Generative AI - The New Reality: How Key Players Are Progressing

1
GENERATIVE AI – CHANGING THE WORLD
KEY PLAYERS AND THEIR PROGRESS
Oct 2023

2
Agenda
Disclaimer: This presentation acknowledges and gives credit to the work of others. Necessary validation has been taken to
avoid copyright infringement. Any instances that violate terms can be removed when notified. All discussed thoughts &
opinions are my own & not those of my employer or other parties.
§ Generative AI
§ Evolution Since 1950, Hype or Potential, Where is the Spend, DCN Spend, Value Chain – E2E View, How it Works,
Deployment Models, Future of LLMs, Key Players
§ TSMC – Chipmaker That Runs The World, Leading the most Advanced Chip, Risks
§ Nvidia – Godfather of AI
§ Why the Market is Bullish, Where is the Growth, Is Dot Com Repeating? Under the Hood – TSMC
§ CEO’s Relation and Taiwan Link, Risk to its Growth
§ Microsoft – The King of AI in Cloud and Software
§ Where is the Growth, Partnership with Open AI, Cost of Training an AI System, Risks
§ Google – The King of Search, Where is the Growth, Economies with Gen AI, Cloud Play, Diversify, $100Bn Blunder, Risks
§ Amazon – The King of Retail
§ Where is the Growth, Flywheel Phenomenon, Multiple Flywheels, AWS The Crown Jewel, Chasing new Flywheel, Risks
§ Tesla – Vehicles That Reinvent Themselves, It’s Not a Car, Secret of Success, Growth? FSD Ambition, AI-enabled Dojo, Risks
§ Oracle - Database Leader to AI Innovator, Facing Headwinds, Growth Focus, Gen AI Play, Risks
§ Salesforce - Cloud and AI-Powered CRM, Where is the Growth, Competition to Cooperation, Risks
§ SAP - ERP Market Leader, One Stop Shop, Scrambling To Catch Up, Risks
§ Palo Alto Networks - Leader in Cyber Security, Dominant Growth Strategy, Risks
§ IBM – A Tech Giant, Growth, Case of Misfires, AI Play is Comprehensive but Sluggish, How it Lost its Way, Risks
§ Board Games - Competition to Cooperation
§ AMD - Pending

3
Attribution
Licensed Under
This work is Licensed Under
No Derivative 4.0 International
Disclaimer: This presentation acknowledges and gives credit to the work of others. Necessary validation has been taken to
avoid copyright infringement. Any instances that violate terms can be removed when notified. All discussed thoughts &
opinions are my own & not those of my employer or other parties.

4 Source: Gartner
Gartner
Emerging Technologies -2023
Key Message
Today, Gen AI stands at the top of extravagant expectations, as per the
Gartner Emerging technologies Hype urve. The question that arises is
whether it will be another overhyped concept that will eventually fade
away, or whether it has significant support behind it to sustain and
bring about a considerable change.

5
Evolution - Since 1950
AI to Generative AI
1950 1960
Birth of AI
Alan
Turing
Markov Chain,
statistical model that
could be used to
generate new sequence
of data based on input
Rule Based
System
2020
Big Data
and
Deep
Learning
AI Boom
Conversational
AI
&
GPT-3
Source: Wikipedia,Seekingalpha, Bloomberg, Forbes, Gartner
Machine
Learning
1970
Knowledge
Rep
1980
Expert
Systems
1990 2000 2010
2014
DeepMind, VAE,
RNN
2017
Nvidia Progressive
GANs, Generates
Images,
2018
LLM Foundation
via
BERT, “Pre-
training of Deep
Bidirectional
Transformers for
Language
Understanding”
2022
Dall-E,
Generative Pretraining
Transfer-3.5
(OpenAI)
2023
Google Bard
GPT-4 (OpenAI)
This Photo by Unknown Author is licensed under CC BY-SA-NC

6
Why Hype Around Generative AI
• In 2022, the global AI market was worth $0.45Tn. It’s expected to significantly expand and reach $2.6Tn by the year 2032, at a CAGR of 19%.
• It's worth noting that by 2030, an estimated 70% of companies will incorporate AI into their operations, marking a significant 35% increase
from the 35% observed in 2023.
• This underscores the current opportune moment to capitalise on AI investments. Today AI stands as the forefront of computing innovation
and is emerging to become the mainstream technology for 90% of companies.
• The increasing need for AI technology in different industries like automotive, healthcare, banking and finance, manufacturing, food and
beverages, logistics, and retail is predicted to strongly contribute to the expansion of the global AI market in the coming years.
• A significant milestone was the introduction of ChatGPT 3.0 in 2022, which brought attention to the potential of generative AI, which has led
to a noticeable rise in interest in this domain.
Generative AI
Hype or Potential
Source: Precedentresearch, AFR, RBA, Seekingalpha, Mosrgan Stanlley, Gartner
0.00T
0.50T
1.00T
1.50T
2.00T
2.50T
3.00T
2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032
Market Size in US $Tn
CAGR 19%
• Key Indicator
• The current valuation of the cloud market stands at $500Bn, with companies
relying heavily on hyper scalers such as AWS, AZURE & GCloud for
effectively outsourcing data management and storage.
• The integration of Generative AI into Cloud Computing will significantly
enhance efficiency, resulting in a further interconnection of the two domains.
• This integration will culminate in a market worth over $1Tn within the next 3
to 5 years.
High
Within
Human
Capability
Beyond
Celestial
Generative AI
New Discoveries
Robots
Human
Low Task Automation

7
Generative AI
Where is the Spend
Source: Precedent research, AFR, RBA, Seeking alpha
Key Message
• Majority of the AI Spend in future is going to be around building ML
capability.
• With AI’s market growth it requires substantial capital, and it's
forecasted that spending on AI-centric IT systems will surge at an
impressive rate of 27% from 2023 to 2026 (IDC). This increase will
propel the investment from $150Bn to $300Bn, nearly doubling the
expenditure on this transformative technology.
• Key sectors to embrace this technology are banking and retail sectors.
This graph excludes spending on IT Application and Services.

8
Key Message
• The Data Centre Market is poised for a steady revenue growth,
projecting a CAGR of 4.66% from 2023 to 2027.
Generative AI
Data Centre Spend
Source: Precedentresearch, AFR, RBA, Seeking salpha

9
Generative AI
Value Chain – E2E View
Source: Precedentresearch, Linkedin, Google, MSFT, Nvidia, iStock, Seekingalpha,
Low
Margin
High
Margin
US
$450
Bn
Annual
Spend
by
2026
US
$500
Bn
Annual
Spend
by
2026
Generative AI: Reshaping the world, one line of code at a time.

10
Generative AI
Value Chain – How It Works
Source: Martinflowler, Databricks.com, Seekingalpha, Google
Process Flow of Data
• Data Loading:
• The process starts with the data loading onto the AI
accelerator. The data can be in a variety of formats, such as
text, images, or audio.
• Create the Model:
• The AI library is then used to create a model. The model is a
mathematical representation of the data. The AI library
provides tools and algorithms for creating different types of
models, such as deep learning models and decision trees.
• Pre-Train the Model:
• The Large Language Model (LLM) is then used to pre-train
the model. Pre-training is a process of training a model on a
large dataset of text data. This helps the model to learn the
statistical relationships between words and improve its
performance.
• Fine Tune the Model:
• The Foundation models are pre-trained AI models that can be
used as a starting point to fine-tune the model. Fine-tuning is
a process of adjusting the parameters of the model to
improve its performance (save time and resources) on a
specific task.
• Manage the Model Lifecycle and Deployment
• The MLOps (similar to DevOps) tools are then used to
deploy the model to production. MLOps tools automate the
process of deploying and managing AI models. This helps to
ensure that the models are deployed correctly and that they
are monitored for performance and accuracy.
• Run the Application
• The model is then used by the applications to make
predictions, solve problems, or generate creative content.
Services
Applications
Model Hub
MLOps
Foundation
Models
(SubsetLLM)
Cloud
Platforms
Hardware and
Accelerators

11
Build Buy Hybrid
Advantages:
• Complete control over the AI model,
including its architecture, training data,
and outputs.
• Ability to customise the AI model to
meet specific needs.
• Potential to build a proprietary AI
model that can be used to differentiate
your business from competitors.
• Faster and easier to deploy an AI model.
• No need to invest in AI and machine learning
expertise.
• Access to a wide range of pre-built AI models
that can be used for a variety of tasks.
• Combines the advantages of both
build and buy approaches.
• Allows you to customise a pre-built AI
model to meet your specific needs.
• Reduces the investment of time and
resources required to build an AI
model from scratch.
Disadvantages
• Requires a significant investment of
time and resources, including expertise
in AI and machine learning.
• Can be challenging to build and deploy
a high-quality AI model, especially for
complex tasks.
• Requires ongoing maintenance and
updates to keep the AI model up-to-
date and performing at its best.
• Less control over the AI model, including its
architecture, training data, and outputs.
• May not be able to find a pre-built AI model
that meets all of your specific needs.
• May be more expensive to buy and maintain a
pre-built AI model than to build your own.
• Can be more complex to implement
than either build or buy approaches.
• May require some expertise in AI and
machine learning.
Use Cases
• Developing new products and services
that are powered by AI.
• Improving existing products and
services with AI capabilities.
• Automating tasks that are currently
performed by humans.
• Solving complex problems that cannot
be solved with traditional methods.
• Experimenting with AI to see if it can be used
to solve a particular problem.
• Deploying AI capabilities quickly and easily,
without having to invest in AI and machine
learning expertise.
• Accessing AI models for specific tasks, such
as image recognition, natural language
processing, or machine translation.
• Deploying AI capabilities quickly and
easily, while still having some control
over the AI model.
• Customising a pre-built AI model to
meet specific needs.
• Reducing the investment of time and
resources required to build an AI
model from scratch.
Magnitude of Cost • High to very high • Low to medium • Medium to high
Example • Tesla, Google, Microsoft • SAP, Salesforce • Oracle
Source: Next Platform, Gartner, Seeking alpha, Google
Gen AI
Deployment Models

12
Indexed
Challenging
Scalable
Easily
Smaller LLM
Retrieval Augmented
Safe, Factual, Adaptable
Llama2 13B (Meta)
Commodity H/W
Big LLM
Contextual
Fluency
Chat GPT 4 > 1Tn
Specialised H/W
Unstructured Data
Future
Today
• The future of Large Language Models (LLMs) will
be Smaller, Smarter, and Accessible, running on
commodity-like hardware instead of specialised
hardware only.
• This will lead to a reduction in both Fixed and
Marginal costs associated with building the
models, while also enabling more targeted and
effective outcomes, such as improved efficiency,
safety, security, and factual accuracy.
Foundation Model
Data Training
Architecture
Quantity
Quality
Specialisation Degree
Pre-Training
Alignment (Behavior
like Safety, Feedback)
Style
(Transformer, Diffusion)
Dense
Size (parameters)
Future of Lang. Models
Maserati to Mazda
Key Message
• Large Language Models (LLMs) are incredibly resource-hungry (compute &
finance), making them only accessible to a select few organisations. Besides their
time to market duration is not desirable.
• Google, Meta, and other major players in the AI industry are actively
working to make LLMs more efficient and affordable for wider use.
• Foundation Model vs Large Language Model
• Large Language models (LLMs) are a type of machine learning model that can
process and generate human language. They are trained on massive datasets of
text and code and can be used for a variety of tasks, such as translation,
summarisation, question answering, and creative writing.
• Foundation Models are a newer type of AI model that is still under
development. They are trained on even larger datasets of text, code, and other
types of data, such as images and videos. Foundation models are designed to be
more general-purpose and adaptable than LLMs, meaning that they can be used
for a wider range of tasks.
• Key Differences
• Foundation models are multimodal, meaning that they can work with multiple
types of data. This enables them to perform tasks that would not be possible for
an LLM alone, such as generating images from text or translating videos
between languages.
• Besides, Foundation models are designed to be more transferable. This means
that they can be easily adapted to new tasks without having to be retrained from
scratch. This makes them more practical for use in real-world applications.
• Building Blocks of a Foundation Model

15
TSMC - The Chipmaker
That Runs The World
Source: Google Finance, Yahoo Charts, TSMC, Seekingalpha
Market Cap. Progression from 1994
Key Indicators
• Market Cap – $423.63Bn
• EV – $406.01Bn
• Debt – $29Bn
• Cash - $48Bn
• P/B – 4.58
• P/E (Trailing) – 15.50 (Low)
• P/E (Forward) – 15.24 (Low)
• Economic Moat – Wide
• Major Shareholder: Berkshire Hathaway
1987- 2018
• Founded TSMC in 1987 and led the company to become
the world's largest semiconductor foundry.
• Pioneered the concept of the pure-play semiconductor
foundry, which has become the dominant model in the
semiconductor industry.
• Oversaw TSMC's technology leadership, with the
company being the first to produce chips using a
number of advanced manufacturing processes.
• Expanded TSMC's global reach, with the company now
having fabs in a number of countries around the world.
• Built TSMC into a highly profitable company, with the
company generating billions of dollars in profits each
year.
$424Bn
2018 – To date
• Continued TSMC's growth and expansion.
• Oversaw TSMC's continued investment in
new manufacturing technologies, with the
company being at the forefront of the
development of new manufacturing processes.
• Expanded TSMC's customer base to 530+
• Maintained TSMC's technology leadership
• Led TSMC through the global chip shortage,
with the company emerging stronger from the
crisis.

16
• Q2 FY23
• $15.7Bn in revenue, down 13.7% YoY.
• $5.93Bn in net income, down 23% YoY.
• 54.1% gross margin.
• 42% operating margin.
• Segment
• Growing
• HPC(AI) 44% & Automotive (TSLA) 8% of total Rev
• Declining
• Smartphone 33% and IoT 8% of total Rev.
• Technology
• 50% of revenue comes from 5nm and 7nm technology.
• Other
• Debt – $29Bn, Cash - $48Bn.
• FCF $5.2Bn, FCF Margin 7.4%
• Ahead
• Guidance for Q3FY23 revenue is $16.5Bn to $17.5Bn.
• FY23 guidance of 10% decline YoY.
2nm technology is on track to begin production in 2025.
• Key Customers (Total 530+)
• TSMC's revenue is made up of 26% from Apple and 7% from Nvidia.
Apple has a 10-year partnership with the chip maker.
• Apple designs chips for iPhones and Mac computers, while Google
designs Tensor Chips for Pixel smartphones. Qualcomm and
MediaTek design processors for Android phones. Nvidia designs
Gaming and Artificial Intelligence (AI) processors and AMD and
Nvidia design advanced processors for Tesla.
• TSMC chips are also used by major cloud providers like AWS, MSFT,
Google, Oracle, and IBM for data centres, networking, and software.
Broadcom designs chips for broadband and wireless markets.
TSMC Leading
The Most Advanced Chips
Key Indicators
• Domain: Semiconductors
• Comp. (Chip Manf) – Samsung, SMIC, GFS, UMC
• Growth Segment – HPC AI Chips (Up)
• Economic Moat – Wide
• Cyclical - Yes
TSMC, 59%
Samsung, 13%
GlobalFoundries, 7%
UMC, 6%
SMIC , 5% Semiconductor Market Share
Seg Revenue Split
HPC (C/GPU), 44%
Smartphone, 33%
IoT, 8%
Automotive, 8%
Digital Consumer, 3%
Others, 4%
TSMC Revenue by Platform
Source: NextPlatform, TSMC, WSJ, Economist, Seekingalpha, CNBC, IDC, Yahoo Charts, Google Fin, Nvidia

17
• Traditional AI
• AI servers are specialised computers designed for AI Training and Inference.
Training involves adjusting the layers of the neural network based on results and
can require a month of computational power. Inference uses trained neural
network models to infer results. AI chips are used for applying trained AI
algorithms to real-world data inputs, which is often referred to as "inference".
• Specialised chips called Accelerators play a crucial role in the field of deep
learning. There are two types of accelerators, Training Accelerators and Inference
Accelerators. Training accelerators are optimised to facilitate the training of deep
learning models by performing intricate calculations and processing extensive
datasets. Inference Accelerators, on the other hand, execute trained models on
fresh data with great speed, making them perfect for real-time applications such as
image recognition in cameras or voice assistants in smartphones.
• Gen AI
• Traditional AI relies on structured, labelled data for training and is confined to
specific tasks such as image recognition, sentiment analysis, and
recommendation systems. Generative AI, on the other hand, aims to simulate
human-like creativity and generate content autonomously. It is versatile and
capable of producing diverse outputs across various domains, including text,
images, music, and even entire applications. The key aspect of Gen AI models is
their ability to generate content that goes beyond the scope of their training
data.
Various types of Gen AI chips are
• GPU (Graphics Processing Unit)
• TPU (Tensor Processing Unit)
• FPGA (Field-Programmable Gate Array)
• ASIC (Application-Specific Integrated Circuit)
• Neuromorphic Chips
Key Message
• TSMC is able to offer its customers its manufacturing capabilities in the areas
of Smartphones, High Performance Computing (HPC), Internet of Things (IoT),
Automotive and Digital Consumer Electronics. TSMC calls its Technology
Leadership, Manufacturing Excellence and Customer Trust as TSMC Trinity of
Strengths.
• TSMC is a major player in three of the top four semiconductor growth sectors
which include Silicon Carbide (SiC), Gallium Nitride (GaN), AI Compute
Processors, and Generative AI.
TSMC - The Silicon Maker
Beholds the Future of Technology
Source: Next Platform, TSMC, WSJ, Economist, Seeking alpha, CNBC, IDC, Yahoo Charts, Google Fin, Nvidia
TSMC 2023

18
• Geopolitical Risk
• TSMC is headquartered in Taiwan, which is a politically sensitive region. If there were to be a conflict between China and Taiwan, it
could have a significant impact on TSMC's business.
• Economic Risk
• TSMC is a cyclical business, which means that its revenue and profits can fluctuate significantly depending on the overall state of the
economy. A recession could lead to a decline in demand for semiconductors, which would hurt TSMC's business.
• In 2022, after pandemic complications led to a global semiconductor chip shortage, the overall industry took a tumble for the worst.
Companies like TSMC were detriment due to supply shortages, creating downturns in business. Some repercussions were still
impacting recent financial returns, viewed by the 13.7% decline in revenue YoY in Q2FY23.
• Technological Risk
• TSMC is constantly investing in new technologies, like 2nm chip manufacturing, to stay ahead of the competition. However, there is
always the risk that a competitor could develop a new technology that makes TSMC's technology obsolete.
• Competitive Risk
• TSMC faces competition from other semiconductor foundries, such as Samsung and GlobalFoundries. These companies are also
investing heavily in new technologies, and they could pose a threat to TSMC’s dominating 59% market share.
• Customer Concentration
• SMC has a few major customers like Apple, Nvidia and Tesla that account for a significant (~40%) portion of its revenue. A loss of a
key customer or a reduction in orders from these customers could have a significant negative impact on the company's financial
performance.
TSMC’s Chip Dominance
Risks to its Growth
Source: NextPlatform, TSMC, WSJ, Economist, Seekingalpha, CNBC, IDC, Yahoo Charts, Google Fin, Nvidia

19
Nvidia - Godfather of AI
Why the Market is Bullish
Source: Precedentresearch, Google Finance, Nvidia, Seekingalpha
Key Indicators
• Market Cap – 1.12 Tn
• EV – 1.1Tn
• P/B – 40.58
• P/E (Trailing) – 108 (Growth)
• P/E (Forward) – 47.6 (Growth)
• Economic Moat: Wide (product sales)
$
8
4
0
B
n
Market Cap. has increased by $840Bn in the last 12 months
Today Nvidia commands more than 70% of the AI Chip market
• Invented the graphics processing unit (GPU), which has revolutionised the computer graphics and gaming
industries.
• Led NVIDIA to become the world's leading supplier of GPUs.
• Pioneered the development of GPUs for a wide range of applications, including Gaming, AI, DCN Computing,
and Auto Industry (Tesla).
• Helped to shape the future of the computing industry by predicting the rise of artificial intelligence and
developing new GPU-based computing platforms (CUDA) to support it.
1993 - Todate

20 Source: NextPlatform, Seekingalpha, Yahoo Charts, Google Fin, Nvidia
• 1st
mover advantage in GPU (parallel) driven chips interacted by the
software interaction layer of CUDA.
• Nvidia in 2007 cleverly provided CUDA free of charge as
long as clients used its GPUs. The move locked in customers
from the get-go. Those who wanted to switch would have to
rewrite existing code, potentially disrupting the application at
a user level. Even more importantly, developers would need
to be retrained.
• Today CUDA is the default industry standard. Nvidia has
erected a fortress which is now generating a helluva of cash.
• As a result, Nvidia’s data centre revenues have grown from
US $317Mn in 2015 to US$15Bn last year. Gross Margins have
increased from 38% in 2006 to 67% in the previous quarter.
• Today Nvidia commands more than 70% of the AI Chip
market.
• Products like the H100 Tensor core GPU, DGX
supercomputers, inference platforms and AI IaaS in the
Cloud, are poised to transform AI delivery, partnered with
major cloud providers.
• Other Key players are AMD, ARM (Softbank), Apple and
Intel.
• CUDA - Compute Unified Device Architecture, a programming
software layer that allows applications to maximise the advantage
of parallel processing and eliminate cumbersome coding.
Key Indicators
• Market Cap – 1.12Tn, EV – 1.1Tn
• P/B – 40.58
• P/E (Trailing) – 108 (Growth)
• Economic Moat: Wide (strong driven by S/W)
•
Why GPU’s
• With lower latency, GPUs are widely recognised as
the most efficient and fastest way to construct, train
and advance machine learning applications.
Parallel
Simple
Applications
Complex
Nvidia
GPU
Intel
CPU
Serial Processing
AMD, Apple
ARM, AMD, Apple,
Asus
Nvidia
What is their Moat

21
• CPU (Central Processing Unit) is a general-purpose processor that can be used to perform a wide range of tasks, including
running applications, processing data, and performing calculations. CPUs are designed to be flexible and versatile, but this comes
at the cost of performance.
• GPU (Graphics Processing Unit) is a specialised hardware device that is designed to perform graphics processing, which
involves manipulating large amounts of data in parallel. This makes GPUs well-suited for AI tasks, which also involve
manipulating large amounts of data. However, GPUs are not specifically designed for AI tasks, and they may not be as efficient as
AI accelerators for certain types of AI workloads.
• AI Accelerator is a specialised hardware device that is designed specifically for AI tasks. They are typically optimized for specific
types of AI tasks, such as machine learning and deep learning. As a result, AI accelerators can be significantly faster and more
efficient than GPUs for certain types of AI workloads.
Source: NextPlatform, Linkedin, Seekingalpha, Google, Nvidia
Nvidia
CPU, GPU, TPU and Accelerators
Feature CPU GPU AI Accelerator
Purpose
General-purpose processing
(Serial)
Graphics processing
(Parallel)
AI tasks
(Parallel)
Flexibility High Medium Low
Performance Good for general workloads Good for general AI workloads Best for specific AI workloads
Cost Less expensive More expensive Most expensive
Example Intel Core, AMD, Apple M1 Max Nvidia GeForce, AMD Radeon, Intel Arc GPUs, TPUs, ASICs, FPGA’s
Controlled By
Software
OS like Windows, Ubuntu, MacOS
(not a library)
Nvidia’s CUDA , AMD’s OpenC, Intel’s
OneAPI
Nvidia’s CUDA
Googles Tensor Flow for TPUs
Key Message
• The best choice for you will depend on your specific needs and budget.
• For a wide range of tasks with flexible & versatile device, CPU is a
good option. To perform AI tasks efficiently, then an AI accelerator is a
better choice. GPUs fall somewhere in between CPUs and are a good
option for general AI workloads but may not be as efficient as AI
accelerators.

22
Nvidia – FY24 H1
Where is the Growth
Key Indicators
• Domain: Microchip (GPU, DPU)
• Comp. – AMD, Intel, Apple, ARM
• Rev Growth – Up
• Growth Segment – DCN (on steroids)
Source: Wikipedia, MIT, NextPlatform, Linkedin, Seekingalpha, Company Data, Oracle, Markettech, TechCrunch, ServiceNow, Khaveen Investments
6.9
9.7
11.7
10.9
16.7
26.9 27.0
7.2
13.5
16.0
28%
33%
32% 26%
27%
37%
16%
42%
58%
53%
2017 2018 2019 2020 2021 2022 2023 Q1 2024 Q2 2024 Q3
2024F
Revenue in US $Bn Op Margin
23%
-12%
41%
61%
-27%
22%
11%
53%
2%
124%
58%
41%
18%
141%
21%
7%
-13%
100%
-29%
31% 28%
2019 2020 2021 2022 2023 Q1 2024 Q2 2024
Gaming Revenue Data Center Revenue
Professional Visualization Revenue Automotive Revenue
OEM and Other Revenue
• Q2 FY24 Results
• FY24 Q2 proved to be highly successful, with a record-breaking reported
revenue of $13.5Bn and an operating margin of 58% - an all-time high for
the company.
• It is on track to generate $16Bn in revenue and maintain an operating
margin of 53% during Q3.
•
• Data Centre Revenue saw a staggering 141% increase between the 1st and
2nd quarters of FY24.
• In contrast, Gaming, Professional Visualisation, Automotive, OEM and
other segments made comparatively modest contributions when
compared to the robust performance of Data Centres.
• It's worth noting that Professional Visualisation revenue also saw a notable
28% increase between quarters, while Gaming experienced a solid 11%
growth during the same period.
• Currently, its total debt amounts to a modest $9.9Bn. This becomes even
more evident when juxtaposed with its substantial cash balance of $16Bn.
• It generates noteworthy free cash flows ("FCF"), boasting a margin of 30%.
While this figure does not surpass the pinnacle reached in 2020 with a 38%
FCF margin, it remains highly commendable.
• Nvidia's financial metrics reinforce what is widely acknowledged that the
company enjoys healthy margins, strong cash flows, low debt levels, and a
substantial cash reserve. These factors collectively position Nvidia
favourably to pursue growth opportunities without exposing itself to
excessive financial leverage.
Segments YoY %Growth
Revenue in US $Bn and Operating Margin

23
Nvidia
Is Dot Com Repeating ? Cisco Sun Mic.
Key Message
• The current bull run in the stock market
has seen Nvidia's market capitalisation
increase by $840Bn over the past 12
months. This is a significant indicator of
the company's overall success and growth
potential in the current market.
• There are concerns within certain circles
that history may be repeating itself, with
the hype surrounding Gen AI and its
demand appearing to mirror the Dot Com
era. This is especially relevant given that
the Revenue Guidance for Q3 is the
highest to date, with a projected figure of
$16Bn.
• If you take the other Nvidia divisions and separate them out and add them together,
revenues were up by 9.9% to $3.18Bn. That’s another way of saying that the Nvidia
data centre business is now 3.2X bigger than the rest of Nvidia, and the data centre
business now accounts for 76.4% of the overall sales for the company.
• Dot Com - What Happened
• We are now at the point in the AI revolution that the Dot-Com Boom was in
1999 and 2000, where every Internet startup lined up money from venture
capitalists and the 1st thing, they did was buy a big, fat Solaris server and a
whole bunch of pizza box Web servers from Sun Microsystems, some storage
from EMC, and a relational database from Oracle. All three of those
companies minted coin for a number of years. And then the crunch came, and
where did we all end up? On pizza box X86 servers running Linux and using
the MySQL database.

24 Source: Foreignpolicy.com, Yolegroup.com, NextPlatform, Linkedin, Seekingalpha
Key Message
• If Taiwan gets invaded, which could consequently affect TSMC, the real
sufferers would be companies like Apple, Alphabet, Amazon, Microsoft,
Nvidia, AMD, Tesla, most other automobile companies, Broadcom, Oracle,
IBM and counting, all of their business customers and all their end
customers as well.
• TSMC’s most advanced N3 microchip is Powering the iPhone 15 Pro Max.
Nvidia
Under The Hood - TSMC
TSMC’s N3 (3nm) Semiconductor Design
Nvidia responded to the US ban on the sale of its
advanced accelerators (A100 and H100 GPUs) to the
Chinese and Russian market by exporting scaled-down
versions, resulting in even higher unit demand. At the
same time, this has increased the pressure for a home-
grown AI acceleration supplier to rise in China.
The Huawei Mate Pro 60 reportedly contains 7nm chips
manufactured in China by the Semiconductor Manufacturing
International Corp (SMIC), which is partly state-owned. This
has sparked concerns in the United States about how China
acquired such an advanced chip technology.
• This is the most advanced semiconductor on the planet, and it is
banned by the US for manufacturing and export to China and Russia.
• Despite the controls’ focus on physical exports, most Chinese technology firms
access chips virtually using services offered by cloud computing companies.
These services are not monitored to prevent usage by blacklisted foreign
entities under the United States’ current system of safeguards.
• According to reports, SenseTime, a facial recognition company that has been
blacklisted, has been using intermediaries to smuggle prohibited components
from the United States. This strategy resembles the one used by China's
leading nuclear weapons laboratory, the state-run Chinese Academy of
Engineering Physics, as its employees have stated. It seems that China has been
exploiting these vulnerabilities.
• And despite being blacklisted for human rights abuses, state-backed artificial
intelligence firm iFlytek has been renting access to controlled NVIDIA chips via
the cloud. There is little practical difference between using a physically
exported chip and using a chip “virtually” through the cloud. However, this
practice is currently completely legal, even for firms like iFlyTek. This tactic
could soon become even easier: NVIDIA has called out its intentions to expand
its cloud supercomputing offerings to China.

25
• Nvidia’s CEO Jensen Huang was born in Taiwan. He immigrated to the United States with his family in 1973, and he is now a
US citizen. Huang currently lives in Los Altos, California.
• Nvidia's headquarters is also located in Santa Clara, California.
• Nvidia has factories in a number of locations around the world, including the United States, China, Taiwan, and South
Korea. However, its largest factory is located in Tainan, Taiwan.
• TSMC’s CEO C.C. Wei was born in China in 1958. He is a Taiwanese citizen and currently lives in Hsinchu, Taiwan.
• TSMC's headquarters is also located in Hsinchu.
• TSMC has factories in a number of locations around the world, including Taiwan, China, the United States, and South
Korea. However, its largest factory is located in Tainan, Taiwan.
• Both CEOs have a close working relationship. TSMC is Nvidia's primary supplier of chips, and the two companies have
collaborated on a number of projects, including the development of new chip manufacturing technologies.
• Huang and Wei have also spoken publicly about their admiration for each other. Huang has praised TSMC for its
manufacturing expertise and its commitment to quality, while Wei has credited Nvidia with its innovative chip designs.
• The relationship between Huang and Wei is important for both companies. Nvidia relies on TSMC to manufacture its chips, and
TSMC relies on Nvidia to be a major customer. The two companies also work together to develop new chip technologies, which
benefits both companies in the long run.
• In addition to their business relationship, Huang and Wei are also friends. They have been photographed together at industry
events, and they have spoken about their shared passion for technology.
• Overall, the relationship between Nvidia CEO Jensen Huang and TSMC CEO C.C. Wei is close and mutually beneficial. The two
companies rely on each other for success, and they have a shared commitment to innovation.
Nvidia – TSMC
CEO’s Relation and Taiwan Link

26
Major Risks To Consider
Low-cost options like TPUs built using ASIC Chips
• Google developed its accelerator chip using ASIC architecture, which is far cheaper to buy and run at scale. Although these chips may be
slower compared to GPUs, it is a trade-off between lower levels of precision and decreased costs. Extensive research has shown that most
applications do not require the very high levels of precision provided by GPUs; thus, their use may be excessive and wasteful.
Overcome Nvidia’s Moat in Software
• An innovative mitigation plan has been proposed by Chris Lattner, a highly regarded software expert and founder of the start-up Modular
(with a valuation of $0.6Bn). Lattner's solution involves a modular and extensible inference engine that is capable of running models in
production. This engine can be easily integrated into any development framework, cloud or hardware without any concerns about
compatibility issues. Regardless of the method used for constructing a model, it can be easily incorporated into the Modular inference
engine, making it adaptable to any hardware and cloud environment. This development has the potential to pose a significant challenge to
Nvidia and their current offerings.
High Concentration of Customer Base
• The big cloud builders – Google, MSFT, and AWS and given the restrictions of sales of full-blown GPU accelerators to China where Alibaba,
Tencent, and Baidu, which also operate clouds – accounted for more than 50% of data centre sales in fiscal Q2.
Supply Chain Bottleneck
• According to the Financial Times, Nvidia is facing supply chain constraints to reach the maximum capacity.
• Analysts have reported that Nvidia might struggle to produce >500K H100 GPU this year, while it was looking to ship 1.5 - 2 Mn.
Geopolitics
• Nvidia outsources all of its production to TSMC in Taiwan, which is at the centre of geopolitical risk.
• The most advanced chips from Nvidia and other companies were banned from direct sale in the Chinese market by the Biden administration
in November 2022. As a result, Nvidia had to create less powerful GPUs that meet the new regulations while also keeping customers in a
growing market.
• In July 2023, the Biden administration added further restrictions on investment into the Chinese tech sector, which made the Chinese
industry executives work on a plan to become more self-reliant. This is not the last time new restrictions are being placed as the ongoing
Sino-American confrontation is unlikely to end anytime soon.
• That’s why a potential disruption of Nvidia’s operations in China, which accounted for around 21% of total revenues last fiscal year, is a
factor that can greatly undermine the company’s growth story and result in a permanent loss of opportunities for its business.
Demand Side - Economic Slowdown (High Inflation, Russian-Ukraine Conflict)
• If the growth of the global economy slows down or the gaming market doesn’t properly recover, then there will be an impact on its growth
as consumers and corporations tackle the rising cost of living, higher IR and WACC.
Nvidia
Risks to its Growth

27
Google (Alphabet)
Evolution Since 2004 - 2023
$1.35 Tn
US $177 Bn
Market Cap Progression from 2004 Onwards
Google Comprises 99% of Alphabet’s Revenue.
U$200 Bn
Key Indicators
• Market Cap – $1.75Tn
• EV – $1.66Tn
• P/B – 40.58
• P/E (Trailing) – 29.27 (Growth)
• Economic Moat: Wide
• Successfully oversaw Google IPO.
• Built a Search and Advertising
Monopoly business with a $203Bn
Market Cap and generating
revenue of $28Bn pa.
• Business expanded beyond search
by launching several
new products, like Google Maps,
and Gmail and successfully
acquired YouTube and Android.
• Built on Android's Continued
Success.
• Ensured Business continued its
Search and Advertising
Monopoly.
• Invested in AI.
• Expanded Google’s Cloud computing
business which became profitable in 2023.
• Launched new products in Mobile and Smart
Home.
• Invested in Gen AI (ML) and launched GPT
based Bard, next is Gemini.
• Under his leadership Digital Advertising
spending has reached to 70% of total
spending.
2001 - 2011 2015 - Todate
2011 - 2015
Source: Wikipedia, MIT, NextPlatform, Linkedin, Seekingalpha, Company Data, Microsoft, Markettech, TechCrunch, ServiceNow, Khaveen Investments

28
Google – King of Search
Where is the Growth
Key Indicators
• Comp. (Search) – Bing, Yahoo, Baidu
• Comp. (Cloud) – AWS, MSFT, IBM
• Comp. (Ads) – Meta, Amazon, LinkedIn
• Comp. (Video) - Netflix, Disney, Amazon, FB, Tiktok
• Rev Growth – Up,
• Growth Segment – Search, Cloud
Q2 FY23 Result
• Revenue grew to $74.6Bn from $69.8Bn in Q1
• Net & Operating income grew 22% & 25% resp.
• FCF continued to grow by 73% YoY, from $12.6 Bn to $21.8Bn. There was a
slight decline in cash flow last year but that was due to increased CAPEX,
where the company invested heavily into servers, DCN & office facilities.
• Cash on hand grew to $118Bn, up from $115Bn in Q1. This cash is enough to
buy 2 of the 5 largest banks in the U.S. by assets, (U.S. Bancorp and
Citigroup). Besides, it has a minimal amount of debt worth $11.9Bn and a
credit rating of AA+.
Google Search & Other
• Revenues grew 13.5% YoY in Q2 (compared to 68.1% last year), driven by
Travel and Retail.
• YouTube Ads (2.6 Bn Users, 1/3 of Earths pop.)
• Revenues grew just 4.8%, after an “uniquely strong” 83.7% growth last year.
Time spent on YouTube globally has continued to grow. Short-form video
consumption is increasing across multiple platforms. Positive results in
monetising its Shorts format. However, revenues did decelerate QoQ, which
primarily reflects pullbacks in spend by some advertisers.
• Google Network Member Properties
• Revenue grew 8.7% YoY, driven by the AdSense product. However, revenues
decelerated on QoQ basis because of pullbacks by advertisers.
• Google Other
• Revenues fell 1.1% YoY, due to lower revenues from the Play store after
Google reduced its commissions on subscriptions from 30% to 15% at the
beginning of 2022. This followed reduction on apps commissions at the start
of July 2021.
• Google Cloud
• Revenues grew 35.6% YoY in Q2, compared to 43.8% in Q1 and 53.9% in the
prior-year quarter, continuing its trajectory of strong revenue growth and
reported 2nd profit in FY23.
.
85.3
98.1
104.1
149.0
162.5
83.0
11.2
15.1
19.8
28.8
29.2
14.4
20.0
21.5
23.1
31.7
32.8
15.3
14.1
17.0
21.7
28.0
29.1
15.6
5.8
8.9
13.1
19.2
26.3
15.5
2018
2019
2020
2021
2022
H1 2023
GoogleSearch & Other Youtube Ads GoogleNetwork
GoogleOther GoogleCloud Other Bets
33.0% 32.3% 32.4%
38.7%
34.1% 35.1% 35.4%
-74.5%
-52.1%
-42.9%
-16.1%
-11.3%
2.6%
4.9%
2018 2019 2020 2021 2022 Q1 2023 Q2 2023
GoogleServices Total GoogleCloud
Operating Margin
Segment Rev in US $Bn
Google's Improved Growth
• Accelerating growth in Google’s core search
business demonstrated that the segment’s network effect
moat in search is intact despite threats from Microsoft and
Open AI.
• YouTube Ad revenue returned to growth due to a more
balanced mix of broad-based and direct response ad
demand, improvement in YouTube Shorts monetisation,
and increasing demand for ads on connected TVs.
• Revenue declined in Google’s Advertising offerings;
however, it is expected that this segment will improve as
economic uncertainty lessens and ad spending across the
internet picks up (Via: Morningstar) Source: Wikipedia, MIT, NextPlatform, Linkedin, Seekingalpha, Company Data, Google cloud, Markettech, TechCrunch, ServiceNow, Khaveen Investments

29
Google’s Search
Economics with Gen AI
• Today cost of Search is nearly zero (<0.4c), aided by
constant innovation and optimisation in the past 2
decades.
• AI is much more expensive, and Morgan Stanley estimates
that if Google had to run every search through Generative
AI averaging 100-word results, it would cost it an extra
$24Bn (in 2023) in operating costs. That will reduce its
Operating income of FY22 by 32%.
• According to Seeking Alpha Analyst, this is not a realistic
scenario. It is expected that search-related traffic costs are
going to rise from $49Bn (FY22) to roughly $80Bn in 2028.
In 2022 Traffic costs are 17% of Annual Rev.
• The proposed proposition appears to carry a significant
cost, albeit with the potential for substantial returns. It is
estimated that sales may increase by over $200Bn, whereas
search-related costs would only see a $31Bn uptick.
• The technology behind the search engine is powered by
Google Cloud. In FY2022, Google Cloud posted an
operating loss of $3Bn, but it is projected to increase to
operating profits of $18Bn by 2027. By 2027, it is estimated
to generate approximately $50Bn in EBITDA, which
accounts for just over 25% of all EBITDA for Google.
Google's EBITDA margin stands at 40%. Google Cloud made
its 1st Operating profit in Q1FY23 and is heading northwards.
• To put this in context, yes AI-powered Search is expensive
today, but with economies of scale, and advancements in
H/W and S/W (like opting for cheaper Accelerators and modular
inference engines), it is rapidly changing the dynamics of
search business and promises to be another Cash Cow
with similar margins if not higher.
Source: Wikipedia, MIT, NextPlatform, Linkedin, Seekingalpha, Company Data, Google cloud, Markettech, TechCrunch, ServiceNow, Khaveen Investments

30
Google’s Cloud Play
AI and ML Centric
• Google Cloud has come into its own over the last two quarters. 1st entering operating profitability in Q1 FY23, it subsequently doubled its
operating income contribution in Q2. YoY revenue growth remained particularly robust at 28% for both quarters. Operating margins nearly
doubled from 2.6% to 4.9%.
• As the Cloud offering becomes matured and with Gen AI rush, on top of the Digital Transformation tailwinds, the company’s Cloud division
has genuinely turned the corner into becoming a profit driver for Google.
• From here on with the economies of scale kicking in, and utilising spare capacity in storage, compute and network, both revenue and
operating margins are expected to head northwards and reach similar levels to that of other cloud providers.
• With Amazon's and Microsoft's Cloud operating margins at 29% and 43.1% respectively. Google has a significant distance to cover to match
its main rivals.
• Having said that there is an upside in the growth potential for Google Cloud and AI. This positivity is reflected in the market, which has seen
a surge in Market Cap by $400Bn since February 2023.
Key Indicators
• Comp. (Cloud) – AWS, MSFT, IBM
• Rev Growth – Up and Accelerating
• Growth Segment – Cloud
High
Low
Level
of
Functionality
High
AWS – 35%
Low Customer Adoption
MSFT– 23%
Google – 12%
IBM – 4%
Salesforce – 3%
Tencent – 2%
Alibaba – 5%
Oracle – 2%
Source: Wikipedia, MIT, NextPlatform, Linkedin, Seekingalpha, Company Data, Google cloud, Markettech, TechCrunch, ServiceNow, Khaveen Investments

31
Google’s AI Journey
Diversify & Make it Affordable
• According to Trading Places Research, Among the big AI players, Google is an anomaly. Unlike its competitors like, Microsoft and Meta which
rely heavily on Nvidia’s Accelerators or GPUs, Google on the other hand has designed and built its own AI Accelerator named TPU – Tensor
Processing Unit.
• Tensor Processing Units (TPUs) are relatively slower than Graphics Processing Units (GPUs) per chip, but they are significantly less expensive to
buy and operate at scale, making them a more cost-effective solution. The solution to this issue is to increase the number of TPUs used for a given
task. The TPU v4 is currently limited to 4,000 chips, whereas the TPU v5e will support tens of thousands of chips. This allows for larger models to
be trained and deployed on TPUs than previously possible, with the capability to handle up to two trillion parameters, which would be sufficient to
support GPT-4. The ability to rent out an entire data centre is one of the most significant advantages of TPU v5e.
• These TPUs were 1st built by Google in 2016 and were consumed internally. Subsequent versions followed and today we have v5e available
via Google Cloud. These new generation of TPUs are designed for cloud consumption keeping hype around Gen AI in addition to internal
consumption.
• As there is currently no data available on the cost-benefit of v5es, extrapolation from previous versions of TPUs suggests that customers could see a
cost savings of around 40-70% compared to Nvidia H100 for the same task on Google Cloud.
• One of the key advantages of these TPUs is their ability to handle much larger models (FM’s >1Tn parameter) than before. It is worth
noting that these TPUs offer significant cost relief to Google’s own extensive AI work.
• After building its own chips (TPUs) for AI and Cloud workloads, the next was to build a software layer for managing TPUs.
• . To overcome this obstacle, Google open-sourced its software library, TensorFlow, in 2015, which has been a success. Amazon is also adopting
a similar approach.
• Nvidia’s Moat lies in Software. Following Apple’s and Microsoft’s footsteps, Nvidia provided CUDA for free, creating a mouse trap since the early
2000s. Since then, Nvidia’s software stack has expanded, which is still being used by its customers.
• During this evolutionary journey, Google soon realised that it had built an advantage over Nvidia in the Cloud Computing and Data Centre
play, as their TPUs are more suitable for this type of infrastructure compared to Nvidia’s accelerators. The reason is, that Nvidia entered the
Cloud and DCN domain around 2014, whereas Google has been building Cloud and Data Centre infra for services like Gmail and YouTube
since the early 2000s.
• With this background on Google’s AI journey with TPUs’, there is another aspect of chips that Google concentrated on to reduce the cost
envelope. Instead of relying on Nvidia, Google collaborated with Broadcom, an early adopter of TPUs and a prior supplier of other
infrastructure gear to Google. This partnership has proven to be highly effective, as evidenced by the success of Google’s AI initiatives forcing
companies like Amazon and Tesla (Dojo Supercomputer) to build their own chip and software stack. The success of TPUs has made Intel, AMD,
and Nvidia all work on their own TPUs, but none of them have yet released a product that is as mature or as powerful as Broadcom’s TPUs.
• In summary, Google has been able to build an affordable AI infrastructure without solely relying on Nvidia which commands 70% of the AI
Chip Market. This will soon be reflected in its AI solutions offered to consumers and enterprises.
• Key Message
• Google has been able to build an affordable AI infrastructure without solely
relying on Nvidia which commands 70% of the AI Chip Market.
Source: Wikipedia, MIT, NextPlatform, Linkedin, Seekingalpha, Company Data, Microsoft, Markettech, TechCrunch, ServiceNow, Khaveen Investments

32
• Bard’s error-prone response, in a video demo posted online, included an answer suggesting the JWST was used to take the very
first pictures of a planet outside the Earth’s solar system, or exoplanets. This claim was identified as inaccurate by experts, which
resulted in investors selling off their shares, leading to a significant decline of over $100Bn in Market Cap.
• Since then, the dust has settled. The market feared that Bing’s integration with ChatGPT would rapidly expand its global search
market share from 3.02%, but that hasn’t happened yet. On the other hand, Google was able to quickly achieve success by
enhancing Bard, and its new AI-powered search function started giving more precise results with immediate sourcing. Bing and
ChatGPT attempted something similar but until now it hasn’t succeeded. Hence, Google continues to remain the dominant player
in the search market with over 90% market share, 92% share of search ads, and 39% share of global digital ads. Furthermore, its
market capitalisation has increased by $450Bn, which includes a $100 billion recovery from a previous decline.
• It is reasonable to assert that Google has taken measures to restore its mojo. Furthermore, recent developments in Gen AI seem to
corroborate this sentiment.
• Launched At Google I/O on May 23
• Autofill for Gmail, 3D path maps in Google Maps, Automated photo editing,
• Two new large language models called PaLM2 (340 Bn parameters) & Gemini (parameters not known).
• These products appeared more iterative and engineering-oriented in nature.
• More Significant Launches were at Google Cloud Next on Aug 23, which puts them on par with Microsoft and Amazon.
• Two new cloud products – Cloud TPU and A3GA
• Cloud TPU is a cloud-delivered ‘Tensor Processing Unit. This is a cloud service of an Accelerator (ASIC application-specific
integrated circuit) designed to be particularly efficient for AI workloads.
• A3GA is a cloud service of an AI-optimised supercomputer that leverages NVIDIA A100 GPUs.
• VertexAI provides APIs (standardised protocols for data transfer between computers) for accessing a large library of LLMs.
This is similar to AWS’ offerings in that it provides more of an infrastructure layer for accessing LLMs, each of which is
tailored to certain types of processing.
• Duet AI is Google’s response to MSFT’s Copilot (live with 600 customers, full launch in Nov). It is an AI integration that fits
into the entirety of Google’s Workspace products (like Google Docs & Sheets) as well as Google Cloud, where it helps
programmers write code.
Google’s $100Bn Blunder
Has it Recovered ?
• On Feb 7, 2023, Google’s competitor to Microsoft-backed
ChatGPT, Bard in a promotional material posted an error
in the response by the chatbot to, “What new discoveries
from the James Webb space telescope (JWST) can I tell my
nine-year old about?”
Source: TechCrunch, Intentwise, Forbes, Google Cloud Blog, Wikipedia, MIT, Bloomberg, Fourweek MBA, CNBC NextPlatform, Seekingalpha

33
Google
Risks to its Growth
Regulatory and Ongoing Lawsuits
• There are currently multiple antitrust cases against Google, in the EU (appealing against $2.6Bn in fine), DOJ, 20 with various State's Attorney
Journal. Besides, Gannett Newspapers recently (Jun 23) joined these ongoing legal battles and Media Alpha may follow in the future. It is
these ongoing lawsuits that are worrisome, esp; from businesses like Gannet Newspapers. Knowing this, the company could face additional
lawsuits in the future, the company is closely guarding $118Bn in cash and to date hasn't paid a single dividend.
Increased Competition from the likes of ChatGPT Powered Bing (Search)
• There is a potential risk if ChatGPT becomes a popular substitute for Google, especially among the younger generation. That
can have a domino effect on other services like Digital Ads. Currently, Bing's attempt to incorporate chat features has not been successful.
Nevertheless, Microsoft is striving to improve its AI-powered search and since It has a strong and dominant presence in the AI cloud it
aspires to expand its 3.02% search market share.
• Google is facing increasing competition from other tech giants, such as Amazon and Meta. These companies are all investing heavily in AI
and cloud computing, and they are all trying to compete with Google in its core markets.
Ongoing Interest rate hikes for managing Inflation
• Google’s stock valuation is strongly susceptible to market conditions like changes in interest rates impacting its WACC and FCF. This is
rightly reflected by its beta being more than 1. As pointed out by Seeking Alpha Analyst,
• Google's discounted cash flow valuation depends heavily on where interest rates go in the future. If we use the treasury yield as the discount rate,
then it's somewhere between fairly valued and undervalued depending on how much growth it can achieve. If we add a 6% risk premium to the
treasury yield, then the stock is overvalued unless massive amounts of growth are achieved. The long-term treasury yield is the absolute lowest
discount rate you can use in valuing a stock; the higher it goes, the less valuable stocks become, holding everything else constant. So, if the Fed
resumes its rate hikes in order to fight inflation, Google will require ever more growth in order to be theoretically "worth it."
Economic Slowdown
• Google primarily makes money from the advertising business, which is vulnerable to cyclicality. The economic slowdown will negatively
affect advertising expenditures, resulting in an adverse impact on both the top and bottom line of the company
Privacy concerns
• Google collects a lot of data about its users, and this has raised privacy concerns. If users become more concerned about their privacy, they
may be less likely to use Google's products and services. The US Congress has raised issues about Google collecting users' data and causing
privacy issues many times in the recent past and is again pursuing this matter. In 2018, Senator Ed Markey introduced a bill called the "Do
Not Track Act," which would have required websites to obtain users' consent before tracking their online activity. The bill was not passed by
Congress.
Technological Change
• Google's business is based on a number of technologies, such as search, advertising, and cloud computing. If these technologies change
rapidly, Google may not be able to adapt quickly enough, and this could hurt its business.
Source: TechCrunch, Intentwise, Forbes, Google Cloud Blog, Wikipedia, MIT, Bloomberg, Fourweek MBA, CNBC NextPlatform, Seekingalpha

The book is available at Amazon.com

Disclaimer: This presentation acknowledges and gives credit to the work of others. Necessary validation has been
taken to avoid copyright infringement. Any instances that violate terms can be removed when notified. All
discussed thoughts & opinions are my own & not those of my employer or other parties.
For further information, please contact:
Name: Vishal
Address: Melbourne, VIC 3000 Australia
Mobile: 0468 675 566
Blog: https://blog.sharmavishal.com/
Thank You

Generative AI - The New Reality: How Key Players Are Progressing

Recommended

More Related Content

What's hot (20)

Similar to Generative AI - The New Reality: How Key Players Are Progressing (20)

More from Vishal Sharma (20)

Recently uploaded (20)

Generative AI - The New Reality: How Key Players Are Progressing