ChatGPT What is it Why it Matters Is it a Google Threat...and Broader Industr.pdf
100墨值下载
Brian.Nowak@morganstanley.com
Kieran.Kenny@morganstanley.com
Chloe.Currie@morganstanley.com
Matthew.Cost@morganstanley.com
Matt.Bombassei@morganstanley.com
Julian.Herrera@morganstanley.com
Dave.Nwokonko@morganstanley.com
Patrick.Ho@morganstanley.com
In-Line
MORGAN STANLEY & CO. LLC
Brian Nowak, CFA
EQUITY ANALYST
+1 212 761-3365
Kieran Kenny, CFA
RESEARCH ASSOCIATE
+1 212 296-5137
Chloe A Currie
RESEARCH ASSOCIATE
+1 212 761-2158
Matthew Cost
EQUITY ANALYST
+1 212 761-7252
Matt Bombassei
RESEARCH ASSOCIATE
+1 +1-212-761-9811
Julian Herrera
RESEARCH ASSOCIATE
+1 212 761-1784
Dave Nwokonko
RESEARCH ASSOCIATE
+1 212 761-4774
MORGAN STANLEY & CO. INTERNATIONAL PLC+
Patrick A Ho
RESEARCH ASSOCIATE
+1 212 761-2764
Internet
North America
IndustryView
InternetInternet || North America North America
ChatGPT: What is it? Why it
Matters? Is it a Google
Threat...and Broader Industry
Implications
Investor questions about OpenAI and ChatGPT have picked up
after its viral launch. We detail what ChatGPT is, why we
believe Google Search is still well positioned (and set to
improve with its own AI/ML offerings), broader industry and
stock implications, and what to watch for next.
Who is OpenAI? OpenAI is an artificial intelligence research company founded in
2015 by Sam Altman, Elon Musk, and several others. Its mission is to ensure that
AI benefits all of humanity by building safe artificial general intelligence. It
started as a non-profit and evolved into a "capped profit" company – hybrid of a
for-profit (OpenAI LP) and nonprofit (OpenAI Nonprofit entity) – in order to raise
additional capital, attract talent, and scale. As a result investors and employees
can receive a capped return. Returns in excess of the capped return are owned by
the OpenAI Nonprofit. We believe OpenAI LP has several hundred employees,
and to-date has raised $2bn+ ($1bn investment from MSFT, and $1bn of grants
from AWS, Infosys, Y Combinator, Peter Theil, and Reid Hoffman). It has three
primary offerings today: ChatGPT, DALL-E 2, and Whisper.
Meet OpenAI's ChatGPT: OpenAI recently launched ChatGPT, a large-scale
natural language model with a prompt for users to submit questions into.
ChatGPT then responds with human-like (coherent/natural) answers. In simple
terms, it's a broad-based, highly-sophisticated chat bot that offers answers that
read like they are coming from a human (see examples in Appendix). GPT was
trained over the last ~3 years by ingesting large amounts of text data across the
internet. It launched to the public on 11/30/22 and amassed 1mn users within a
week. Its interest on Google Trends (See Exhibit 1) and Twitter speak to the high
level of interest in this notable next step in the use of AI.
Is Chat GPT a Structural Threat to Google Search? The viral launch of ChatGPT
has caused some investors to question whether this poses a new disruption
threat to Google Search. The Google bear case would be that language models
could continue to gain users and take share of Search queries and disrupt
Google's position as the entry point for people on the Internet. While we believe
the near-term risk is limited – we believe the use case of search (and paid search)
is different than AI-driven content creation – we are not dismissive of threats
from new, unique consumer offerings. Think about the potential AI-driven use
cases around product reviews or travel research (to name just two), which over
Morgan Stanley does and seeks to do business with
companies covered in Morgan Stanley Research. As a
result, investors should be aware that the firm may have a
conflict of interest that could affect the objectivity of
Morgan Stanley Research. Investors should consider
Morgan Stanley Research as only a single factor in making
their investment decision.
For analyst certification and other important disclosures,
refer to the Disclosure Section, located at the end of this
report.
+= Analysts employed by non-U.S. affiliates are not
registered with FINRA, may not be associated persons of
the member and may not be subject to FINRA restrictions
on communications with a subject company, public
appearances and trading securities held by a research
analyst account.
1
December 12, 2022 11:34 PM GMT
time could cause a change in consumer behavior. That said, we now detail two
key points that (for now) leave us confident in Google's durable position.
1. Scale and the 10X Rule Matter: Google has an estimated ~4bn+ daily users,
with billions of searches per day. While we are not dismissive of disruptive
technologies, we continue to believe that technologies need to offer 10X better
solutions than the second best offering to drive significant (aka billions of
people) behavioral change. The volume and consistency of consumers' activity
speaks to how they derive utility from Google (see even our recent AlphaWise
survey on Google as a starting point here)...so driving significant differentiation
and change will be challenging. Google Search also continues to improve with
thousands of changes made to paid search per year.
2. Google Invests Hard in AI/ML, Already Has Multiple Products and (We
Believe) More To Come: Google has spent $100bn/$79bn on R&D and capex
over the past 3 years alone and has ~187k employees (up from 119k in 2019). We
don’t see this letting up either, modeling R&D to grow at a 13% '23-'25 CAGR and
for infrastructure annual capex to reach ~$30bn by '25. We estimate roughly
50% of GOOGL's headcount is engineers, with our industry conversations
continuing to indicate GOOGL is the big cap tech leader of AI/ML research and
investment. Google is also building similar natural language models such as
LaMDA, which it showed powerful examples of in 2021 (see here). Google hasn't
released LaMDA broadly to the public, but we look for further products to come
over time. Google has also been investing in BERT, a machine learning
architecture that helps Google process language and better understand context,
for 5+ years. There are likely more products we don't even know about too, as
consider that last Friday, GOOGL's Deepmind announced the release of
Dramatron, a vertical specific language model enabling writers to co-create
theatre and film scripts (complete with title, characters, location descriptions,
and dialogue).
Broader Industry Implications from ChatGPT? Expect AI and ML Investment to
Continue to Ramp: GOOGL has been speaking about investing in AI/ML for
nearly 20 years (first mentioned in 2005 10k) and on its 3Q:22 call spoke to AI
powered Search (and large language models) at the top of its 4 key investment
initiatives (AI, YouTube, Hardware, Cloud). META's step up to $35.5bn in
estimated '23 capex is largely driven by further investment in AI/ML and higher-
end GPUs to allow for more analytics and computations of its leading first party
data. We also expect PINS/SNAP to spend more with third party cloud services in
'23 in order to use more AI/ML models and capabilities. Expect the investment in
AI/ML to continue to ramp...and for the engagement and monetization
capabilities of these large, leading platforms to expand in '23/'24 and beyond.
From our perspective, this will put the scale leaders (with more ability and
willingness to invest) at an advantage. As a read across, this would also be a
tailwind to component vendors such as NVDA, AMD, MRVL, INTC, AVGO, ANET,
and PSTG. NVDA/AMD/MRVL/INTC/AVGO are covered by Joe Moore, ANET/PSTG
are covered by Meta Marshall.
What to Watch for Next From ChatGPT?
1) What Will Improve With GPT-4, the next iteration of OpenAI's GPT algorithm?
The launch timing is uncertain (some sources suggest 2023), but what
improvements come will matter. We will be very focused on improvements to
2
of 14
100墨值下载
【版权声明】本文为墨天轮用户原创内容，转载时必须标注文档的来源（墨天轮），文档链接，文档作者等基本信息，否则作者和墨天轮有权追究责任。如果您发现墨天轮中有涉嫌抄袭或者侵权的内容，欢迎发送邮件至：contact@modb.pro进行举报，并提供相关证据，一经查实，墨天轮将立刻删除相关内容。
下载排行榜
文档被以下合辑收录
评论