没有合适的资源?快使用搜索试试~ 我知道了~
【PitchBook-2024研报-】新兴技术未来报告:更新我们的人工智能展望(英).pdf
1.该资源内容由用户上传,如若侵权请联系客服进行举报
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
版权申诉
0 下载量 11 浏览量
2024-10-25
19:22:32
上传
评论
收藏 1.4MB PDF 举报
温馨提示
行业研究报告
资源推荐
资源详情
资源评论
1
PitchBook Data, Inc.
Nizar Tarhuni Executive Vice President of
Research and Market Intelligence
Paul Condra Head of Emerging
Technology Research
Institutional Research Group
Analysis
The Emerging Tech Research Team
Publishing
Designed by Jenna O’Malley
Contents
pbinstitutionalresearch@pitchbook.com
Published on September 30, 2024
Introduction 1
Enterprise applications 3
AI & ML 3
Crypto 6
Data analytics 8
Enterprise SaaS 10
Fintech 12
Information security 15
Insurtech 17
Consumer applications 19
E-commerce 19
Gaming 21
Industrial applications 23
Agtech 23
Climate tech 25
Defense tech 27
Foodtech 29
Mobility tech 32
Space tech 34
Supply chain tech 36
Healthcare applications 38
Biotech 38
Digital health 40
Healthcare IT 43
Medtech 46
EMERGING TECH RESEARCH
Emerging Tech Future Report:
Updating Our Generative
AI Outlook
Models thrive while complexity, costs
impact applications
PitchBook is a Morningstar company providing the most comprehensive, most
accurate, and hard-to-find data for professionals doing business in the private markets.
Brendan Burke; Robert Le; Derek Hernandez;
Rudy Yang; Eric Bellomo; Alex Frederick; John
MacDonagh; Ali Javaheri; Jonathan Geurkink;
Kazi Helal, Ph.D.; Aaron DeGagne, CFA; Rebecca
Springer, Ph.D.
Introduction
In May 2023, our team published perspectives on how Generative AI (GenAI) was
poised to impact various industries and technologies. This note revisits those
perspectives with fresh takes on how GenAI is (or is not) manifesting itself and how
expectations have changed or evolved.
The rise of ChatGPT in early 2023 was a pivotal moment, marking the point when
AI became understood as an easily adaptable technology with the potential for
broad application. Since then, investment in related technologies and startups has
skyrocketed, highlighted by intense competition in the foundation model space as
new startups and technology incumbents have aggressively jumped into the fray.
The GenAI-infrastructure landscape is stratifying across several use cases, such as
on-device inference, domain-specific knowledge, and simply raw power to produce
the best results the fastest. Some form factors, such as general consumer search
and chatbots, have emerged as battlegrounds for tech giants like Google and Meta
as they seek to keep users within their ecosystems. Other strategies include more
specialized applications, such as personal assistants for enterprise use cases and
software development.
2
Emerging Tech Future Report: Updating Our Generative AI Outlook
However, as our analysts describe in this note, several blockers to adoption remain
despite intensified efforts toward AI transformation. These include high compute
costs, data availability, data security, and overall system complexity. Whereas much
progress has been made at the foundation model level, where investment capital
appears endless, application-level startups face a more challenging fundraising
environment as they feel near-term pressures to demonstrate commercial viability.
Paul Condra
Head of Emerging Technology Research
paul.condra@pitchbook.com
GenAI software spending estimate ($B)
$7.5 $17.0 $32.4
2023 2024 202 5
Source: IDC • Geography: Global • As of August 20, 2024
GenAI VC deal activity
$3.8 $14.3 $9.3 $26.0 $23.9
349
544
582
877
508
2020 2021 2022 2023 2024
Deal value ($B) Deal count
Source: PitchBook • Geography: Global • As of August 20, 2024
3
Emerging Tech Future Report: Updating Our Generative AI Outlook
Enterprise applications
AI & ML
Prior expected impacts
At the outset of GenAI’s irruption, we saw the limitations of large language models
(LLMs) for enterprise use cases and early signs of trends that are now maturing.
Expected innovations included the maturation of a supporting LLM operations
(LLMOps) industry, including foundation model orchestration and vector search,
along with AI agents. The need for supporting software for LLMs has been
exacerbated by the limitations of successive model releases after OpenAI’s GPT-4
and the proliferation of copycat open-source models, requiring users to get higher-
quality outputs from similar generative models. In the long term, we expected
foundation models to create more decacorns valued at over $10.0 billion and
code generation to progress the field to a greater extent than image generation
or chatbots.
Reality one year later
GenAI has transformed the existing AI & machine learning (ML) vertical in
fundamental yet still limited ways. While new LLMs represent the future of the
industry, they have not extinguished legacy approaches, and pre-existing models still
outnumber LLM applications. GenAI software will contribute only about 14% of AI
software spending in 2024 with $14.5 billion and is on pace to contribute only 32.3%
of spending by 2028, according to IDC estimates.
1
Even so, companies building legacy
ML models have seen their estimated valuations plunge, including DataRobot’s by
over 90% and H20.ai’s by over 80% in the face of GenAI disruption.
2
New research
into model techniques can further progress the field in deterministic areas of
software, as covered in our analyst note on foundation models. LLM innovators are
capturing mindshare and market share from their predictive predecessors.
Brendan Burke
Senior Analyst, Emerging Technology
brendan.burke@pitchbook.com
1: “Worldwide AI and Generative AI Spending Guide,” IDC, Karen Massey, et al., August 20, 2024.
2: “Caplight MarketPrice,” Caplight, August 28, 2024.
AI-centric software spending estimate by type ($B)
$0
$20
$40
$60
$80
$100
$120
$140
2023 2024 2025
GenAI Predictive AI
Source: IDC • Geography: Global • As of August 20, 2024
4
Emerging Tech Future Report: Updating Our Generative AI Outlook
Enterprise rollouts have progressed more slowly than initially forecast, and models
have not continued to take on new capabilities. Because of the tepid adoption
of applications, the infrastructure layer, including model architecture labs and
semiconductor startups, has achieved 25 of the 39 unicorn valuations we have
tracked in the LLMOps space. Semiconductors, model research labs, and startup
cloud providers have shown the need for a new stack and have demonstrated
the ability to create new software innovations without a significant surrounding
ecosystem. We did not anticipate the demand for startup cloud providers that have
achieved high valuations, including CoreWeave, Crusoe, Lambda, and Together AI.
New semiconductors have achieved breakthroughs in LLM inference and datacenter
networking, building on the NVIDIA GPU ecosystem, including those of Astera Labs,
Cerebras, and Groq.
Other software categories face challenges to prove their legitimacy. We tracked a
doubling of VC deal count for GenAI operations software in 2023, including in data
preparation, model orchestration, and application deployment, and 2024 is on pace
for a further 50% growth. Deal value has not kept up with the infrastructure layer,
however. These LLMops companies have not grown large independently, given
the spectrum of configuration options and continuously improving features from
hyperscalers. Vector databases in particular have become commoditized and are
unlikely to present a growth category as open-source options extend their network
effects and incumbent databases offer vector support. The AI agent space has
become crowded, yet we believe it will be disrupted by more action-oriented model
capabilities. Few acquisitions of model orchestration companies have been made
to justify early-stage VC investments as acquirers monitor the monetization of
LLM tools.
Real-world progress
In the long term, commercial gains will likely come before artificial general
intelligence (AGI) potentially renders software irrelevant. Incumbents have taken
more commanding positions than was clear last year via aggressive startup
investments. We predicted that more $10.0 billion companies would be created
after OpenAI, which has proven true with Anthropic, CoreWeave, and Scale AI,
yet other contenders have fallen short of that total before succumbing to Big
Tech offers, including Adept AI, Character.AI, and Inflection AI. AI in software
development has accelerated to widespread adoption, with large customers relying
on AI code generation. Coding assistant startups raised over $1.0 billion in H1 2024
after raising only $480.6 million in 2023, showing the success of the technology
and scale of the market. Generative media lags expectations, facing VC funding
declines in multimedia content suites and video generation. Vertical-focused
companies face accusations of vaporware as they align general-purpose LLMs with
customer workflows and occasionally wait for base models to improve before their
products do.
PitchBook users can access a full list of AI
agent startups here.
5
Emerging Tech Future Report: Updating Our Generative AI Outlook
Key recent GenAI VC exits and talent acquisitions
Source: PitchBook • Geography: Global • As of June 30, 2024
Company Close date (2024) Segment Category Exit value ($M) Acquirer
Character.AI August 22 AI core Model architecture $2,500.0 Alphabet
Adept AI June 28 AI core Model architecture N/A Amazon
Clickable June 26 Visual media Content suite N/A Beehiiv
Argilla June 13 AI core Model architecture N/A Hugging Face
Uizard May 24 Code Testing N/A Miro
Deci May 2 AI core Deployment $300.0 NVIDIA
Mirage April 8 Visual media 3D models N/A Harvey
PartyKit April 4 AI core Orchestration N/A Cloudflare
Inflection AI March 21 AI core Model architecture $650.0 Microsoft
DarwinAI January 1 Vertical applications Industrial N/A Apple
剩余46页未读,继续阅读
资源评论
soso1968
- 粉丝: 2270
- 资源: 1万+
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
最新资源
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功