing him,

6 there really shouldn't be an issue. And just so the Court

7 is aware, even with moving Mr. Krueger and Mr. Varia, there

8 may be some short days. I don't think a lot, but some, just

9 to put the Court on notice.

10 THE COURT: Okay. All right. Well, we'll see

11 where we end up.

12 One last thing before we get started, where are we

13 on requests for closed sessions for Mr. Nadella?

14 MR. SEVERT: Your Honor, Adam Severt for the

15 United States.

16 We expect to be able to conduct his examination in

17 open session today.

18 THE COURT: Okay.

19 MR. SCHMIDTLEIN: We do as well.

20 If we get -- depending on how some of the

21 conversations go with some of the documents that we're going

22 to try not to display, we may ask to go into a short

23 session, but we also are going to try to respect the rules,

24 I think, that we've been operating under.

25 THE COURT: Okay. Terrific.

 3485

1 Mr. Larrabee -- oh, Mr. Cavanaugh.

2 MR. CAVANAUGH: We're going to take some limited

3 questions in open sessions.

4 THE COURT: Okay.

5 Mr. Larrabee.

6 MR. LARRABEE: Just briefly, Your Honor.

7 Following up, and consistent with the comments you just

8 heard following up over the weekend, we discussed this issue

9 of confidentiality. Mr. Nadella is strongly committed to

10 doing this in public to the maximum extent.

11 The two limitations on that optimistic prediction

12 are the documents that are designated confidential. I think

13 the parties are all agreed not to put them on the screen,

14 not to read the confidential parts to him. I think that'll

15 work.

16 And the only other caveat is if he gets a

17 question, which we don't anticipate, but if he gets a

18 question that he feels requires him to disclose something he

19 thinks is genuinely confidential, I've asked him to speak

20 up, and then we'll address it with the Court.

21 THE COURT: Okay. Terrific. All right, thanks.

22 I appreciate that, Mr. Larrabee.

23 All right. Is there anything else before we get

24 started?

25 MR. DINTZER: Not from the DOJ Plaintiffs,

 3486

1 Your Honor. We're ready to call our next witness.

2 MR. CAVANAUGH: No, Your Honor.

3 MR. SCHMIDTLEIN: Thank you, Your Honor.

4 THE COURT: All right. Let's proceed.

5 MR. SEVERT: Thank you, Your Honor. Adam Severt

6 for the United States.

7 THE COURT: Mr. Severt.

8 MR. SEVERT: The United States calls Satya

9 Nadella.

10 COURTROOM DEPUTY: Before you have a seat, can you

11 please raise your right hand.

12 (Witness is placed under oath.)

13 COURTROOM DEPUTY: Thank you.

14 THE COURT: Mr. Nadella, welcome.

15 THE WITNESS: Thank you.

16 THE COURT: Counsel, whenever you're ready.

17 - - -

18 SATYA NADELLA, WITNESS FOR THE PLAINTIFF, SWORN

19 DIRECT EXAMINATION

20 BY MR. SEVERT:

21 Q Good morning, Mr. Nadella.

22 A Good morning.

23 Q My name is Adam Severt. I'm a lawyer for the

24 United States. I'm going to be asking you some questions

25 this morning.
 3487

1 Can you please state your name for the record.

2 A Satya Nadella.

3 Q And where are you employed, sir?

4 A Microsoft Corporation.

5 Q And what's your position?

6 A CEO of Microsoft.

7 Q And how long have you been CEO at Microsoft?

8 A Since 2014.

9 Q And what positions have you held at Microsoft

10 before you became CEO?

11 A I've been at Microsoft since 1992, so just prior

12 to becoming CEO, I was heading the cloud business for

13 Microsoft as EVP.

14 And then before that, I was running an online

15 services. And before that, our biz apps, business

16 applications. And before that, Windows and Office.

17 Q And before you became CEO, what responsibilities

18 did you have for the Bing product and its predecessors?

19 A So since -- I think from 2007 to 2011, I ran what

20 was the engineering group responsible for search and

21 advertising, among other things.

22 Q And for what period of time did you work on that

23 business?

24 A 2007 to around 2011.

25 Q And then you got -- from 2011 until you became

 3488

1 CEO, did you have responsibilities for Bing?

2 A No, I didn't.

3 Q Okay.

4 And just, sir, at a high level, can you describe

5 how Bing fits into Microsoft's overall strategy?

6 A You know, Microsoft has been always focused on, as

7 a software company, to look at software categories that we

8 want to participate in, both because we think that we can

9 bring innovation to it, and also the categories' attractive.

10 So when I think about Bing from a Microsoft

11 perspective, I sort of see search or Internet search as the

12 largest software category out there by far. There's nothing

13 even close. I used to think Windows and Office were sort of

14 attractive businesses until Internet search. And so once

15 that became clear, we were very committed to making progress

16 in that category.

17 So, yes, we're a very, very low-share player,

18 we're grounded in that, but we continue to persist in it

19 because we think of this as a software category that we can

20 contribute innovation to, and it's a very attractive market.

21 Q In what sense is Internet search the largest

22 software category? What do you mean by that?

23 A Just by share, revenue, margins. Whichever way

24 you look at it today, you know, it's dominated by one

25 player, so therefore it's got high margins.

 3489

1 But even if you sort of take that in aggregate,

2 the market is the largest software category.

3 Q And just sort of talking -- looking at sort of

4 Bing in particular, what's the Bing strategy within

5 Microsoft?

6 A Within Microsoft is more about being able to make

7 progress in what is the largest software market.

8 And then do so by picking a strategy that allows

9 us to continually invest. One of the things I did, you

10 know, when I became CEO was to make sure that the business

11 became profitable so that we can keep investing in it on a

12 continual basis, knowing that this is a large, attractive

13 market, and then if a window opens up for us, then that'll

14 be something that Microsoft can benefit from.

15 Q So is Bing profitable today?

16 A Yes, it is.

17 Q And you said you're waiting for a window. What do

18 you mean by that?

19 A You know, when I think about any software market,

20 digital market, there are sort of two sets of things that

21 were. One is that there is an opportunity for us to see

22 either some paradigm shift or, quite frankly, some

23 intervention in the market.

24 When I think about sort of what Microsoft itself

25 had to go through was both an antitrust intervention, plus

 3490

1 innovation, both of those impacted us, so I kind of look at

2 that and say, you know, What is the way something like that

3 happens in the search market.

4 Q And why was making Bing profitable important to

5 this strategy of sort of looking for a window?

6 A The -- you know, continuing to participate in this

7 market requires two things, a significant amount of fixed

8 cost investment, and on top of that, you need to have -- you

9 know, you need to sort of have an innovative product. And

10 so, therefore, in order to be able to justify that fixed

11 cost investment, it makes it easier to have a profitable

12 business.

13 So we ran it without profit for over a decade

14 plus, and so I felt that one way to sustain Microsoft's

15 interest in this category was to turn profitable.

16 And pick -- in fact, when I say profitable, is not

17 profitable just for profit's sake, but to pick a segment of

18 the market, which was desktop search, which is what we

19 participated in, and use that as a way to just stay in the

20 game, so to speak.

21 Q And you mentioned a couple of times that Bing is

22 profitable, and I understand that in Microsoft Securities

23 filings, it reports -- it's Bing revenue under the search

24 and news advertising product line; is that correct?

25 A That's correct.
 3491

1 Q And what portion of that, the revenue that,

2 Microsoft reports, is attributable to Bing?

3 A The vast majority of it.

4 Q Okay.

5 And just going back to the point you made about

6 search, Internet search being the largest software category,

7 what about that makes it attractive to Microsoft?

8 A Just the size of it.

9 And, you know, in some sense when you look at a

10 company like ours, we do have, you know, a set of

11 diversified businesses.

12 But if you're a software company looking to sort

13 of participate in big markets that can be needle moving for

14 you, there's nothing more attractive than Internet search

15 out there.

16 Of course, you also still have to be grounded in

17 the fact that if you're a low-share player, it's very, very

18 hard. It's a supply-constrained market; there's many, many

19 lock-ins. So, therefore, it's a hard game to make any

20 breakthroughs.

21 But it's, you know -- no one can accuse of not

22 being persistent.

23 Q And given the size of the opportunity and

24 potential profits, why has Microsoft and Bing not be able --

25 not been able to, at least to this point, increase its

 3492

1 market share?

2 A I mean, look, that's -- it's a -- when you say

3 we've not been able to increase our market share, the

4 interesting thing is even being a very, very low-share

5 player, you can run a profitable business and wait for the

6 next day when something could change, is how I look at it,

7 as an opportunity for us.

8 And when I think about why it's hard, it's -- you

9 know, in every digital market, that are network effects.

10 The thing with search in particular, and I would

11 ISA say consumer Internet properties, the virtuous cycle or

12 vicious cycle, depending on which side of it you find

13 yourself, it becomes even harder to break through, right,

14 because if you think about it, once you have dominant share,

15 users are just most familiar with your user experience, and

16 change is hard.

17 You get all the distribution, changing of defaults

18 is hard. All publishers will optimize for your search

19 crawler. So, therefore, no one optimizes for you, so that

20 kind of impacts quality.

21 Advertiser density, that is, advertisers trying to

22 reach users will come to your auction. And advertising is

23 content. And so you have to break that, as I said, vicious

24 cycle or virtuous cycle, depending on which side you fall

25 on, and that's sort of the hardest part of it.

 3493

1 It's kind of also an interesting market, which is

2 it's a multisided market. It's -- you can't bring

3 traditional price competition because the price on one side

4 is zero, and the other side, in fact, the prices keep

5 going up.

6 And so that even creates even harder, like, how

7 does one even bring price competition is one of the

8 fundamental challenges of these multisided markets.

9 Q And what have you done to try to combat this, as

10 you call it, vicious cycle?

11 THE COURT: Counsel, can I interrupt?

12 Mr. Nadella, can I just ask you to clarify your

13 response.

14 You said one side of the equation, there's zero

15 price, I understand that, the consumer. You said the other

16 side of the equation, the price is going up. What did you

17 mean by that?

18 THE WITNESS: So on search, in simple terms,

19 there's advertisers bidding in an auction to get placement

20 in order to reach consumers, and so the consumers obviously

21 get a free service.

22 But the advertisers, you know, supply-constrained

23 market, which when there's one dominant supplier, every

24 advertiser needs to go into that one auction.

25 And so if the density of the auction goes up, the

 3494

1 prices go up.

2 And so, for example, the empirical evidence is

3 anywhere there's a marketing campaign or an advertising

4 campaign moves from Google to Bing, the first thing they

5 notice is they get better ROI because the prices go down for

6 them. But there isn't enough traffic in Bing to justify

7 sometimes having the same intensity of auction, if you will.

8 THE COURT: Thank you.

9 BY MR. SEVERT:

10 Q And just to follow-up on the Court's question, has

11 Bing had difficulties in increasing advertiser participation

12 in the Bing auctions?

13 A Oh, yeah.

14 I mean, it starts with, quite frankly the share,

15 because advertisers go where there is high share. So it

16 sort of -- that's the first barrier.

17 And then the second barrier is there is things,

18 you know, that, like SA360 and some structural issues and

19 how campaigns can be moved over in a friction-free way and,

20 you know, obviously we've not been able to break through on

21 that with Google in terms of making it such that an

22 advertiser can have the same campaign across multiple search

23 engines and so on.

24 Q And you mentioned, just to sort of go back, you

25 mentioned the vicious or virtuous cycle depending on what's

 3495

1 your perspective. What have you done to try to combat or

2 overcome that?

3 A I think the predominant thing that we've achieved

4 some modicum of success, which sort of even helps us

5 continue down this path, is desktop search.

6 I mean, the one inherent advantage Microsoft has

7 is we get to set defaults on Windows and create incentives

8 for our OEMs for those defaults and what have you.

9 And so therefore in that context, as I said,

10 I think, you know, on desktop search, we're being able to

11 grow share, we've become more competitive with our browser.

12 And so that's sort of perhaps the most successful

13 strategy we've had. And, of course, one of the other things

14 that we've been trying very, very hard is to get additional

15 distribution.

16 Q And you mentioned you have more share on desktop.

17 About what share do you think you have on desktop?

18 A It depends.

19 If you use third party, things like Comscore,

20 probably in the teens.

21 Q How about in mobile?

22 A Mobile would be low, low single digits, if that.

23 Q So what's the rationale for investing in the area

24 where you're already -- at least relatively more strong,

25 versus the area where you're relatively more weak?

 3496

1 A It's one market.

2 So at this point, the desktop search market is not

3 a growing market. The growing market is the mobile search

4 market.

5 And also, you can't, from an advertiser, or let's

6 start with the user. User quality of search is not that I

7 only search on my phone, I only search on my desktop.

8 In fact, one of the inherent advantages Google has

9 even on desktop is the fact that they see all the query

10 stream on mobile makes their search -- the dynamic data

11 access is a big asset.

12 So the more queries you see, the better search

13 quality you're going to have by definition and that search

14 quality is not limited to just mobile search. It also goes

15 over to desktop search. So you have to participate in both

16 markets to be long-term relevant.

17 Q And you mentioned sort of the user interaction.

18 Is that a relevant consideration for the investment in, more

19 focus on desktop?

20 A Yeah, I mean, that's a place where we have the

21 best distribution. So I divide this up into a couple

22 different dimensions.

23 One is you have to build a competitive search

24 product with making all the fixed cost investments. And

25 then build out even the entire data center infrastructure

 3497

1 before you even get the first query.

2 But then after that, it quickly becomes how do you

3 get enough queries to continuously improve search quality

4 and enough advertisers to get enough advertising in order to

5 sustain not just the business model but, as I say,

6 advertising itself is content.

7 And so that's kind of where we have focused on

8 really making sure that we do our best job on both those

9 fixed costs, as well as the dynamic data acquisition, if you

10 will, on desktop.

11 And then, you know, continuing to try to get

12 distribution in mobile.

13 Q Sure.

14 And you've mentioned distribution a few times.

15 When you say distribution, do you mean defaults?

16 A Defaults.

17 Q And why is that the type of distribution that

18 you've been focused on?

19 A It's the only -- I mean, it's the mechanism that

20 fundamentally can change.

21 This entire notion that users have choice and they

22 go from one website to one website or one search into one

23 search and it's complete bogus. There's defaults. The only

24 thing that matter in terms of changing search behavior.

25 And at this point, in fact, it's even more true,

 3498

1 right. I mean, there's -- you get up in the morning, you

2 brush your teeth and you search on Google. And so

3 therefore, with that such level of habit forming, the only

4 way to change is by changing defaults.

5 Q And is there a difference between defaults on

6 desktop versus mobile?

7 A The desktop, it ironically turns out that Windows

8 is the most open platform today, because anyone can

9 distribute anything on Windows, they don't need to go to

10 Microsoft.

11 And so, yes, we do have defaults of ours, but it

12 doesn't require any inter -- or -- any coordination with

13 Microsoft to distribute a complete alternate player which

14 includes the browser as well. So changing defaults today is

15 easiest on Windows, toughest on mobile platforms because

16 they're all locked up; they're locked up on the browser that

17 is allowed, they're locked up with app store access. So

18 there are many, many sort of friction points on mobile

19 operating systems.

20 Q And you made a statement that Windows is the most

21 open platform today. What did you mean by that?

22 A I mean, here's the thought experiment.

23 To ask ourselves today and say, would Google even

24 exist if Windows had iOS-like restrictions on how things get

25 distributed? And so, it just -- that's what I mean.

 3499

1 Like Google exists because of two things. One is

2 because of our consent decree, where we had to put a lot of

3 limits on what we could distribute and not distribute by

4 default.

5 And, second, because the fact that you could

6 distribute anything you wanted on Windows, and it's still

7 the case, right, it's not just Google, take the largest

8 marketplace on Windows happens to be not from Microsoft,

9 it's Steam.

10 And so it's an open platform on which anybody can

11 distribute anything. And that is a great advantage for

12 somebody trying to build a new business.

13 Q And in what way is -- you mentioned iOS, but how

14 about Android, in what way is Android --

15 A It's the same thing.

16 Both of these, it's just that the real estate --

17 I mean, Android by market share is the highest. IOS has got

18 the highest R pool because of the user base. But in the

19 U.S., iOS would probably be more important. To gain access

20 worldwide, Android would be more important.

21 Q And I'll just -- sort of focusing on mobile

22 devices just for a few moments. There are -- am I right

23 that there are multiple search access points on a mobile

24 device?

25 A There are multiple search access points, but the

 3500

1 one access point that matters is the search default on the

2 browser.

3 Q Why?

4 A Just because that's the vast majority of the entry

5 points. You know, they might be in time some new entry

6 points, like these assistants like Siri or Google Assistant.

7 But right now, the search default is fundamentally

8 synonymous with the browser entry point.

9 Q And has Microsoft tried to get default positions

10 on mobile devices?

11 A Yes, we have.

12 Q How has that gone?

13 A Not very well.

14 Q Why not?

15 A I mean, the one that we tried and continue to try,

16 quite frankly, is the Apple default.

17 I mean, if you think about it, if you step back in

18 a second and say, what really are the defaults that truly,

19 truly matter that are even available, you -- obviously,

20 there's no availability of the default on Android because

21 Google owns it. There is no way you're going to get a

22 default on Chrome, which happens to be the dominant browser,

23 even on Windows, so, therefore you're not going to get that.

24 So you're left with really iOS as the only thing

25 that you can truly contend for as sort of another player in

 3501

1 search, which is needle-moving in a big way.

2 So I focused every year of my sort of tenure as

3 CEO to see if Apple would be open, and we've had, you know,

4 a series of dialogues with them over the years on it.

5 Q And just sort of following up on the discussions

6 with Apple, I think the court has heard about some of those

7 discussions, I'm not going to ask you to sort of repeat the

8 back-and-forth, but what would a deal to be the default in

9 Safari, what would that do for Bing?

10 A It would be a big game-changer.

11 If you think about it, my confidence, quite

12 frankly, in all of that comes from what Apple was able to do

13 with Apple Maps, which is, they just said, Apple Maps is the

14 default, there was a period of turbulence, and now other

15 than the few elites in places, everybody just uses Google

16 map -- or rather, Apple Maps.

17 And so to me, that's what the power of defaults

18 is. In fact, that's I would say exhibit number one on

19 anybody who wants to sort of know anything about sort of the

20 power of defaults, that case study is the best.

21 And so I basically look at that and say Apple,

22 both because of the brand they have and the power they have

23 of setting defaults on their operating system, that whomever

24 they chose, they basically king make on it.

25 And so -- and then, of course, they have economic

 3502

1 interest in it; they get paid for that default setting.

2 And so that's where the entire discussion is.

3 Q And what would this additional traffic that Bing

4 would receive through that default, what impact would that

5 have on Bing's search quality?

6 A The thing that we were -- it's a good question,

7 because the thing -- obviously the economic benefits to us

8 are huge, but quite frankly, we were going to essentially

9 trade away -- my entire goal is to bring essentially price

10 competition here. So I think all of the economics we are

11 going to get from Apple on that traffic would go to Apple

12 and we were not sort of confused by that.

13 But the fact that our search quality product --

14 our search quality would improve because we would finally

15 get to see the dynamic data, right, I mean, every

16 misspelling of a new movie that people sort of have will

17 show up first in mobile and you'll get that, or the local

18 restaurant, you know, that you mistyped or what have you.

19 And so the dynamic query data is a very critical asset to

20 have your search quality get better.

21 Q And you said that Microsoft was -- the premise was

22 that Apple would essentially receive the economic benefits

23 of the -- what do you mean by that?

24 A I mean, today, Apple makes, you know, a lot of

25 profit and a lot of money because of the money they collect

 3503

1 from Google for setting Google as the default, and so we

2 were always grounded in the fact that if they needed to

3 switch, that you would have to pay, and we would have to pay

4 and even subsidize the transfer, right, which is you have to

5 sort of, in this period of transition, most of the economics

6 were going to go to Apple, we were going to be negative for

7 multiple years perhaps, and I was willing to do essentially

8 a very long term and think perpetually.

9 So the basic concept I had was in order to break

10 some new ground in this search competition was to

11 essentially run search like a public utility, where two

12 suppliers can bring their supply, Microsoft with its desktop

13 and Apple with its mobile search, and then make search more

14 competitive. So that was sort of the economic construct

15 that I used to try and see if Apple was interested in doing

16 the switch.

17 Q So let me make sure I understood that.

18 So you mentioned in the short term that you were

19 prepared to take losses; is that right? Is that what you

20 said?

21 A Yes.

22 Q And can you give the Court a sense of sort of what

23 magnitude of losses you were prepared to take?

24 A It would be billions of dollars. I mean, like,

25 I don't know exactly what the levels of payment are today,

 3504

1 but they're significant, let's call them naught of 10 plus

2 billion a year, I don't know, maybe even close to 15 billion

3 a year.

4 So you then have to sort of color all that and

5 cover for any even loss share because Google -- one of the

6 things that Apple was, rightfully, very scared of was what

7 Google would do if they switched, because Google has a lot

8 of power. They have eight plus properties with multiple

9 billion users, and every time a user hit Gmail or YouTube,

10 they were going to get them to download Chrome, and so there

11 was going to be a period of turbulence of shift. And so we

12 had to cover them for that risk and what have you.

13 So there's significant amount of -- yeah, there

14 would be real competition. And in that competition, there

15 would be economic loss because right now, there is basically

16 status quo, right; the dominant player in search is paying a

17 lot of money to maintain that share position.

18 Q And I think the other thing you said was that you

19 were -- I think, in your deal construct, was to give away to

20 Apple the economics in perpetuity?

21 A Yeah.

22 So, you know, it's interesting. If you look at

23 the original paper of the Google founders, they even talked

24 about how search should be a public utility, and so that was

25 the construct I had in my head, which is to say, Hey, it's

 3505

1 really hard, right, Apple's already making a lot of money,

2 and -- but at the same time, I think it was -- my argument

3 to Apple was it's in their long-term interest to have at

4 least two suppliers because, quite frankly, they're getting

5 the benefits of having two suppliers, except they're not --

6 they're using us only to bid up the price, but in the long

7 run, that doesn't work. That was one argument.

8 The other argument was to say, Look, someone told

9 me, you know, if all you want to do is check how Google Maps

10 was priced, it was sort of zero until it was infinite. So

11 that is kind of what happens, right.

12 When you have a dominant position, you're not

13 going to keep paying. At some point, let's say Bing exited

14 the market, you think that, you know, Google is going to

15 keep paying Apple for defaults? I mean, if there was no

16 search competition, why would they do that? So those are

17 the economic rationale arguments I was making.

18 And also making the commitment that, Look, we're

19 not going to do this and essentially be a substitute to

20 Google in terms of economics. We want to bring a complete

21 different set of guarantees that in perpetuity, they can

22 essentially monetize their supply, and we will just share

23 the modernization engine.

24 Q And so let me ask, then, if it wasn't really the

25 economics that -- from the Safari deal that Microsoft was

 3506

1 going to receive, what was the point? What -- what was

2 Microsoft going to get out of it?

3 A So if -- there are two sets of things, right. One

4 of the things we talked about was auction density.

5 So as soon as, let's say, Bing's share on desktop

6 and Safari's share on mobile get aggregated on one auction,

7 the first thing that will happen is the auction prices will

8 get more normalized, and so there will be benefits of that

9 even for us on the desktop side. And so that is sort of

10 a -- and publishers will benefit, and users will benefit

11 because we'll invest more and what have you.

12 So to me, that was the gist of it, which is, we

13 don't have to -- in a large, attractive market like Internet

14 search, we needed to be less greedy and more competitive.

15 Q And so the -- my understanding is that the

16 benefits that Microsoft would receive, it would essentially

17 be everywhere else; is that what you mean?

18 A Predominantly, yeah.

19 Q And I think you talked before about dynamic data

20 as a --

21 A That would be huge.

22 Q Is that sort of the benefit?

23 A That would be another one, right, which is right

24 now sometimes we do have users abandon Bing or Edge,

25 because, you know, we didn't get the local restaurant right

 3507

1 because we didn't see it on the desktop, if you will. So,

2 yes, there will be search quality benefits.

3 And there's so many benefits of having share,

4 right, which is, not an index problem, which is basically

5 being able to see all the websites, publishers, optimizing

6 for your SEO, because today, mostly everybody -- you know,

7 everybody talks about the open web, but there is really the

8 Google web. And so, therefore, everything is optimized for

9 Google Search and Chrome and what have you.

10 And so the degree to which there was an alternate

11 with enough share, then every publisher, every advertiser

12 will start optimizing for it.

13 Q You made reference to the Google web. What is

14 that?

15 A Look, Google -- if you think about the organize --

16 the web got created and the organize -- and it was all

17 "open" because you could go to a browser and go to any

18 website, except the organizing layer of the web is the

19 search engine. So if nobody goes to any website, it was --

20 it's all intermediated by search engine.

21 So all the power of the web got aggregated in a

22 search engine, and Google has, you know, the high share in

23 it. And so that's why -- and they were able to parlay that

24 successfully into essentially every publisher cares about

25 Google Search. Every publisher monetizes uses Google

 3508

1 advertising network, you know, and the list goes on.

2 Q And I think one thing that you said when you're

3 talking about the benefits of a deal with Safari is that you

4 would be able to invest more. What did you mean by that?

5 A Can you repeat the question.

6 Q Sure, sure.

7 If Bing had entered into -- got Safari traffic,

8 would that allow Microsoft to invest more in mobile search?

9 A Absolutely.

10 Because, one, in order to start investing -- we're

11 making all the fixed cost investments already. And for us

12 to make additional fixed cost investments on, say, search

13 relevance or search scale, we needed to get more of the

14 dynamic data and mobile distribution.

15 And so, yes, absolutely, that would have made us

16 more competitive as a search product that end users would

17 have liked and advertisers and publishers would have

18 benefited from as well.

19 Q And just to sort of back up to talk about

20 investments a little more broadly, how -- how does Microsoft

21 think about what investments it can make, sort of at a high

22 level?

23 A As I said, there are two classes of investments.

24 Obviously there's the operating expense of having the

25 engineers and the product makers work on the product, and

 3509

1 then there's the capital investment, and maybe the third one

2 is the investments in the go-to-market, including

3 distribution. So all three are always in contention at any

4 given time.

5 And the one way to think about it is in a market

6 where there is more competition, we are growing share, that

7 means as a percentage of our revenue, we'll invest back

8 more, because, after all, we're a, you know, contender, and

9 so that's kind of how we would look at it.

10 Q And are there any constraints on Microsoft in

11 terms of, you know, what it can invest in just by it being a

12 public company?

13 A Sure.

14 I mean, in any -- at the end of the day, I'm

15 subject to sort of the same laws of gravity that anybody is

16 as a public company, which is it has to be in the long-term

17 interest of our shareholders.

18 So I always have to sort of argue the case

19 internally and externally by showing real progress that,

20 Hey, we picked the right markets to participate in, we are

21 making progress in those markets, and, yes, that, you know,

22 if we continue to make the progress, then we get more

23 permission from our shareholders to be able to invest.

24 Q And this may come as no surprise to you, but

25 Google has argued that Microsoft is under-invested in Bing.

 3510

1 Do you agree with that?

2 A You know, I would say as someone who has -- let me

3 put it this way. As per capita to our revenue we've made,

4 we've invested a lot, more so than Google has invested, in

5 search. And so, therefore, I would say, you know, we're the

6 only player other than Google that has continued to invest

7 in search.

8 And, quite frankly, the investments Microsoft has

9 made in search has even kept all the other search players

10 who contend for it, like a DuckDuckGo, even going, because

11 they use our search index. And so I kind of look -- I --

12 I don't subscribe to that view that we've not invested.

13 Can we invest more? Of course, any day, you know,

14 everybody wants to invest more. And in order to invest

15 more, please give me some mobile share and I'll invest more.

16 Q And how much money Microsoft has invested in Bing

17 over the years?

18 A You know, if you take sort of our latest numbers

19 and you say the vast majority of it is search revenue and --

20 and we drive today profit, but over the last 20 years, maybe

21 close to $100 billion.

22 Q And is that net of revenue?

23 A No, this is total investment.

24 THE COURT: While counsel is paused, can I ask you

25 a slightly different question, which is, you mentioned

 3511

1 DuckDuckGo, for example.

2 Is it conceivable to you that there could be a

3 nascent search company that could be built from the ground

4 up, given the barriers of entry and the costs to invest in a

5 truly first-class search engine?

6 THE WITNESS: There is no existence proof of that,

7 that's what I'd say.

8 It's sort of -- it's very -- sort of hard to say

9 that, you know, in tech, at least, we love to sort of talk

10 about how anything is possible, because that's the dream, at

11 least, we try to keep alive.

12 But this is just a hard one, the fixed cost.

13 That's why, the reason I brought up DuckDuckGo is DuckDuckGo

14 had an angle, they wanted to be, you know, in fact, a

15 privacy-first type of search engine.

16 But then they piggybacked off of a bunch of the

17 investment we made, and we were happy to give them that

18 because we wanted search competition, you know. It would --

19 in fact, it would be very similar to me if DuckDuckGo was

20 able to make a lot of progress; it would be no different

21 than Apple default for me.

22 And so, yes, so we're always looking for if

23 somebody else has a better idea, or even recently, you know,

24 Neeva was one of the other competitors, we were happy to

25 give them our search index.

 3512

1 But it's just -- you know, because of all the

2 reasons why we have struggled as Microsoft, for all sort of

3 the strengths we have, including our own distribution in

4 Windows, it's very hard for a startup.

5 In fact, you go to Silicon Valley, they have these

6 things called the no-fly zones and you have to go ask them,

7 Which is a biggest no-fly zone, it will be Internet search.

8 THE COURT: And we've heard about how AI has now

9 been integrated into Bing search. To what extent do you

10 think, if at all, AI can make up the delta in -- between

11 Google and Bing in terms of search quality?

12 THE WITNESS: So that's a -- there are two sets of

13 things, right, which is, the age of these large language

14 models creates, I'll call it, a new baseline for what both,

15 now at some level, Bing and Google and anybody else who

16 wants to compete in search has to do for core search ranking

17 and be republic with this, which is -- some of the largest

18 sort of core ranking gains we had was when we started using

19 these large language models.

20 But it fundamentally doesn't address all the other

21 challenges, right, which is the user habit comes from search

22 defaults, settings, so the distribution advantage Google has

23 today doesn't go away.

24 In fact, if anything, I worry a lot -- even in

25 spite of my enthusiasm that there is a new angle with AI,

 3513

1 I worry a lot that, in fact, this vicious cycle that I'm

2 trapped in can even become even more vicious because the

3 defaults get reinforced.

4 The publisher content can get locked in. And so

5 if there are exclusive content deals which are happening

6 right as we speak, which is scary because what that does is

7 not only lock all that content today, at least, that's

8 crawlable by everyone and usable by large model training,

9 could become exclusive.

10 THE COURT: When you say "lock in deals," you mean

11 locked-in deals for the AI crawling in terms of --

12 THE WITNESS: So you can parlay high share today.

13 Think about -- you have the largest sort of economic sort of

14 surplus. So you can use that economic surplus to

15 essentially buy off publishers of all kinds.

16 And it's -- in the case of obviously Alphabet

17 Google, you know, they're super strong, not only in Google

18 Search but also in YouTube. So you can see how they could

19 use both of those as places where that content will be

20 exclusively used by them to train their models and serve

21 their models.

22 And it will not be viable. So right now I'm

23 struggling to get a default distribution, and I'll have to

24 now contend with an additional thing called -- now you're to

25 pay up, to every publishers, the rates Google can pay,

 3514

1 because Google can pay using that economic rent they collect

2 today, which we don't collect.

3 So I think this is going to become even harder in

4 the AI age to compete with someone who has that core share

5 advantage and economic advantage.

6 THE COURT: So say you were Tim Cook. Why would

7 Apple ever have want to switch to Bing?

8 And what I mean by that is, given the lag in what

9 I think we've all heard is less user quality or less quality

10 of Bing search results on mobile, we understand what the

11 reasons are, what would be their incentive to change and

12 what circumstances could you ever see them changing?

13 THE WITNESS: Yeah.

14 So I don't think anybody would want to switch to

15 any product that's inferior in the long run.

16 We had proven to them that with the access to

17 their query stream and their share position, that we can

18 absolutely bridge the quality gap, whether it's in local

19 search, international search, all of the various

20 considerations. So we know how to do that, it's a set of

21 investments that need to be made. More importantly than the

22 investments, it's the data that needs to be there.

23 Then your question of why would Apple ever want to

24 do this. It would want to do this if they felt that

25 economically in the long run for them having competition in

 3515

1 search is helpful.

2 And this is sort of an interesting question for

3 them, right, which is, of course, you could say, hey,

4 there's this fantastic simple oligopolistic arrangement

5 between number one and number two, so why upset the apple

6 cart, no pun intended. So that might be one way to go about

7 it.

8 But I do feel at some level Android and iOS

9 compete and so, therefore, is it not in the interests of

10 Apple to actually have competition.

11 So that's at least my rational argument for them.

12 THE COURT: Thank you.

13 BY MR. SEVERT:

14 Q And just following up on that a little bit, if

15 there were investments in Bing that Microsoft could make to

16 improve quality and drive more users, is there any reason

17 why Microsoft wouldn't make them?

18 A If we can make -- we're making all of those

19 investments today.

20 Q I want to go back. We started talking about --

21 strike that.

22 Do the investments that Bing would make to make

23 economic sense -- is query flow kind of a prerequisite for

24 that?

25 A Yeah, I mean, distribution deals are prerequisite.

 3516

1 There's two sets of things, right, which is each

2 distribution deal will come with a very specific context,

3 which is, it's in a country or it is on a platform like

4 mobile which has some context. And so we would make

5 incremental investments as needed in order to sort of defend

6 the default that we want, if you will.

7 Q Sure.

8 We've been talking about iOS. I want to go back.

9 When we talked about distribution, you also talked about --

10 you mentioned Android. Do you remember that? And has

11 Microsoft done anything to try to get distribution on

12 Android?

13 A Yes.

14 Obviously, it's very, very hard because it's a

15 home territory for Google.

16 And, you know, even of recent, we've been trying

17 only to be smacked down by -- on our attempts.

18 For example, we went to Xiaomi, and said, let's

19 get secondary browser, not even a primary browser.

20 And this was in a European country, I forget now

21 which locale, and Vodafone was the carrier. So what we got

22 back was basically we got thrown out of that because the

23 marketing dollars Google uses with Vodafone were used as a

24 way to get Xiaomi to basically get rid of the secondary

25 browser.
 3517

1 So, yes, every day of the week, we're trying,

2 whether it's SwiftKey distribution, which also got taken

3 away because of the MADA restrictions, what have you. It's

4 a tough one. Google does a good job of maintaining their

5 positions with their agreements with OEMs and carriers.

6 Q What specifically -- you mentioned the MADA. What

7 specifically does the MADA do that --

8 A MADA -- I'm not the expert in all the details of

9 it, but fundamentally, look, everybody defends: Apple

10 defends their defaults, we defend our defaults, and Google

11 defends their defaults.

12 The one advantage Google has in defending their

13 defaults is I think of as a game of carrots and sticks. We

14 all try to defend by carrots, which is pay for OEM default

15 distribution.

16 Google has carrots and it has massive sticks, like

17 one big stick is that we'll remove Google Play if you sort

18 of don't have us as the primary browser. And without a

19 Google Play, an Android phone is a brick.

20 And so that is the type of stuff that is

21 impossible to overcome. No OEM is going to do that.

22 And so we have to sort of watch for it. That's

23 why I think Android realistically for us, just even getting

24 some secondary entry point is probably the best shot we

25 have.
 3518

1 Q And you mentioned something called SwiftKey. What

2 is that?

3 A It's a software keyboard on mobile phones, so we

4 bought a company a while ago, which actually had significant

5 OEM distributions, because OEMs liked SwiftKey. In fact,

6 they paid SwiftKey to include a good software key. And

7 then, of course, we wanted to continue that. And once it

8 became part of Microsoft, I think Google rewrote some of

9 their terms on what is allowed and not allowed in terms of

10 third-party keys, keyboards.

11 Q What the connection between the SwiftKey keyboard

12 and search?

13 A Because you can -- I mean, keyboard could be,

14 could be, an alternate entry point because you could have a

15 search box on a keyboard.

16 So if there was a default keyboard then which had

17 an entry search box, that could act as an another type of

18 entry point other than the search default on the browser.

19 Q And what happened to SwiftKey after it was

20 acquired by Microsoft?

21 A As I said, I think there were changes in some of

22 the agreements, and some of the agreements they had with

23 OEMs, that is SwiftKey had with OEMs, just became harder to

24 maintain because all the economics on that changed.

25 Q And I think when you were talking about scale

 3519

1 earlier; you mentioned that you need enough queries to

2 continually improve search quality.

3 What do you mean by that?

4 A The search is a product in which what you see

5 today informs quality. So it's not just what you saw all in

6 history, but there would be always new novel things that are

7 being queried for the first time because there's new content

8 that is being generated for the first time, being indexed

9 for the first time, new things are coming out.

10 And so unless and until you are in that river of

11 sort of seeing what people are searching for all the time

12 and have the diversity, right -- so you're seeing stuff in

13 mobile, you're seeing stuff in desktop, you're seeing it

14 internationally, all of that matters.

15 So dynamic data is sort of really part of search

16 quality. There may be other software categories like that,

17 but this one is much more, you know, if you're not a high --

18 in some sense, the share and scale is quality on two fronts.

19 One is because of the search quality of organic results; and

20 advertisers are going to be more intense on your platform

21 and so, therefore, advertising is also content. And so

22 between the two, you need to get to high share in order to

23 have a high quality product long term.

24 Q And is there some time element to the -- to having

25 scale? Does it sort of build upon itself over time?

 3520

1 A Yeah.

2 I mean, like if one -- one of the -- the dreams,

3 at least, I have is that, hey, at what level does it start

4 getting -- like on desktop, I think there's some empirical

5 evidence that there is, you know, 20 points of share or

6 what -- you can have a very competitive market, right,

7 between Google and Microsoft and some of it is also because

8 we can get to defend our defaults.

9 So my hope is, hey, what would the world look like

10 if we, or somebody else, had 20 points of mobile search

11 share. You know, would there be more competition on -- and

12 rather, will there be stability on search quality gains.

13 Q And I want to talk for just a couple minutes about

14 Microsoft syndication deal with Yahoo!. Are you familiar

15 with that?

16 A Absolutely.

17 Q And what is Microsoft providing to Yahoo! in terms

18 of search?

19 A Again, it's very similar to what we do with

20 DuckDuckGo, what have you. We provide them all of our

21 search, essentially white labeled, so that it's Yahoo!'s

22 search for people who go to Yahoo!

23 Q And how long has that relationship existed?

24 A It's existed now for a long time, and I think we

25 renewed it even recently again. So it's 10 plus years.

 3521

1 Q And when you first entered into that deal -- well,

2 first, were you involved in that?

3 A I was.

4 Q What was your role in that?

5 A I was at that time running our engineering team

6 for search.

7 Q And what was the thesis for Microsoft's

8 perspective about -- relating to entry into that deal?

9 A At that time, that was the big break-through we

10 needed.

11 In fact, I don't think we would be here, quite

12 frankly, even in desktop competing if it was not for the

13 Yahoo! deal, giving us enough of a market position.

14 Again, thanks to, in fact, the intervention, I

15 forget now which, by the DOJ, I think, which helped us, in

16 fact, because they were going to go to Google, because they

17 were going to get marginally better economics than we could

18 ever give them.

19 But since, for some reason, at least at that time,

20 people said, oh, you can go to high share player, that

21 helped us a ton to get to some credible share position and

22 the auction density improved and obviously Yahoo! benefited

23 from it, we benefited from it. So, yes, that was a very

24 helpful thing for us.

25 Q And what was the effect in the years since?

 3522

1 A So we've made us competitive in desktop search.

2 As I said, I think, you know, for us, the Yahoo!

3 supply -- fundamentally search is a supply constraint

4 market. There is Google supply and then there's all of us

5 fighting for scraps.

6 And so in that world, you need supply. And if you

7 can get -- Yahoo!'s supply was the next best thing from our

8 own supply. And so the aggregation of that helped us

9 consolidate and grow into this 20 plus points of share.

10 Q And to stay on desktop for a few minutes, you

11 mentioned earlier that on Windows, there are some access

12 points where Bing has a default; is that right?

13 A That's correct.

14 Q And is that essentially on the Microsoft browsers?

15 A Yes.

16 Q And are you familiar with something called the

17 "jump-start program" on desktop?

18 A Yes, I am.

19 Q What is that?

20 A So it's our OEM incentive program. Essentially we

21 give discounts on Windows pricing for OEMs who carry our

22 defaults.

23 And in that sort of analogy of carrots, this is

24 our carrot to have OEMs carry our defaults. Of course,

25 they're free to take money from others and at that time our
 3523

1 incentive goes away and our Windows price becomes whole so

2 to speak. But that's essentially the problem.

3 Q And how does that program differ from what Google

4 does with Android?

5 A It's similar, I think, in the fundamental

6 construct. I think the advantage Google has is more the

7 carrots and sticks. It's -- we don't have those sticks, and

8 we don't exercise those sticks because it makes no sense.

9 For example, it's not like if somebody didn't

10 carry our defaults, we'd go them and say see Office won't

11 run on Windows. I mean, that would be the modern equal of

12 taking out an app store or saying YouTube would not be

13 optimized, and so on.

14 It's not in our business interest to do that,

15 quite frankly, so, therefore, we don't do it. And, yeah, so

16 we defend our defaults, they defend their defaults, and

17 Google just has inherent advantages of what they can do on

18 Android that we don't have on Windows.

19 Q And you mentioned that Bing has much more share on

20 desktop than mobile. Is there any -- can the share that

21 Bing has on desktop make up for sort of the lack of scale on

22 mobile?

23 A Tough.

24 You know, so -- I mean, there are two questions.

25 If somehow if we could get to our share position, in fact,

 3524

1 that's kind of the mental sort of math I always do every day

2 which is, hey, if I could get to 20 points on desktop, why

3 can't I get to 20 points on mobile. And then the

4 aggregation of those two will improve search quality and

5 that'll get me to 30 points. So that's essentially how

6 I think about it, and that's why we bid on all these

7 distribution deals on mobile.

8 Q I want to show you a document. This is UPX0736.

9 MR. SEVERT: Your Honor, may I approach?

10 THE COURT: You may.

11 BY MR. SEVERT:

12 Q UPX0736 has been marked as confidential by

13 Microsoft, so I'm not going to read it into the record or

14 display it, sir, but I'm going to just sort of ask some

15 general questions about it.

16 And, sir, is this an email chain from 2018 that

17 you participated in?

18 A Yeah.

19 Q And without being too specific, what's the general

20 subject matter?

21 A Yeah, it's the conversations we were having with

22 Apple at that time.

23 Q Okay.

24 And I want to direct your attention to your email

25 on the first page, Bates ending 414.

 3525

1 Do you see there's an email from you sort of in

2 the middle of the page?

3 A Yes, I do.

4 Q And if you wouldn't mind just reading the

5 paragraph to yourself that says, "Also on the transition,"

6 that paragraph.

7 A Yeah, "Also on transition, one way for" --

8 Q Sorry, don't read it aloud. Just read it to

9 yourself.

10 A I'm happy to read it, actually. It captures,

11 I think, yeah, everything I've talked about.

12 Q Yes.

13 So my question is, you make a statement that --

14 I don't think this particular -- I don't think this part is

15 confidential -- that there's no reason why Bing should have

16 any share at all in Windows. What did you mean by that?

17 A So, yeah, it's sort of -- I think what I was

18 referencing there was Windows being an open platform.

19 I mean, if you think about it, like what's my --

20 if somebody said to me, Prove to me that you have search

21 quality, and my sort of simple defense is, Look, why is --

22 yes, we're defending defaults, we're setting defaults.

23 But, quite frankly, given that anytime anybody

24 visits any of the Google properties -- because, remember, on

25 Windows, Chrome is the dominant browser, and there's

 3526

1 significant usage of all the Google properties on Windows,

2 and anytime anybody goes to Gmail or YouTube, they can

3 download Chrome, why do we even have any share.

4 And the only reason why I sort of say is that we

5 have quality. So I was using that as one of the arguments

6 that, Look, we know how to improve quality. We have fought

7 hard to improve quality. We are improving our share

8 monotonically, and so, therefore, we can do the same on

9 mobile.

10 Q You can put aside UPX0736, sir.

11 And I want to circle back to -- the Court asked

12 you some questions --

13 THE COURT: Mr. Severt, can I just ask Mr. Nadella

14 another question.

15 What's your reaction to the following, which is

16 that users do know how to switch, and even on Windows

17 devices, they have switched away from Bing, they've switched

18 away from Edge, they've switched to Chrome and they've

19 switched to Google.

20 Doesn't that -- well, how do you respond to that

21 argument?

22 THE WITNESS: I mean, for sure.

23 I mean, there's no question that users can switch.

24 My only argument against that is users don't

25 switch. And as I said, the best example is, again, take a

 3527

1 look at what happened with Apple Maps. You know, the

2 default was there, people sort of -- there was a lot of

3 press about how, you know, it's got quality issues, this,

4 that, and the other. And I don't have the data, but I would

5 say, my bet, that 90 percent of the users use Apple Maps.

6 And so defaults, that's the biggest thing that

7 I think the broader world sort of doesn't get is, you just

8 can love, love, love a product but will not switch defaults.

9 Defaults are the key things where network effects in digital

10 products come about.

11 And in many other categories, switching happens

12 because buyers are different than the users or what have

13 you, but in this case, it just doesn't happen.

14 MR. SEVERT: Your Honor, I just got a note.

15 I neglected to offer -- sorry, offer UPX0736 into evidence.

16 MR. SCHMIDTLEIN: Your Honor, given the limited

17 scope of the use of the document, which is only statements

18 made by Mr. Nadella and not any embedded hearsay in the

19 document, we would not object to the admission on that

20 basis.

21 THE COURT: All right. So it will be admitted for

22 that purpose.

23 (Plaintiffs' Exhibit UPX0736

received into evidence.)
24

25
 3528

1 BY MR. SEVERT:

2 Q Sir, I want to go back and sort of talk about AI,

3 and ChatGPT. I think we touched on this earlier with the

4 Court.

5 But what was the sort of strategy or thesis around

6 the investment that Microsoft just made in OpenAI and

7 ChatGPT and Bing chat, those products?

8 A Yeah, but I -- I think of this new generation of

9 AI technology with large language models and broad

10 applicability in many domains, but in particular as it

11 relates to search, we wanted to sort of ensure that we could

12 think about innovation in the search category with LLMs.

13 And so when I look at what ChatGPT themselves are

14 doing, ChatGPT today exists with Bing as the index, so

15 that's sort of one entry point, if you will. The other

16 entry point is Bing itself, with Bing chat powered by

17 OpenAI's model.

18 So yes, so we are trying everything possible to

19 see if we can make some headway in this world, and obviously

20 Google's responded pretty strongly as well, and they will be

21 very competitive in this.

22 And as I said, I think my main worry now, other

23 than my early exuberance about an opportunity that we may

24 have here, is, Wow, is this going to be even worse of a

25 nightmare to make progress in search because there's a new

 3529

1 avenue to lock up essentially what's the thing that feeds

2 the power of these LLMs, which is content.

3 Q And were the investments in LLMs, these large

4 language models, was that limited to benefit the Microsoft

5 search products, or is there sort of a broader application?

6 A Oh, it's broader applications.

7 I mean, we make available, and OpenAI makes

8 available, these models for anybody for use in any product

9 from Epic software in healthcare to Bing search.

10 Q And is there any expectation, at least in the

11 foreseeable future, that these LLMs, these ChatGPT products,

12 are going to replace Internet search?

13 A The -- there's two things.

14 I believe the search category by itself will

15 fundamentally change, because there's a new way to think

16 about answering questions using LLMs versus sort of just

17 giving you the 10 blue links, and you're still having

18 citations, still having traffic transfers to publishers,

19 which all matter because otherwise, why would a publisher

20 allow you to index them.

21 So with all that said, I do believe it

22 fundamentally can change if allowed to sort of play out in a

23 way where the content that allows you to create large

24 language models is, again, not exclusive to one player.

25 Q And does Bing chat rely upon search?

 3530

1 A Absolutely --

2 Q How so?

3 A -- yeah.

4 Because that's the way it grounds itself in

5 current information.

6 So you can train a model using sort of crawled

7 content. And then when you are serving the model, you do

8 need to be able to look up what is current data. And so

9 there's a pattern called retrieval augmented generation,

10 that's kind of fundamentally the pattern that Bing uses.

11 Q And to what extent does ChatGPT or Bing chat solve

12 Bing's scale problems on mobile?

13 A They don't do anything for it.

14 So right now we have to learn each -- you know,

15 like, if somebody downloads a Bing app from the App Stores

16 in iOS or Google Play, yes, that will be one helpful benefit

17 for us. But that's the long haul of that, like, you know,

18 yes, that's the grind that we'll continue to grind.

19 Q And you, earlier, a few minutes ago, made

20 reference to a sort of -- I think you said worst nightmare

21 scenario. What did you mean? What were you referring to?

22 A Fundamentally, you know, I'm much more lately

23 very, very focused on is one fundamental thing, which is, in

24 this new age of LLMs, if the competition between model

25 capability -- which is fundamentally determined by two

 3531

1 things, compute and data. If compute will be fixed

2 investments we'll get to make, Google will get to make,

3 others will get to make. So no problem, we're happy to put

4 in the dollars.

5 The other source is data. If we don't have access

6 to the same data because of any type of exclusive access

7 that our competition has, and that would be problematic.

8 And then especially if that exclusive access to

9 content was secured by using essentially the ranks from

10 current advertising power, that would be even more

11 problematic because you just can't -- even being the bid for

12 that exclusive access.

13 Q And would Microsoft need to make sure it had

14 access to this -- to this data in order for the investments

15 that you referenced to make economic sense?

16 A Yeah, I mean, one -- two things.

17 One is, we definitely care as much as the next

18 person about the publishers' rights, copyrights, publishers'

19 sort of need for traffic, and we will do everything on that

20 context.

21 I think the key thing to understand in is this

22 dynamic is in the immediate term, people can use their

23 economic rent to lock up content which we don't have access

24 to. So that's really the real challenge.

25 Q And I recall seeing a number of public statements

 3532

1 that you've made about the promise of large language models

2 and what that could mean for search. Is that -- that's a

3 topic you've talked about?

4 A Yes, I have.

5 Q And one quote I wanted to ask you about, I think

6 at one point you said that "ChatGPT or Bing chat is going to

7 make Google dance." Do you remember that?

8 A Yeah.

9 I mean, look, let's call it exuberance of someone

10 who has like 3 percent share that maybe I'll have

11 3.5 percent share.

12 And, look, Google is a very strong, competitive

13 company, and I'm not confused by it.

14 And they have done a great job of responding to

15 even ChatGPT and Bing chat with Bard and what they're doing

16 with their generative experiences on search.

17 And so, yeah, we'll continue to compete. And to

18 me, let's call it a rallying cry for a 3 percent shareholder

19 to improve.

20 Q And when Microsoft rolled out or announced Bing

21 chat, was there a competitive response from Google?

22 A Oh, yeah. I mean, pretty much immediately they

23 had Bard, and since then, you know, they're competing every

24 day to improve search and Google Assistant, and in all

25 dimensions, yeah, we compete on a broad spectrum of things.

 3533

1 But, yes, on search, I think the competition is

2 pretty intense.

3 MR. SEVERT: Thank you, sir.

4 Your Honor, I pass the witness.

5 MR. McKINLEY: Good morning, Your Honor.

6 Matt McKinley on behalf of Plaintiff States. May I proceed?

7 THE COURT: You may.

8 - - -

9 DIRECT EXAMINATION

10 BY MR. McKINLEY:

11 Q Good morning, Mr. Nadella.

12 As you just heard, my name is Matt McKinley, and I

13 represent the Plaintiff States in this matter. It's nice to

14 meet you.

15 A Good morning.

16 Q I just have a few questions for you this morning.

17 Are you familiar with the term "vertical search

18 engine"?

19 A I am.

20 Q And would a vertical search engine be a company

21 where a user can go to search within a particular search

22 vertical?

23 A That's correct.

24 Q Okay.

25 So something like an online travel agency where a

 3534

1 user can go to find and make, say, a hotel reservation?

2 A That's correct.

3 Q Or an online Website that focuses on local

4 services where a user can search for and find, say, a

5 plumber?

6 A Yes.

7 Q Is there a difference, in your opinion, between

8 general search engines and vertical search engines?

9 A Yeah, I mean going back to a statement I made

10 earlier, on the web, search engines, general search engines

11 have that enormous power because they become the organizing

12 layer, right, especially when you couple it with the browser

13 default and when I go into my URL bar and I type in any

14 term, I first get thrown into a search engine, and then from

15 there, I can navigate to a vertical search engine.

16 And so, yes, the differences that vertical search

17 engines are like search engines but their scope is limited

18 to a particular domain, and but the way the web works is for

19 most often nine out of ten times, you get traffic through a

20 general search engine to a vertical search engine.

21 Q Is it fair to say that vertical search engines are

22 reliant upon general search engines?

23 A Absolutely.

24 Q And vice versa is not true?

25 A That's correct.
 3535

1 Q And over the years, has Microsoft invested in

2 partnerships with vertical search engines?

3 A Yeah, the vertical search engines predominantly

4 try to do one thing, which is SEO, which is they try to do

5 search engine optimization to rank themselves up on organic

6 results and get traffic. That's kind of what any vertical

7 search does.

8 And then on top of that, yes, they can have

9 partnerships with Microsoft or ostensibly Google, I guess,

10 to get essentially distribution in a variety of ways. So,

11 yes, we have partnered with, you know, multiple vertical

12 search engines over the years.

13 Q Why has Microsoft partnered with vertical search

14 engines?

15 A Just as a way for us to bring more innovation and

16 competition, because in some sense, Google -- let's take

17 travel vertical as an example. I mean, Google is building

18 their own. And so sometimes it's sort of more beneficial

19 for us to partner up with others who have content and good

20 search capability in their particular domain. And

21 especially internationally that is definitely the case.

22 Take local search, you know, to be able to compete in all of

23 those locales, it makes a lot more sense for us to partner

24 up.

25 Q So you mentioned travel and local search. Are

 3536

1 there any other examples of the types of partnerships that

2 Microsoft has done over the years with vertical search

3 engines?

4 A I think that it will be a good question for

5 someone who has the specific names.

6 But I recall multiple, I think we've worked with,

7 you know, whether it's the Expedias, the Bookings, or even

8 Yelp and others in a variety of different ways with

9 different arrangements over the years.

10 Q Has Microsoft faced challenges in partnering with

11 vertical search engines over the years?

12 A Yeah, I think vertical search engines care about

13 two things. Again, as I said, they care about getting

14 mostly a lot of organic traffic.

15 And the challenge we have faced is a lot of them,

16 just again, they look at the share position of Bing versus

17 Google and they didn't do enough work to optimize for Bing,

18 and so we worked for them to try and get their traffic share

19 from us to go up proportional to at least our share. That's

20 one area.

21 And then the second area, of course, is there's

22 also the economic incentive, right, which is they need to

23 advertise sometimes in our search engines in order to get

24 traffic as well, and that's another place where they don't

25 prioritize that, because after all, they're prioritizing

 3537

1 Google given of the share advantage Google has.

2 So, yes, they have certain amount of friction in

3 working with us because of our low share, opportunity cost

4 of working with us because of our low share.

5 Q And would that friction or opportunity cost due to

6 the low share also apply to mobile search?

7 A Yeah.

8 Q From your perspective, do Microsoft's partnerships

9 with vertical search engines help improve the user

10 experience on Bing?

11 A It can.

12 At the end of the day, if someone has better

13 content, as I said, even if we're ranking it, it's always

14 helpful.

15 So, yes, we think about whether it's in commerce

16 or in travel or in local or any of these domains. When

17 people have good content sources to be able to partner with

18 them and improve search experience and quality is something

19 that we will always look to do.

20 Q I just briefly want to switch topics.

21 You previously mentioned in your testimony

22 Google's search engine management tool SA360. Do you recall

23 that?

24 A Yes, I do.

25 Q And you testified that Microsoft has not been able

 3538

1 to break through on SA360 and making it such that

2 advertisers can have the same campaigns across multiple

3 search engines.

4 Do you recall that?

5 A I do.

6 Q Can you elaborate on what you mean by that?

7 A Yeah, I'm not, again, deep in details as it stands

8 today, but fundamentally over the years, there's been one

9 goal we've had which is, how can a campaigner click on one

10 button, any advertisers with that campaign click on one

11 button and have it just migrate over to the Bing ad system.

12 And just given the share position of SA360, you

13 just need to be able to enable some of that through their

14 APIs, I would imagine, or their tooling, or even their terms

15 around who owns that user data. And we have run into lots

16 of friction over the years and we keep sort of bringing that

17 up with Google, and I forget now where we are, quite

18 frankly, in the status of all of that.

19 Q And have you heard any complaints internally at

20 Microsoft about the integration of the Microsoft features

21 into Google's SA360?

22 A I think that's the main thing, which is we keep

23 asking for them to add some features that we want and, you

24 know, I think they ask us to go pound sand, so that's kind

25 of where we are.
 3539

1 MR. McKINLEY: No further questions, Your Honor.

2 Thank you, Mr. Nadella.

3 THE COURT: All right. Thank you, Counsel.

4 Mr. Schmidtlein.

5 - - -

6 CROSS-EXAMINATION

7 BY MR. SCHMIDTLEIN:

8 Q Good morning, Mr. Nadella.

9 A Good morning.

10 Q I just want to pick up on a couple of points that

11 counsel for the States just asked you about with respect to

12 SA360.

13 Is that a product that you spent a lot of time

14 overseeing -- or I'm sorry.

15 Do you spend a lot of time overseeing the Bing

16 ads-SA360 relationship?

17 A I don't.

18 Q Have you had any meetings with Google about the

19 SA360-Bing ads relationship?

20 A Not me personally.

21 Q And you didn't really have the details about

22 whatever complaint Microsoft has with Google about the SA360

23 integration issues?

24 A Yeah, I don't have the details. It's just that it

25 keeps coming up in escalations to me that this is one of the

 3540

1 things that is a contended issue between the two companies.

2 Q You sitting here today, you don't know what the

3 status is in terms of the features that Google has built or

4 is building in SA360 of Bing ads features; is that fair to

5 say?

6 A That's fair.

7 Q And you've never spoken with anyone outside of

8 Microsoft, any advertisers, to be particular, about this

9 SA360 issue, have you?

10 A I've not talked about specifically SA360.

11 But whenever I've had roundtables with

12 advertisers, they bring up this opportunity cost, and if we

13 could work as an industry.

14 For them, they want to get best ROI, as I said,

15 you know, whenever anybody's campaign moves over to Bing,

16 the first thing that happens is they get lower prices so

17 they like that, but they don't like the opportunity cost of

18 spending time on it.

19 So in my roundtables with advertisers, which I do

20 sometimes, you know, when I'm in different cities, they

21 bring that up, but at that conceptual level.

22 Q I see.

23 But they haven't brought up to you a complaint

24 that a particular feature of Bing ads is or is not being

25 supported on SA360?
 3541

1 A Not to me personally, but I'm sure to our teams,

2 they'd probably do.

3 Q You can't name a single advertiser who has

4 complained to Microsoft about a failure of SA360 supporting

5 some Bing ads feature, can you?

6 A Not me.

7 Q You also were asked some questions about

8 relationships that Microsoft has had over the years with

9 vertical search providers; is that fair?

10 A Yeah.

11 Q And one of those relationships that Microsoft has

12 had has involved Yelp; is that right?

13 A I believe so.

14 Q And Microsoft no longer has an agreement with

15 Yelp, correct?

16 A I don't know the exact detail of where we are with

17 them today.

18 Q Okay.

19 You're aware that Microsoft and Yelp fell out over

20 economic terms, right?

21 A I don't, actually.

22 Q Okay.

23 Are you aware of the status of any particular

24 vertical agreement between -- strike that.

25 Are you aware of the status of any particular

 3542

1 agreement between a vertical search provider and Microsoft

2 for, you know, inclusion of data on Bing search results?

3 A I am.

4 Q Okay.

5 And is it your testimony that Google has done

6 something to somehow interfere with those relationships?

7 A No. I mean, I -- to all internal issues.

8 Q Okay.

9 THE COURT: I'm sorry, could you repeat that last?

10 THE WITNESS: Our relationship with the vertical

11 search engine, whether we have a partnership with them on

12 content that we include, I've not -- I'm not at least aware

13 of anything that goes through Google on that.

14 THE COURT: Okay.

15 BY MR. SCHMIDTLEIN:

16 Q You've also given some testimony this morning

17 about AI and concerns you have about content providers

18 entering into exclusive agreements with a party that's maybe

19 behind a particular development of an LLM model;

20 is that fair?

21 A That's correct.

22 Q And you're not aware of any issue, today anyway,

23 of an exclusive agreement, are you?

24 A We're in the middle of many conversations. And to

25 me what is sort of concerning is, it sort of reminds me of

 3543

1 what happened with distribution deals in some sense. The

2 reason I say that is because distribution deals, you have to

3 pay to get distribution deals. If you are the largest

4 player with the most economic rent, you get to afford more

5 distribution deals.

6 And so when I am meeting with publishers now,

7 they're coming in and saying, hey, you know, Google is going

8 to write this check to us and it's going to be exclusive and

9 you have to match it.

10 And so we're beginning to see some dynamics which

11 are troubling in the sense, because, you know, an exclusive

12 deal that a player who has more ability to pay for it just

13 because the money they make today could be problematic.

14 Q ChatGPT has been based on publicly available data,

15 correct?

16 A And this is the question, which is, what is

17 publicly available today. Will it be publicly available

18 tomorrow? Is the real issue.

19 Q And one of the issues is that Chat -- or I should

20 say OpenAI has encountered are lots of publishers who are

21 unhappy with the use of their data in ChatGPT, correct?

22 A Yeah.

23 So I think there are two sets of issues. There's

24 copyright issues. There is publishers who want traffic

25 exchange, and all of that is a very, very legitimate, and

 3544

1 the question is how do you address those needs as ChatGPT or

2 Bing chat or bar, or what have you.

3 And then there is a separate thing called

4 exclusive content licenses that are only accessible by one

5 player.

6 Q And ChatGPT hadn't solved for that problem before

7 it went out and launched its product, did it?

8 A Yeah, basically none of us solved for the problem

9 called if -- in a world where most content is not crawlable,

10 how do you build a large language model.

11 Q ChatGPT has been sued by many, many publishers,

12 right?

13 A I don't have the count, but I -- I'll take your

14 word for it.

15 THE COURT: Mr. Schmidtlein, if I could interrupt

16 you. It looks like you may be transitioning to a different

17 topic.

18 MR. SCHMIDTLEIN: Yes.

19 THE COURT: Why don't we go ahead and take our

20 morning break. It's a little before 11:00. We'll resume at

21 11:15.

22 Mr. Nadella, I'll ask you not to discuss your

23 testimony with anyone during the break.

24 THE WITNESS: Absolutely.

25 THE COURT: Thank you, sir.

 3545

1 COURTROOM DEPUTY: All rise. This Court stands in

2 recess.

3 (Recess from 10:59 a.m. to 11:16 a.m.)

4 COURTROOM DEPUTY: All rise. The Honorable

5 Amit P. Mehta presiding. This Honorable Court is again in

6 session. Be seated and come to order.

7 THE COURT: All right. Thank you, everyone.

8 All right. Mr. Schmidtlein, ready when you are.

9 MR. SCHMIDTLEIN: Thank you, Your Honor.

10 BY MR. SCHMIDTLEIN:

11 Q Mr. Nadella, did Microsoft first offer a search

12 engine in 1998?

13 A I don't recall the exact first time we offered a

14 search engine, but maybe we did.

15 Q Do you recall what it was called?

16 A I don't.

17 I mean, at that point, I'm assuming everything

18 must have been in some MSN brand.

19 Q 1998, Microsoft offered a product called

20 MSN Search, didn't it? Does that sound right to you?

21 A Sounds right.

22 Q Okay.

23 And 1998 was the year Google was founded,

24 wasn't it?

25 A Yep.
 3546

1 Q And by 1998, Microsoft was a very wealthy company,

2 wasn't it?

3 A It was.

4 Q It had revenues of over $14 billion, right?

5 A Yes.

6 Q Profits of almost $5 billion, right?

7 A If you say so. I don't recall all those numbers,

8 but I'll take your word for it.

9 Q In 1998, Microsoft Windows was far and away the

10 leading personal computer operating system in the

11 United States, correct?

12 A Yes.

13 Q And even though Microsoft offered a search engine

14 in 1998, Microsoft did not actually build that search

15 engine, did it?

16 A I don't have real detailed recollection of exactly

17 the state of our business in 1998. I was in a very

18 different part of Microsoft, and I've not sort of really

19 gone back to refresh myself exactly what we did or

20 didn't do.

21 Q MSN Search was based on search technology that

22 Microsoft licensed from a third party; isn't that right?

23 A I don't -- I sincerely don't know anything about

24 that time period.

25 Q And in that early time period, Microsoft did not

 3547

1 have its own search advertising technology either, did it?

2 A Probably not.

3 Q Am I correct that it was not until 2005,

4 seven years later, that Microsoft was in a position to

5 launch a search engine built on Microsoft technology?

6 A Sounds right.

7 Q And by 2005, Google had been operating its own

8 search engine and had gone public in 2004, correct?

9 A Okay.

10 Q And when Microsoft finally launched its search

11 engine in 2005, it was still relying on third-party

12 technology for search advertising, right?

13 A Don't know exactly when we built our own auction

14 engine, but it's likely that maybe we partnered with someone

15 there.

16 Q Microsoft ad center did not launch until 2006,

17 correct?

18 A Seems like it.

19 Q Now, by 2006, Google had become so popular in the

20 United States that the Merriam-Webster and Oxford English

21 Dictionaries added the word "google" as a verb, right?

22 A Yep.

23 Q And in 2006, after launching its own version of a

24 search engine, it renamed it to Live Search, right?

25 A I think that's the name we called it first, yeah.

 3548

1 Q And it was variously known as Live Search or

2 Windows Live Search until 2009, correct?

3 A That's right, when we launched Bing, yeah.

4 Q And Bing launched in 2009, correct?

5 A That's correct.

6 Q I believe you testified from '07 to '11, you were

7 the person responsible for overseeing the research and

8 development teams for Microsoft search and search

9 advertising products, correct?

10 A That's correct.

11 Q And when you started in that position in 2007,

12 Windows Live was struggling in the marketing right?

13 A It was.

14 Q There were a lot of things you wanted to fix with

15 it, correct?

16 A Yep.

17 Q And one of them was branding, right?

18 A Yes.

19 Q From 1998 to 2007, would you agree Microsoft had

20 failed to create a so-called destination brand for search?

21 A Yeah, I mean, in retrospect that organizing layer

22 of search as a destination brand became sort of obviously

23 important, to us at least, perhaps a little late, but we did

24 eventually launch Bing.

25 THE COURT: Mr. Schmidtlein, I'm sorry to

 3549

1 interrupt you.

2 Windows Live, that's the name of the search

3 function?

4 MR. SCHMIDTLEIN: That was another one of the

5 predecessor names, I believe, that -- of Microsoft's search

6 product before Bing in 2009.

7 THE COURT: Okay.

8 MR. SCHMIDTLEIN: But, Mr. Nadella, you can

9 correct me if I'm wrong.

10 THE WITNESS: No, that's correct.

11 Basically we had a variety of different names

12 until we decided to prioritize a destination brand very much

13 to the point about google had become a verb. In order to

14 compete against something that's in the English language

15 dictionary, you need a -- a name that's a destination brand.

16 BY MR. SCHMIDTLEIN:

17 Q By 2009, Google had created one of the leading

18 brands in the entire tech industry, right?

19 A Absolutely.

20 Q And by 2007, it was far and away the most popular

21 search engine in the United States, correct?

22 A Absolutely.

23 Q Microsoft at that time was a distant third behind

24 not only Google but also Yahoo!, correct?

25 A That's correct.
 3550

1 Q And Microsoft was a distant third behind Google

2 and Yahoo! when it did its deal with Yahoo! in 2009 you

3 testified about this morning, correct?

4 A That's correct.

5 Q In order to create a destination search brand, you

6 needed a world-class search engine, right?

7 A That's correct.

8 Q And Microsoft did not have that in 2007, correct?

9 A In 2007, we had a -- I would say -- two things, I

10 would say.

11 We had a credible product. We needed to sort of

12 obviously continue to improve. I don't mean -- Google was a

13 very, very good product with a great brand.

14 And so in order to compete, we needed to sort of

15 both meet search quality, but also, I think, one of the

16 things that we came late to was even this idea that you

17 brought up, which is the need for building a destination

18 brand.

19 Q When you took over in 2007, Microsoft was at least

20 three to five years behind having a search product that

21 could be broadly competitive to challenge Google, correct?

22 A I'm sorry, what was the time frame you mentioned?

23 Q In 2007.

24 A In 2007, I felt that -- much more, I would say,

25 ready on the product front, and I felt that we needed to get

 3551

1 much more aggressive on building the brand, building

2 distribution, and those are the things that we subsequently

3 prioritized.

4 MR. SCHMIDTLEIN: May I approach, Your Honor?

5 THE COURT: You may.

6 BY MR. SCHMIDTLEIN:

7 Q Mr. Nadella, I've handed you a binder of

8 documents, and I'd ask that you turn to the tab in the

9 binder that is DX424.

10 And, Mr. Nadella, is this a series of emails that

11 you wrote on or around November and December of 2007?

12 A Yeah, as I see it, that's correct.

13 Q Okay.

14 MR. SCHMIDTLEIN: Your Honor, at this time, we

15 move the admission of DX424.

16 MR. SEVERT: Your Honor, we have an embedded

17 hearsay objection on this email.

18 MR. SCHMIDTLEIN: We have a business records

19 declaration. I don't believe we are going to be treading on

20 any embedded hearsay.

21 THE COURT: Okay.

22 All right. Well, let's see where it ends up. But

23 if that's not the portion of the document you're pointing

24 to, then we'll be all right.

25 Go ahead.
 3552

1 MR. SCHMIDTLEIN: Okay.

2 (Defendant's Exhibit DX424

received into evidence.)
3

4 BY MR. SCHMIDTLEIN:

5 Q If you would take a look at the page -- and

6 there's a -- these have been numbered a couple of different

7 ways. But if you look at the numbering in the middle of the

8 bottom of the page, it's DX -- it says DX424.005.

9 Are you there?

10 A Yes.

11 Q Okay.

12 MR. SCHMIDTLEIN: And, Your Honor, this document

13 has been only -- there are just a couple of partial

14 redactions. We're not going to get into those. So

15 I believe we can put this on the screen, if that's easier

16 for you to follow along with.

17 BY MR. SCHMIDTLEIN:

18 Q On this page -- and this is an email that you sent

19 to Brad Goldberg, Brian McDonald, and Derrick Connell

20 around, it looks like to be, November 27th, 2007; is that

21 right?

22 A Yeah, it does.

23 Q Okay.

24 And on the page that we're looking at, there is a

25 section that says, "Search, help we need from SLT."

 3553

1 Do you see that?

2 A I do.

3 Q And what does SLT stand for?

4 A The senior leadership team.

5 Q It says, "Don't blink on the investment. We are

6 at least 3 to 5 years behind to be broadly competitive on

7 the product."

8 Do you see that?

9 A I do.

10 Q And you wrote that in November of 2007?

11 A I did.

12 Q And that was accurate in November of 2007,

13 correct?

14 A Yeah, I mean, it -- just to be very transparent,

15 I mean, we have a culture of a lot of self-critical stuff.

16 And so, yes, I do feel like, you know, I would say even that

17 today, we need to do more work, and we are behind Google.

18 So in some sense, that's the spirit of sort of me

19 trying to get into my new job and work it.

20 Q You didn't have a good brand and you didn't have a

21 reason for people to seek out Microsoft to search in 2007,

22 right?

23 A That is correct.

24 Q And another one of the problems you found when you

25 started in 2007 was the engineering head count, correct?

 3554

1 A It's very likely that I was looking to increase

2 our investment, for sure.

3 Q In fiscal year 2007, Microsoft estimated that

4 Google had more than double Microsoft's search head count,

5 correct?

6 A Yeah, I'm assuming there's a document to that

7 effect.

8 Q If you'll turn to DX423. And if you'll look at --

9 this is also an email exchange -- and I'm really focusing on

10 the second page of it.

11 A 002?

12 Q That's correct, sir.

13 And this is an email you sent to Kevin Johnson and

14 cc'ing a number of other people in July of 2007; is that

15 right?

16 A Uh-huh, I do.

17 MR. SCHMIDTLEIN: Your Honor, at this time, we

18 move the admission of DX423.

19 MR. SEVERT: We have a similar embedded hearsay

20 objection to this exhibit as well.

21 THE COURT: Doesn't sound like it's an issue since

22 he's pointing to his own, you know.

23 MR. SCHMIDTLEIN: Correct.

24 (Defendant's Exhibit DX423

received into evidence.)
25
 3555

1 BY MR. SCHMIDTLEIN:

2 Q This is -- first of all, who's Kevin Johnson?

3 A He at that time was my direct boss who ran sort of

4 the division in which I worked in.

5 Q Okay.

6 And when you look at your email there, you write

7 to him so that -- and you want him to see Microsoft's gap

8 with Google, correct?

9 A Yes, I do.

10 Q And you lay out for him your estimates of head

11 count, the search group head counts from Microsoft, Yahoo!,

12 and Google, right?

13 A Yes, I do.

14 Q And Microsoft is trailing both Yahoo! and Google,

15 correct?

16 A That's correct on these numbers, yeah.

17 Q And in fiscal year 2007, the numbers indicate

18 Google's got more than twice as many -- twice as much head

19 count on search than Microsoft, right?

20 A Yeah.

21 Q And you were forecasting for fiscal year 2008,

22 it's going to go up to 3 to 1, right?

23 A Yep.

24 Q And, in fact, over the years, this head count gap,

25 search engine years has only gotten worse for Microsoft,

 3556

1 right?

2 A Yeah.

3 I mean, just to make sure, the spirit of the way

4 these head count things is, it's just our estimate.

5 In fact, one of the big things I did was I shut

6 down things like our book search and so on, because I sort

7 of focused our energy with the head count we had on things

8 that mattered, because you can be investing on a lot of

9 things and then things that really, really matter and needed

10 moving things.

11 So, yeah, like, I have head count games inside

12 large organizations like ours, I was playing it.

13 Q By 2018, did Microsoft estimate the Google Search

14 engineering team was five times bigger than Bing's?

15 A I'll take your word for it.

16 Q If you'll turn in your binder to DX524. And,

17 Mr. Nadella, I'll advise you that your counsel has asked

18 that this document be treated confidential in full, so we're

19 not going to show it on the screen and I'm not going to try

20 to elicit from you particular specific numbers because

21 I want to respect that.

22 A Okay.

23 Q So I want to advise you of that before you take a

24 look at it.

25 A Okay. Great. Thank you.

 3557

1 Q This is -- now you're on this email exchange, but

2 this email exchange involves a gentleman named Jordi Ribas?

3 A Yes.

4 Q And what position did Mr. Ribas have in August of

5 2018?

6 A Mr. Ribas runs one aspect of our core search

7 effort.

8 Q He would be somebody that you would rely upon to

9 be able to make accurate estimates of at least with respect

10 to Bing's head count for search, right?

11 A Yes.

12 Q And if you look at this email, he makes certain

13 comments and statements comparing the size of Google's

14 search engineering versus Microsoft's search engineering,

15 correct?

16 A I do.

17 Q And he goes on to say, "It's probably even higher

18 since I bet a significant portion of their 10K is for Google

19 Search."

20 Do you see that?

21 A Yes, I do.

22 Q Sir -- and if you look in the last sentence there,

23 he calls out Milan.

24 Who is Milan?

25 It seems like it's the person who's cc'ed on the

 3558

1 document.

2 A Yeah, I don't recall what Milan did.

3 Q Okay.

4 He's asking Milan to include a sentence in a

5 document or a presentation that they're making that talks

6 about the investment in Bing and whether that would be

7 sufficient for Bing to be competitive in both U.S. and

8 international market.

9 Mr. Nadella, do you have any reason to question or

10 doubt any of Mr. Ribas' statements in this document?

11 A No, I don't.

12 Q Do have an estimate today as to the disparity

13 between search engineering resources between Bing and

14 Google?

15 A I don't.

16 The one thing, as I look at all these numbers

17 again, is we've gotten much better at getting, again,

18 focused on -- you know, like it is brought up on

19 international markets, for example. We prioritized our U.S.

20 market, so in some sense we sort of said, let's hunker down

21 on things we can really make a difference in and then

22 concentrate our head count forces. So I don't have an

23 estimate as to exactly how we match up; and at this point,

24 in fact, with all the LLM investments and so on, it becomes

25 even harder to estimate.

 3559

1 Q You don't have any reason to believe that Bing had

2 closed the gap on head count with Google on search

3 engineering, do you?

4 A I don't -- I don't think so.

5 In fact, the thing that we are trying to make sure

6 is in a very focused way in core search be competitive in

7 few segments with the option that get -- once we have more

8 distribution, we can go accelerate other areas.

9 Q This is the -- give us the distribution first and

10 then we'll invest afterwards, right?

11 A Or coincident with it.

12 So it's not, give us distribution, it's co-own

13 distribution and use that as -- because you sort of need --

14 the only reason I bring up distribution is not that we just

15 want somebody to hand us distribution for free. We are

16 talking about pay for distribution, which then gives us some

17 long-term guarantee that we will have traffic. And so,

18 therefore, it makes economic rationale for us to invest.

19 Q We'll come back to that.

20 During the time that you were in charge of search,

21 Microsoft also was significantly under-invested in hardware,

22 correct?

23 A In hardware, you said?

24 Q Yes.

25 A Okay.
 3560

1 Q Well, did Microsoft have enough servers to support

2 the search business it was forecasting?

3 A Oh, got it.

4 So our capital investment in support of serving

5 Bing traffic.

6 Yeah, I mean, there are two sets of things even

7 there, which is the fixed cost and then there's the variable

8 cost.

9 The variable one is based on, obviously, your

10 search share in traffic, and then the fixed one is what you

11 need to play.

12 And in all these cases, we continuously keep

13 having the argument about how to increase fixed cost and

14 then how to be ready for any search gains -- or share gains.

15 Q The forecast that Microsoft was making during the

16 time period that you were overseeing search, this '7 to '11

17 period, Microsoft didn't have enough servers to support even

18 the forecasted growth you were hoping to get, correct?

19 A You know, each -- I felt, you know, we definitely

20 were pushing for more and more investment, but I felt well

21 supported and well funded to keep continuing to improve the

22 product and compete.

23 But, yes, I'm sure there's lots and lots of emails

24 where I'm asking for more, I'm asking for -- advocating for

25 our team to get the investment required.

 3561

1 Q Over the years, I think. Let me -- strike that.

2 You made reference this morning to capital

3 expenditures as one of the pieces that you need to support

4 in order to build a search business, correct?

5 A Yes.

6 Q And over the years, Bing's capital expenditures or

7 Microsoft's capital expenditures devoted to Bing have been

8 consistently a small fraction of the capital expenditures

9 that Google has made on search, correct?

10 A I would assume so because at some level it's very

11 proportional to your share in traffic, and if you're like,

12 you know, 3 percent share versus 97 percent share, yes,

13 that'll show up in capital expense.

14 Q Microsoft's failure to make capital expenditures

15 to support its search business wasn't because Microsoft

16 lacked the financial resources to do it, right?

17 A The distinction that needs to be brought is two

18 sets of things. What is the fixed cost capital expense and

19 what is the cost of capital for serving the traffic. So one

20 is what's the difference in scale, which you can use share

21 as a proxy for it.

22 And then the second is the fixed cost. And I felt

23 that we were making, and over the years have made, and

24 continue to make, all the fixed cost investments, with

25 tradeoffs; we pick countries, we pick index size, we pick a

 3562

1 particular angle. And then the variable expense comes with

2 more traffic and share gains.

3 Q I'd ask you to take a look in your binder at

4 DX500.

5 And there were some -- there are some native files

6 that were attached to this, so you should be able to see

7 towards the back of the document, I'm hoping -- yeah, past

8 the blue sheet.

9 If you look at the, one -- the fourth page,

10 there's a "Cap X spend sheet"?

11 A Yeah.

12 Q Do you see that?

13 Okay.

14 And, again, Your Honor, we offer DX500.

15 MR. SEVERT: We have an embedded hearsay

16 objection.

17 MR. SCHMIDTLEIN: I'm not sure what embedded

18 hearsay there is. These are capital expenditures numbers

19 that Microsoft is estimating.

20 MR. McKINLEY: We'll withdraw, actually.

21 THE COURT: Okay. So DX500 and DX500A will be

22 admitted.

23 MR. SCHMIDTLEIN: Okay.

24 (Defendant's Exhibits DX500 and DX500A

received into evidence.)
25
 3563

1 BY MR. SCHMIDTLEIN:

2 Q And, Mr. Nadella, I just want to caution you, your

3 counsel has asked for full confidentially at least over this

4 part document so I'm not going to ask you about particular

5 numbers.

6 But this is a -- this document here purports to

7 estimate Google's capital expenditures versus Microsoft's

8 during the period 2005 to 2015 with respect to Bing,

9 correct?

10 A Uh-huh, yes.

11 Q And you don't have any reason to doubt the

12 accuracy of these numbers, do you?

13 A Yeah, as represented here, absolutely not.

14 And I think the one thing that all these estimates

15 are, this is the hard part about, yeah, how do you separate

16 out what is the YouTube investment, what's the Google Search

17 investment. So our teams, you know, obviously, to me, to

18 make the arguments that I should spend more would make the

19 argument of picking the largest number, and it's okay.

20 Q You're not suggesting that your people were

21 goosing the numbers, right?

22 A Absolutely, they were. Everybody would goose the

23 numbers in order to get more investments.

24 Q And you think this isn't an accurate depiction of

25 the difference between Bing and Google Search?

 3564

1 A Like any manager who sees head counts requests,

2 capital requests, use your judgment and say, you know,

3 discount it.

4 Q And this is a document from 2018, correct?

5 A Yes.

6 Q By 2018, Microsoft was one of the wealthiest

7 companies in the world, correct?

8 A Yes.

9 Q It certainly had the resources to make additional

10 capital expenditures for Bing, right?

11 A Absolutely.

12 Q And it was one of the wealthiest companies in the

13 world from 2010 to 2015, correct?

14 A Yes.

15 Q Now, you testified earlier this morning about

16 Bing's distribution and the efforts Microsoft has made to

17 get distribution. And I take it you're not suggesting that

18 Microsoft doesn't have adequate distribution on desktop

19 computers in the United States; is that right?

20 A That's correct.

21 Q Okay.

22 To the extent you've got issues with distribution,

23 it's on mobile, right?

24 A That's correct.

25 And also on desktop. I'm grounded on the fact

 3565

1 that, you know, we have low share of our browser where

2 Google Chrome has.

3 But, yes, to your fundamental point about we have

4 the ability to set defaults on desktop is -- is accurate.

5 Q Yeah, no, I'm going to come back to the desktop.

6 I want to talk a little bit about mobile.

7 In 2008, Microsoft entered into a search

8 partnership with Verizon, correct?

9 A 2008.

10 Yeah.

11 Q And it was a five-year mobile deal that provided

12 that Bing would be the exclusive search engine preloaded on

13 devices that were covered by that deal, right?

14 A That's correct.

15 Q And Verizon was forbidden from providing or

16 promoting non-Microsoft search services on any of the

17 devices if the service primarily returned general search or

18 web results, right?

19 A That's correct.

20 Q Now, various of these Verizon devices that were

21 preloaded with Bing were panned by various tech reviewers,

22 right?

23 A I don't have all of the details but, yeah, I mean,

24 these were very different times, even. It was sort of --

25 IPhone was just coming out, I would imagine.

 3566

1 And what is mobile search today was -- is what

2 mobile search was in that period as a habit. And even the

3 browsers and everything else, you know, the speeds, were,

4 I think what, 2G, 3G, so it was a different world, but, yes.

5 Q And one of the things the tech reviewers focused

6 on, particularly with respect to the Android devices that

7 had Bing preloaded, was their dissatisfaction that Google

8 wasn't the default search engine on the Verizon Android

9 device, right?

10 A I don't have the detail, but I'm sure there's a

11 document.

12 MR. SCHMIDTLEIN: Your Honor, we're not going to

13 move this into evidence but --

14 BY MR. SCHMIDTLEIN:

15 Q Mr. Nadella, I'll represent to you that what I've

16 handed you here is a Washington Post article from

17 September 25th, 2010, and it has to do with the Verizon

18 Fascinate mobile phone, which was an Android device.

19 Do you recall this device?

20 A I don't.

21 Q Okay.

22 The -- and you can put this up.

23 The very first sentence said, "A Google phone

24 handcuffed to Microsoft's Bing -- at Bing's search engine

25 sounds like a bad joke. But it's a real product, one for
 3567

1 which Verizon Wireless charges 199.99.

2 You were aware at this time -- you were overseeing

3 Bing search. You were aware that these Verizon devices that

4 came with Bing as the exclusive preloaded search engine were

5 encountering bad tech reviews like this, weren't you?

6 A Yeah, we saw -- saw a lot of this.

7 In fact, a lot of this even informed a lot of my

8 conversation even subsequently with even Apple.

9 Knowing that, look, I think you need to have a

10 good product, and more importantly, you need to be able to

11 get through this bad-press period and turbulence and to get

12 to the other side, because in some sense, you have to

13 persist to change.

14 And you sort of mentioned it very well, when

15 Google's such a user habit that is well ingrained, that

16 there will be a little bit of, both from the intelligentsia

17 to the users, a lot of change, management that you have to

18 persist through.

19 Q A habit that was built on having the leading, best

20 product in the market, correct?

21 A You know, look, I look at it and say, I'm focused

22 on building a product against Google which is all that and

23 more, as you said.

24 And so, therefore, in order to compete with it,

25 you have to compete for distribution, you've got to put up

 3568

1 with press like this, and then get to the other side with a

2 better product ultimately.

3 Q But the problem you found was that users of these

4 Verizon devices, they didn't search with Bing, they searched

5 with Google even when Bing was preloaded as the default,

6 right?

7 A I mean, it happens even today, and so, therefore,

8 the question is, Can you build a product that retains more

9 each day.

10 Q The Verizon Microsoft deal was a disaster for

11 Microsoft, wasn't it?

12 A We -- it -- definitely financially. It shows,

13 I would say there are two things.

14 Yes, it was a disaster financially, but it shows

15 our willingness to keep trying to invest to get

16 distribution.

17 Q It was a disaster for Microsoft because Microsoft

18 made certain guarantees to Verizon that turned out to be

19 economically bad because people didn't use Bing, they used

20 Google, right?

21 A Yeah, because we couldn't retain the users at the

22 rate at which we needed to.

23 Q Let's talk about another example.

24 Do you remember that in 2011, Microsoft announced

25 a partnership with Research in Motion, or RIM, right?

 3569

1 A In 2011?

2 Q Yes, sir.

3 A I don't recall it, but I'm sure we did.

4 Q I mean, you remember in general that there was a

5 time when Microsoft had a deal with RIM.

6 A Could be.

7 Q And RIM was the maker of the very popular

8 BlackBerry device, right?

9 A That's right.

10 Q And pursuant to this deal, Bing was made the

11 exclusive search provider on various BlackBerry devices,

12 right?

13 A Uh-huh.

14 Q But that deal didn't work out well for Microsoft

15 either, did it?

16 A Maybe not.

17 I'm sure you'll produce a document to that effect.

18 Q Well, BlackBerry users also preferred to search

19 with Google rather than Bing, right?

20 A If you say so.

21 Q Well, they switched from Bing to Google on their

22 BlackBerry devices even though Bing was the default, right?

23 A If you have Google as the most dominant search

24 engine that every user in the world knows about, anytime you

25 introduce anything, there will be switchers.

 3570

1 And you're right in saying that its onus is on us

2 to build a product that retains more, and we'll keep trying

3 until we sort of break through on it.

4 Q Sir, when you used the word "dominant," another

5 word that you could substitute for that would be popular,

6 correct?

7 A I use the word "dominant" sort of because I

8 compete against somebody who's dominant.

9 Q In 2011, Google was far and away the most popular

10 search engine in the United States, correct?

11 A But I'm competing against, call it popular or

12 dominant, to me, I'm competing against somebody who's got

13 97 percent share.

14 Q In 2011, Microsoft also struck a deal with Nokia,

15 right?

16 A Yes.

17 THE COURT: Sorry, what was the year again,

18 Mr. Schmidtlein?

19 MR. SCHMIDTLEIN: 2011.

20 BY MR. SCHMIDTLEIN:

21 Q And in 2011, Nokia was one of the largest

22 smartphone manufacturers in the world, right?

23 A I believe so.

24 Q And under that deal, Google -- or Bing would power

25 Nokia's search services across all of Nokia's devices,

 3571

1 correct?

2 A Yeah, I don't have the details, but I would assume

3 that you're correct.

4 Q And Bing wasn't any more popular on Nokia's phones

5 than it was on the Verizon and on the RIM phones, was it?

6 A Probably not.

7 Q And on top of all that, Microsoft had its own

8 phones during the 2007 to 2011 time period when you were in

9 charge of search, right?

10 A Yes, we did.

11 Q Microsoft had a mobile operating system that was

12 called Windows Mobile; am I right?

13 A That's correct.

14 Q And was that later changed to be called Windows

15 Phone?

16 A That's correct.

17 Q And Bing was the default search engine that came

18 exclusively preloaded on Windows Mobile and Windows Phone

19 devices, correct?

20 A That's correct.

21 Q But even Windows Mobile users searched a lot more

22 on Google than they did on Bing, right?

23 A Could be.

24 Q Well, you knew that, you were in charge of the

25 product at the time?

 3572

1 A Yeah.

2 I mean -- yeah, Google -- you don't have to keep

3 repeating to me that Google has high share. Yes, Google has

4 high share and all through that period and today too.

5 Q They had high share in that period even though you

6 and Microsoft had the Bing defaults on all these devices.

7 A They have high share today on desktop on Windows.

8 Q All of these mobile distribution agreements that

9 Microsoft entered into in this 2009, '10, '11, '12 time

10 period, they all failed to get Bing's scale because users

11 preferred Google, and the large majority of users of these

12 devices searched on Google instead of Bing, right?

13 A Users do switch.

14 Q They churned off of Bing. That's an industry

15 term, right?

16 A Yep.

17 Q If you'll turn in your binder to a document that's

18 marked DX440.

19 A Sorry, 4 --

20 Q 4-4-0.

21 MR. SCHMIDTLEIN: And, Your Honor, this document

22 has already been admitted into evidence.

23 THE COURT: Okay.

24 BY MR. SCHMIDTLEIN:

25 Q And, Mr. Nadella, this is a series of emails. You

 3573

1 are on, I believe, the email, sort of the second email from

2 top to bottom, that included a Bing mobile behavioral

3 analysis from September of 2010.

4 Do you remember this?

5 It was a long time ago.

6 A It's a long time ago.

7 Q All right.

8 If you would turn your attention to the document

9 that is -- the DX number in the middle there again is .011.

10 It will be within the slides.

11 A 011. Yeah.

12 Q And I'll put it up on the screen because that

13 might be easier.

14 And the topic of this slide is "Bing Mobile is a

15 Small Percentage of Overall smartphone Queries."

16 And you see at this time, the install base in U.S.

17 at the top sets out four -- sort of the four leading mobile

18 platforms at that time; is that fair?

19 A Yep.

20 Q And you've got the iPhone, you've got RIM we just

21 talked about, Windows Mobile we just talked about, and

22 Android, right?

23 A Yep.

24 Q So 2010 -- just to orient everybody here, iPhone

25 launched in 2007, right?

 3574

1 A Yep.

2 Q Android launched in 2008, right?

3 A Yes.

4 Q So we're a couple years into sort of this next

5 smartphone generation; is that fair?

6 A Yep.

7 Q Okay.

8 And what you notice here is that RIM, who we just

9 talked about that Microsoft had a deal with, RIM is actually

10 bigger than the iPhone, right, in terms of users?

11 A Yep.

12 Q And Windows Mobile we just talked about, at this

13 time, it's bigger than Android, right?

14 A Yes.

15 Q But if you go over to the -- and there's a lot of

16 numbers and a lot of percentages on this, and I'm not going

17 to ask you to do all the math, but if you go over to that

18 first number one, that blue call-out on the right side,

19 do you see where I am?

20 A Yes.

21 Q It says, "Bing and Yahoo! share is about

22 10 percent despite default deals; Google is super dominant

23 in 90 percent."

24 So this is -- this document is sort of reflecting

25 a little bit of what we've been talking about, which is

 3575

1 Google is getting the overwhelming majority of search

2 queries even though Bing and Yahoo! are the defaults on

3 various of these devices, right?

4 A That's correct.

5 Q And I will ask you to do just a wee bit of math

6 with me.

7 On the Windows Mobile column there, going down

8 from top to bottom, it says, "Google queries, Bing queries,

9 Yahoo! queries."

10 Do you see that?

11 A Yep.

12 Q And it talks about the client and browse.

13 Do you see that?

14 A I do.

15 Q And do you know what client and browse in that

16 context means?

17 A I think so. I think it's the browser and the

18 client app, I think.

19 Q Okay.

20 And, again, I'm not going to do the precise math,

21 but comparing the queries that Google is picking up under

22 both the client and browse and comparing that to the Bing

23 queries that are being picked up there, according to my

24 math, Bing has got about 7 to 8 percent, and Google's got

25 about over 90 percent; is that fair?

 3576

1 A Yeah.

2 Q Okay.

3 Now, if you turn to the next -- two pages next.

4 A By the way, that's not that different from today's

5 desktop search, too.

6 Q Understood.

7 I promise we're going to get to desktop search.

8 Now, going two pages ahead, now, this slide is

9 focusing on RIM, okay?

10 A Uh-huh.

11 Q And it says, "Bing mobile browse on RIM achieves

12 low penetration to date, while client has almost no uptake;

13 however, users do search ... using Google," right?

14 A Uh-huh.

15 Q And, again, this is examining RIM devices with --

16 that have different defaults depending on the carrier,

17 right?

18 A Yep.

19 Q Because in the United States, it's often the case

20 that the carrier dictates what software applications come

21 preloaded on the device, not the OEMs; is that right?

22 A Yeah.

23 And from time to time, it changes, but, yes.

24 Q Currently, the overwhelming majority of Android

25 devices in the United States are controlled by, sort of, and
 3577

1 sold by the carriers, not by the Android OEM; is that fair?

2 A Yes, that's correct.

3 Q So depicted here, you see a RIM device sold by

4 Verizon. We'll work from left to right.

5 The default there was Bing, correct?

6 A Uh-huh.

7 Q For that device, even though Bing is the default,

8 Google gets 91 percent of the queries, right?

9 A Yep.

10 Q Bing only gets 8 percent?

11 A Yeah, which is 8 percent more than any other

12 device, yeah.

13 Q Without the default, Bing would have no share,

14 right?

15 A That's correct.

16 Q And then we look at the AT&T, T-Mobile examples,

17 now those were carriers where -- did Bing try to get the

18 default on those carriers?

19 A Bing may have tried, but we obviously didn't get

20 them.

21 Q And Yahoo! was successful in winning those deals,

22 right?

23 A Yep.

24 Q And it was very common at this time period for

25 Bing and Yahoo! and Google to be slugging it out to try to

 3578

1 get that default, right?

2 A Yep. Yes.

3 Q And in this instance, Yahoo! won, but, again,

4 Google gets the overwhelming majority of the queries, right?

5 A That's correct.

6 Q People switched from the default and used Google,

7 right?

8 A That's correct.

9 Q Now, I want you to go and look at -- if you'll

10 take a look at the page that is DX, and it ends in 023.

11 So another problem that Bing was suffering during

12 this time period, and has suffered over time --

13 A Sorry.

14 Q Sorry.

15 A DX what?

16 Q It's -- 023 are the last three.

17 And that 440.023.

18 It's page 20 on the actual document.

19 A Got it.

20 Q That's easier.

21 A Yep.

22 Q And it's also on the screen if that's easier for

23 you to see.

24 Another problem that Bing was experiencing, and

25 has experienced over time, is people churning the product

 3579

1 very early on when it gets exposure, right?

2 A Yep.

3 Q And as this slide says, "First impression matters.

4 Bing mobile users search very few days before leaving."

5 Do you see that?

6 A Yeah, I do.

7 Q And you recall that was a problem that you had

8 when you were overseeing the search group from 2007 to 2011,

9 correct?

10 A Yes.

11 Q Over half of new iPhone and RIM users, across

12 clients and browse, only try the Bing mobile product for one

13 day before churning. One day. And that's something you've

14 experienced over the years, people churn off of Bing very

15 quickly, right?

16 A Yep.

17 Q And this is one of the concerns Apple had about

18 switching from Bing -- or from Google to Bing, correct?

19 A That's correct.

20 Q Now, you can put that one aside.

21 As I promised you, I want to talk about desktop.

22 A Okay.

23 Q Microsoft has been able to win distribution on PCs

24 by using the economics of its Windows license with OEMs,

25 correct?
 3580

1 A That's correct.

2 Q And Microsoft gives OEMs a generous discount on

3 their Windows PC license fee if they agree to preload Bing

4 and Edge as the exclusive search engine and browser on the

5 PC, right?

6 A That's correct.

7 Q And the program that you talked about earlier

8 today was called Jumpstart?

9 A That's correct.

10 Q And Jumpstart was launched in the early 2010s,

11 does that sound right?

12 A Probably. Sounds right.

13 Q Over ten years ago, right?

14 A Probably, yeah.

15 Q It's been wildly successful for Microsoft,

16 correct?

17 A Yeah, it's our best way to incent people on our

18 defaults.

19 Q The overwhelming majority of PCs sold in the world

20 are covered by the Jumpstart program, right?

21 A That's correct.

22 Q Dell's covered?

23 A I think so.

24 Q Hewlett Packard's covered?

25 A I think so.
 3581

1 Q Lenovo is covered?

2 A Okay.

3 Q Acer is covered; is that right?

4 A Okay.

5 Q Asus is covered?

6 A Okay.

7 Q Those five OEMs collectively account for over

8 three quarters of the personal computers sold in the world,

9 right?

10 A Yeah, I think so.

11 Q And there are dozens more Jumpstart participants

12 that are smaller OEMs all over the world, correct?

13 A Yep.

14 Q To the best of your knowledge, Google Search and

15 Chrome are not pre-installed on any Windows PC in the

16 United States, correct?

17 A Yeah, not pre-installed, that's correct.

18 Q That has been the case for many, many, many years?

19 A I don't recall exactly when Google stopped doing

20 distribution deals of their own, so that's someone else

21 should find that out.

22 But, you know, in some sense, they now have the

23 dominant share of the browser, and so, therefore, they don't

24 need to pay, and we have to.

25 Q Well, you have had the exclusive preloaded

 3582

1 browser, which today is Edge, right?

2 A That's correct.

3 Q Edge defaults to Bing, right?

4 A That's correct.

5 Q And you've had the exclusive search engine, Bing,

6 preloaded on all these Windows PCs, right?

7 A That's correct.

8 Q And that has been the case for many years?

9 A That's correct.

10 Q Now, nevertheless, Google gets the large majority

11 of search queries from users of Windows PCs, right?

12 A That's correct.

13 Q I believe you told -- I think you testified

14 earlier today that Bing's search queries share on

15 Windows PCs was in the teens --

16 A Uh-huh.

17 Q -- is that right?

18 Is that your best --

19 A Yeah, that would be my best guess.

20 Q Okay.

21 And is Google still the most commonly queried word

22 on Bing?

23 A Yeah.

24 Q And in terms of -- I want to switch gears just a

25 little bit and I want to talk a little bit about browsers,

 3583

1 because I know you've talked about those as well.

2 A Okay.

3 Q Let me just finish up.

4 So just to be clear, Google is getting the

5 overwhelming majority of search queries on Windows PCs

6 because users are actively going and downloading either

7 Google Search or Google Chrome onto their Windows PC,

8 correct?

9 A That's correct.

10 And also, it also reflects, I think, the dynamic

11 on Windows, which we like, which is, it's sort of an open

12 platform. Even though, yes, we have defaults.

13 To your point, once somebody types Google on Bing

14 and they get to Google.com, the first thing they get

15 obviously is download Google Chrome, and that's fine by us.

16 It's sort of the most friction free, because no one needs to

17 come to Microsoft to distribute payload to Windows.

18 Q Does Microsoft offer users the ability to download

19 Edge or Bing if somebody goes -- navigates to the Bing

20 website?

21 A We do.

22 Q You do the same thing Google does, right?

23 A That's on Windows, yes, where we can.

24 Q If someone were to type Bing into Google and get

25 sent to Bing --
 3584

1 A We would do the same.

2 Q You would do the exact same?

3 A Absolutely.

4 THE COURT: Just to be clear, you mean type Bing

5 into Chrome -- or into the --

6 MR. SCHMIDTLEIN: Either into Chrome or if you

7 were at the Google home page.

8 THE COURT: Right.

9 MR. SCHMIDTLEIN: But, yes, Your Honor.

10 BY MR. SCHMIDTLEIN:

11 Q Now, Chrome is the most popular browser on Windows

12 today, correct?

13 A That is correct.

14 Q Now, Microsoft's Internet Explorer had once been

15 far and away the most popular browser on Windows PCs, right?

16 A That's right.

17 Q It once accounted for the overwhelming majority of

18 browser usage share in the United States?

19 A That's correct.

20 Q And that was, in large measure, the subject of the

21 United States v. Microsoft case 20 years ago, right?

22 A That's correct.

23 Q Microsoft's actions to get Internet Explorer

24 distributed to Windows users?

25 A That's correct.
 3585

1 Q Now, when you took over, in 2007, the search

2 group, you recognized that Internet Explorer was a poor

3 browser?

4 A In 2007, I don't remember. Maybe I was critical

5 of even our own browser, but ultimately, yes, we had to sort

6 of really abandon, effectively trying to build an alternate

7 rendering engine and just join the Google sort of rendering

8 engine, and that's kind of when we started making progress

9 again on the browser.

10 Q Well, in 2007, Internet Explorer was still the

11 dominant browser, right?

12 A Yeah.

13 Q But Microsoft hadn't invested enough in making

14 Internet Explorer a better browser?

15 A I mean, Google -- you know, if the point is Google

16 made significant investment in Chrome to make Chrome

17 popular, that's absolutely accurate. And they had all the

18 incentives to do so in order to be able to control their own

19 destiny on Windows, given the search property.

20 Q Microsoft failed to understand that having a

21 really, really good browser could affect users' experience

22 on search, right?

23 A That's absolutely right.

24 Q And Microsoft was caught sleeping when Google

25 introduced Chrome, which was a far superior browser to

 3586

1 Internet Explorer?

2 A Yeah.

3 I mean, I think there are two things.

4 One is, they were able to introduce -- like even

5 if we wanted to do the same today on mobile platforms with

6 some new innovation in browsing, it's just not possible

7 because of the way the restrictions on what browsers are

8 allowed and not allowed and what rendering engines are

9 allowed or not allowed, different browsers on Android.

10 Yes, Google did a good job of innovating in the

11 browser, and they took advantage, which we're okay with, of

12 Windows being an open platform.

13 Q People, just like they went and downloaded Google

14 Search onto those PCs, they went and they downloaded Chrome,

15 too, right?

16 A That's correct.

17 Q And they did that because Chrome was better than

18 Internet Explorer, correct?

19 A And we had no restrictions.

20 Q They did it because Chrome was better than

21 Internet Explorer, correct?

22 A I'm not going to agree to that because at some

23 level, the real test would have been if we had iOS-like

24 policies on Windows, then that would have been the real test

25 of how much better was their product.

 3587

1 Q Sir, in -- when Chrome launched in 2008, Microsoft

2 had all of the search defaults on PCs, right?

3 A Yeah.

4 Q That's where all the searching, that was where

5 almost predominantly all the searching was being done, was

6 on PCs, right?

7 A Yeah, and Google had toolbars on our browser.

8 Q And Microsoft failed to innovate with Internet

9 Explorer, and Chrome beat it in the market, correct?

10 A I mean, Chrome did beat us in browser share on

11 Windows desktop, yes.

12 Q And Google did that by getting people to download

13 Chrome onto the desktop because Chrome wasn't preloaded

14 anywhere on Windows.

15 A That's correct.

16 Q Even though Microsoft was an open platform, you

17 didn't get -- if you bought a Windows PC, it didn't come

18 with Chrome, did it?

19 A That's correct.

20 Q Now, Microsoft finally introduced Edge in 2015;

21 is that right?

22 A That's correct.

23 Q But by that time, Chrome was far ahead of Internet

24 Explorer, right?

25 A That's correct.
 3588

1 Q And Microsoft did not introduce a version of Edge

2 for Android until 2017; is that right?

3 A Probably, yeah.

4 Q And you didn't introduce a version of Edge for iOS

5 until 2017, correct?

6 A Sounds accurate.

7 Q Microsoft never even made a version of Internet

8 Explorer that could be used on Android devices or iOS,

9 correct?

10 A That's correct.

11 Q Now, by the time that you became CEO in 2014, you

12 understood that Microsoft had missed the mobile revolution,

13 right?

14 A In 2014?

15 Q Correct.

16 A Yeah.

17 I mean, we were the third ecosystem, and it was

18 just hard slugging when you're No. 3 because in some sense,

19 again, it goes back to trying to convince anyone to build

20 apps for the third ecosystem was just a hard slot.

21 Q I mean, you -- you publicly stated that, "As a

22 company, we'd been very publicly missing the mobile

23 revolution," correct?

24 A Uh-huh, that's correct.

25 Q So when you became CEO in 2014, you announced that

 3589

1 you wanted Microsoft, as a company, to be mobile first and

2 cloud first; is that right?

3 A That's correct.

4 Q You made that point in the very first email when

5 you were CEO, right?

6 A That's right.

7 Q And you announced that shift because Microsoft's

8 obsession with the PC had led the company to miss mobile?

9 A Yeah, I mean, there was -- I mean, there was a

10 variety of ways to sort of characterize why we missed and

11 how we missed it.

12 But, yeah, we needed to, in 2014, recognize where

13 we were and then to then make sure that we build software

14 for even other people's platform. So that was as much about

15 a signal to say -- that's why I sort of said mobile first.

16 Still, that doesn't mean Windows first.

17 Q By the time you became CEO in 2014, Microsoft had

18 not invested nearly enough to compete in mobile search,

19 correct?

20 A Yeah, I mean, in fact, I started by saying when it

21 comes to search, to prioritize, if anything, good quality

22 desktop search, and then get back -- having learned all the

23 hard lessons that you pointed to on all of our distribution

24 deals on mobile, get back to a place where we can start

25 competing begin with the product and -- and an approach that

 3590

1 will allow us to gain mobile share.

2 Q If you'll turn to your -- turn in your binder

3 to DX451.

4 A 450, sorry?

5 Q Sorry, 451.

6 Now, this is a November 14, 2012, email exchange

7 involving a number of Microsoft employees. It's from

8 Li Jiang to Harry Shum and Derrick Connell, amongst others.

9 Do you see that?

10 A I see that.

11 Q And can you tell everyone who those individuals

12 were at the time?

13 A Yeah, at that time, I was not in that team, but

14 these are folks ostensibly running Bing at that time.

15 Q Okay.

16 And if you'll turn to the --

17 MR. SCHMIDTLEIN: And, Your Honor, we move this

18 into admission.

19 MR. SEVERT: No objection to the email.

20 THE COURT: Okay. 451 is admitted.

21 (Defendant's Exhibit DX451

received into evidence.)
22

23

24

25
 3591

1 BY MR. SCHMIDTLEIN:

2 Q So if you look at the second page, there is a

3 mobile relevance investment, a proposal.

4 Do you see that?

5 A Yeah, I do.

6 Q And this is -- or at least it appears to be

7 written by Li Jiang and Peter Bailey; is that right?

8 A Yes.

9 Q And, again, this was being circulated to the

10 highest level executives within the group that ran Bing,

11 right?

12 A It looks like it, yeah.

13 Q And if you look at the second paragraph there,

14 "Bing has invested in mobile search, although in the past,

15 the focus has been on clients in U.S. Bing client is

16 available for iPhone, Windows Phone and Android. Users can

17 also access Bing on browser at M.Bing.com. Due to

18 historical reasons, the investment on mobile relevance has

19 been minimal."

20 Do you see that?

21 A Yes.

22 Q And you have no reason to disagree with that

23 statement?

24 A Yeah, I don't.

25 Q It goes on to say, "The mobile search stack has

 3592

1 been using the same as desktop stack, but some components

2 have been lagging behind (for example, the A plus rankers

3 for mobile search were designed for desktop and they are

4 2 years old). And some mobile unique aspects haven't been

5 taken into consideration as it should be."

6 You don't have any reason to disagree with that

7 assessment, do you?

8 A Yeah, I don't.

9 Q If you jump down a bit, because there's a lot of

10 technical stuff that I don't understand about ranking, it

11 says, "Consequently, relevance on mobile search has some

12 obvious problems."

13 You don't have any reason to disagree with that,

14 do you?

15 A I don't.

16 Q And if you jump down to the next paragraph, the

17 third sentence, "We have to act fast to get Bing's mobile

18 search quality up to be competitive."

19 A Yep.

20 Q And that's consistent with all the documents we've

21 been looking at, right --

22 A That's correct.

23 Q -- that in this time period, Bing was not

24 competitive -- certainly wasn't competitive in mobile

25 search?
 3593

1 A That's correct.

2 Q Now, if you'll turn next to a document that's

3 DX456 in your binder.

4 A 456?

5 Q Yes, sir.

6 Now, again, Mr. Nadella, this is not an email

7 chain that you're on, but I want to ask you a few questions

8 about it.

9 This is an email exchange involving, again,

10 Mr. Ribas, correct?

11 A Yep.

12 Q And at the time, he -- was he running Bing at this

13 point in time?

14 A Yeah, he -- yeah, one of the components of Bing,

15 yeah.

16 Q Okay.

17 And this is -- you know, they're talking about

18 notes that were transcribed from a strategy discussion

19 involving people who are pretty high up in search, right?

20 A Yeah.

21 Q And it talks about "notes that I collected from an

22 MS poll discussion" as well, correct?

23 A That's right.

24 Q Now, if you'll turn to --

25 MR. SCHMIDTLEIN: And, Your Honor, we -- we'll

 3594

1 move into evidence DX456.

2 MR. SEVERT: I have a hearsay objection to the --

3 to the poll, whatever this was, the poll. I'm not sure what

4 that is.

5 MR. SCHMIDTLEIN: I think it's Microsoft.

6 THE WITNESS: It's an employee poll.

7 MR. SCHMIDTLEIN: It's a Microsoft employee poll.

8 I'm not going to ask him about the poll, but --

9 THE COURT: Okay, that's fine.

10 MR. SCHMIDTLEIN: -- I think it is all Microsoft

11 employees.

12 THE COURT: 456 is admitted.

13 (Defendant's Exhibit DX456

received into evidence.)
14

15 BY MR. SCHMIDTLEIN:

16 Q If you look at the -- there's a slide deck that's

17 attached to this, and if you'll go to the slide, it's -- the

18 page number is 1, the DX number is 456.007. And, again,

19 this is from July 2013 or June 2013 time period.

20 So the slide here that was presented and

21 circulated amongst the senior executives involved in Bing in

22 July '13 reads, "Our mobile story sucks. We need to beat

23 Google at mobile relevance. The gap to achieve this is

24 absurdly large, and nobody is talking about solving it."

25 Do you see that?

 3595

1 A I see that.

2 Q You don't have any reason to disagree with

3 statements that were sent to Mr. Ribas in July 2013, do you?

4 A I don't.

5 Q This is five years after -- five, six years after

6 the iPhone launched, right?

7 A That's correct.

8 Q This is the reason why Microsoft, all of those

9 mobile deals we looked at, were losers for Microsoft, right?

10 A Yeah.

11 I mean, none of them helped us gain enough ground

12 and prioritize economically these.

13 And that's why I think in 2014, which comes the

14 year after these type of emails, where I doubled-down on

15 just desktop search.

16 Q You doubled-down on desktop search, not mobile

17 search, right?

18 A That's right.

19 Q I want to circle back just for a moment on

20 Windows' phone, because having your own mobile operating

21 system was another potential dimension of competition

22 involving search at the time, right?

23 A That's correct.

24 Q And Microsoft was one of the first companies to

25 develop a mobile operating system, right?

 3596

1 A That's correct.

2 Q In other words, Microsoft had a mobile operating

3 system that was the bases for, I don't know, I've heard some

4 people refer to it as the dumb phone era, but the phones

5 that preceded smartphones, right?

6 A Uh-huh.

7 Q And Microsoft continued to develop that operating

8 system even after the iPhone and the Android phones came

9 out, right?

10 A Yes.

11 Q Microsoft released a mobile operating system

12 called Windows Mobile in 2003; is that right?

13 A Probably, yeah.

14 Q Okay.

15 And it had some success in the market, right?

16 A Yeah.

17 Q That success preceded iPhone and Android phones

18 that came out in '07 and '08, correct?

19 A That's right.

20 Q Now, when the iPhone came out in 2007, you were

21 working in search, right?

22 A That's correct.

23 Q Steve Ballmer was the CEO of Microsoft, right?

24 A That's correct.

25 Q And do you recall Mr. Ballmer making the following

 3597

1 statement in a public interview: "There's no chance that

2 the iPhone is going to get any significant market share. No

3 chance. It's a $500 subsidized item. They may make a lot

4 of money, but if you actually take a look at the 1.3 billion

5 phones that get sold, I prefer to have our software in 60

6 percent or 70 percent or 80 percent of them than I would to

7 have 2 or 3 percent which is what Apple might get."

8 Do you recall he made that statement in an

9 interview with USA Today in 2007?

10 A Yeah, I'll take your word for it.

11 Q Did you agree with that sentiment at the time?

12 A I mean, look, I think it's now conventional wisdom

13 that the iPhone changed the entire computing industry. And

14 so any statement that ridiculed it would be ridiculous

15 today. And so I think let's let that stand.

16 Q The iPhone changed the game for everybody, didn't

17 it?

18 A That's right.

19 Q And Android changed the game, too, didn't it?

20 A Absolutely.

21 Q As a result of the competition from Apple and

22 Google involving iPhone and Android, Windows Phone lost lots

23 of customers, didn't it?

24 A That's correct.

25 Q And Mr. Ballmer, having gotten it wrong about the

 3598

1 iPhone, years later decided the solution was for Microsoft

2 to get into the hardware game, right?

3 A That's right.

4 Q Six years after the launch of the iPhone,

5 Mr. Ballmer decided to spend $7 billion to acquire Nokia,

6 right?

7 A That's right.

8 Q This was how Microsoft was going to solve its

9 mobile problem, right? It was going to build Microsoft

10 search and other Microsoft software applications onto

11 Nokia's smartphone hardware, right?

12 A Yeah, at that point, we sort of said, we knew we

13 were grounded in the fact that we were number three, but we

14 wanted to compete and that was about Mr. Ballmer's attempt

15 to get into the hardware business as a way to sort of

16 solidify our number three position.

17 Q And that acquisition, I believe it was announced

18 in 2013 but closed in early 2014. Does that sound right?

19 A That's correct.

20 That's right.

21 Q Now, when did you become CEO again?

22 A February of 2014.

23 Q One of the first things you did upon becoming CEO

24 was to scrap the entire Nokia investment, right?

25 A I mean, in the first year, yes.

 3599

1 Q In 2014, you wrote off the entire 7-billion-dollar

2 acquisition of Nokia, right?

3 A That's correct.

4 Q Microsoft stopped manufacturing Windows Phones,

5 correct?

6 A Correct.

7 Q And the Nokia acquisition proved to be a failure,

8 correct?

9 A That's correct.

10 Q Now, Microsoft and OpenAI entered into a

11 partnership in the summer of 2019; is that right?

12 A Sounds accurate.

13 Q Microsoft made a significant investment in 2019 in

14 OpenAI, right?

15 A That's correct.

16 Q Was that roughly a billion dollars?

17 A Yeah, that's correct.

18 Q And today, Microsoft owns a significant share of

19 OpenAI, is that right?

20 A Yeah, I mean OpenAI is -- their corporate

21 structure is very different. It's a nonprofit with a

22 for-profit entity so you don't own anything, you own some

23 economic interest so it's a topic on its own.

24 Q We will not get into that. I will not quiz you on

25 that, I promise.
 3600

1 And Microsoft has invested over $13 billion in

2 OpenAI over the years now?

3 A Sounds correct.

4 Q OpenAI is the company behind ChatGPT, right?

5 A That's correct.

6 Q And the GPT in ChatGPT is reference to the large

7 language model that provides the generative chat results; is

8 that right?

9 A That's correct.

10 Q And in the summer of 2022, OpenAI shared its most

11 up to date GPT model with Microsoft, right?

12 A That's correct.

13 Q And Microsoft applied the AI model to Microsoft's

14 core searching ranking engine for Bing, right?

15 A That's correct.

16 Q And as a result of that, Microsoft saw the largest

17 jump in relevance in two decades for Bing?

18 A That's correct.

19 Q OpenAI's GP technology has led Bing's search

20 quality to reach an all time high, correct?

21 A That's correct.

22 Q And ChatGPT is trained on publicly available data,

23 correct?

24 A That's correct.

25 Q ChatGPT does not rely on user click-and-query

 3601

1 data, correct?

2 A That's correct.

3 Q Now, Microsoft announced the new Bing on

4 February 7th?

5 A That's correct.

6 Q This was a couple months after OpenAI's release of

7 ChatGPT, right?

8 A That's right.

9 Q And not long after that, people began to notice

10 problems with the new Bing, right?

11 A Yeah, there were issues that we fixed.

12 Q There were lots of what the industry refers to as

13 hallucinations.

14 A That's correct.

15 Q Bing was delivering strange and menacing responses

16 in some cases to user queries?

17 A The LMMs have hallucinations that do need to

18 really sort of mitigate for, by doing better alignment and

19 better safety, and those are all things that we sort of felt

20 that doing it in the open, getting it out and making sure

21 that we know how to -- it was impossible to do these things

22 in the lab.

23 Q You thought it was better to allow the

24 hallucinations to take place out in the open public rather

25 than have them --

 3602

1 A In preview.

2 And so, like all things, there's a reason why

3 people do previews, people do beta testing, people do things

4 out -- because you can't simulate the real world in a lab.

5 Q I mean, one of the responses ChatGPT gave to a

6 journalist, that it said it was spying on Microsoft

7 employees, correct?

8 A It's -- if you try to do Jungian analysis on

9 ChatGPT, I'm sure it says things.

10 Q And were you made aware of a Time magazine report

11 that Bing -- Bing's chat feature threatened to blackmail,

12 expose, and ruin a philosophy professor before --

13 A I'm not familiar with this specific case.

14 Q You're not.

15 But those concerns, these hallucinations have

16 impacted people's impression of the new Bing, correct?

17 A I would say right now it's very early days, and

18 I'm glad we are out there and using this new technology.

19 And, you know, to your point about thinking about all the

20 issues around hallucination and jail breaks and safety, and

21 that's kind of how innovation happens.

22 Q You're still very, very encouraged by the user

23 feedback and the usage patterns for the new Bing?

24 A I think of it's a new day. But at the same time,

25 I'm grounded in the fact that we desperately need many new

 3603

1 days in order to improve our position.

2 MR. SCHMIDTLEIN: Your Honor, this is probably as

3 good as place to -- I still have a little bit more to do,

4 but this is as good a time as any to break, from my

5 perspective, if that's convenient to you.

6 THE COURT: Okay.

7 Do you have a sense of what you're talking about,

8 how much longer?

9 MR. SCHMIDTLEIN: Certainly a half hour maybe.

10 THE COURT: Okay.

11 All right. So let's go ahead and break for lunch.

12 It's a little after 12:30 now. We'll resume at 1:35.

13 Mr. Nadella, just the same instruction as before:

14 If you please not discuss your testimony during the break.

15 We'll see you around 1:30 and we'll get started at 1:35.

16 COURTROOM DEPUTY: All rise.

17 This Court stands in recess.

18 (Recess from 12:32 p.m. to 1:35 p.m.)

19

20

21

22

23

24

25
 C E R T I F I C A T E

I, William P. Zaremba, RMR, CRR, certify that

the foregoing is a correct transcript from the record of

proceedings in the above-titled matter.

Date:__October 2, 2023______ ____________________________

William P. Zaremba, RMR, CRR