Silicon Valley Observations: Startups, Investment, and Tech Giants — What Changes and What Doesn't in the Generative AI Wave | Yunqi Capital Tech Notes

Early summer in the Bay Area isn't particularly hot, but the Palace of Fine Arts in San Francisco was already swarming with intense energy.

This was the scene at GenAI Summit 2024. The conference had been underway for over half an hour, yet long queues of attendees still stretched outside the venue waiting to enter. Despite some grumbling about the event organization, the star-studded speaker lineup and the promise of networking opportunities proved irresistibly attractive — reportedly, tens of thousands of tickets were sold, starting at $100 per day. Among the crowd were many Chinese attendees who, like us, had traveled from China.

This is a snapshot of Silicon Valley in the throes of the GenAI frenzy. In this wave that few dare to miss, Silicon Valley — that enduring beacon of innovation — remains a critical coordinate system for gauging market temperature and direction.

Yunqi Capital's frontier technology team maintains an international perspective and has sustained high-frequency engagement with the Silicon Valley venture ecosystem over the long term (see "Recommended Reading" at the end for past Silicon Valley observations). Nearly two years into the GenAI wave, what has changed and what hasn't in Silicon Valley's AI venture scene? This article shares our firsthand observations across dimensions including venture atmosphere, startup directions, investment styles, and the competitive posture of tech giants.

AI Venture Atmosphere

Year Two of the Wave: Early-Stage Entrepreneurship and Angel Investment Remain Active

It has been almost two years since ChatGPT broke into the mainstream. The generative AI surge has reshaped the product landscape considerably. On the model layer, the capability iteration curve of the first tier led by GPT has visibly flattened. On the application layer, legions of AI apps that failed to achieve product-market fit have fallen by the wayside — AI Graveyard, a site tracking discontinued AI tools, has already cataloged over 700 such projects.

Yet we observe that compared to the early days of the wave in 2023, both entrepreneurs and investors at the early stage maintain considerable enthusiasm for entering the AI赛道.

1. New AI Projects Keep Pouring In

Large numbers of entrepreneurs continue to flock to AI. One telling data point: AI projects accounted for over 50% of YC's Winter 2024 batch.

And for most entrepreneurs, securing a first round of funding isn't too difficult. From our sample of conversations, roughly 80% or more have raised seed rounds in the hundreds of thousands of dollars.

In terms of founder profiles, our observation is that those active at the early stage tend to be post-95s and post-00s, many fresh out of school or even still enrolled. This differs from China's ecosystem, where senior executives from major tech companies or university scholars typically take center stage.

Additionally, Chinese entrepreneurs remain a significant force in this AI wave. For a long time, Chinese engineers have been well-represented across Bay Area tech giants like Google, Apple, Facebook, Microsoft, and NVIDIA. This pattern persists today — beyond Chinese engineers from major companies, newly graduated Chinese students constitute another active cohort. Pika, HeyGen, and others are relatively representative examples.

2. Non-Institutional Angel Investors Are Writing Checks Enthusiastically

A mature funding environment is crucial soil sustaining Silicon Valley entrepreneurs' passion and confidence. For a substantial portion of founders, securing first-round funding isn't difficult because at the seed/angel stage — where funding needs typically don't exceed $3-4 million — a large volume of non-institutional individual investors serves as an important capital source alongside institutional investors.

Compared to institutions, these backers have "looser hands." Passion and personal interest are significant drivers of their investment decisions. In many cases, they may simply be moved by a founder's background or a business plan not yet substantiated by product, and proceed to put up $500,000 to $1 million. Such diverse funding sources mean lower barriers to entry for entrepreneurs.

Meanwhile, Silicon Valley's entrepreneurial culture, carried forward to this day, has cultivated relatively forgiving room for trial and error. This is another important factor keeping both startup and investment activity vibrant at the early stage.

AI Startup Directions

ToB: The Mainstream Choice for Both Entrepreneurs and Investors

The US toB market, rooted in the 1950s, is stable and well-established. From our observations in Silicon Valley, AI has been applied earliest in toB contexts during this wave, and it is also the direction preferred by both entrepreneurs and investors.

1. AI Applied First in ToB

Compared to consumer-facing applications, AI as an efficiency-enhancing technology has more often been adopted "silently and invisibly" first in toB scenarios like customer service, marketing, finance, legal, and healthcare.

Mature toB software also embraced AI at the earliest opportunity. Leading SaaS companies like Salesforce and Databricks announced their AI strategies and features in early 2023.

2. 70-80% of Startups Target ToB, Cutting In from Niche Scenarios

We observe that toB is the choice of 70-80% of AI entrepreneurs, with very small, very vertical entry points. Beyond customer service, marketing, healthcare, education, legal, and finance, numerous sub-sectors and niche scenarios host many AI startups — they typically don't pursue breadth and comprehensiveness.

In the legal赛道, for instance, Harvey — incubated by OpenAI's fund — primarily serves lawyers with contract analysis, contract clause revision, and legal research assistance. It has attracted clients including law firm Allen & Overy and accounting firm PwC, and was valued at $700 million as of July this year. Another legal service product, DoNotPay, positions itself around consumer rights protection, playing the role of an "AI lawyer" by assisting defendants in fighting traffic ticket dismissal appeals and other legal services. Its $36 annual fee is far below the cost of hiring a lawyer.

3. Growth-Stage Investors Favor ToB Projects

ToB projects tend to have clearer user profiles and PMF validation, so institutional investors from a rational perspective also lean toward toB investments. Such projects offer more predictable returns and lower risk. Overall, from Series A onward, especially at later funding stages, mature institutional investors show stronger willingness to invest in toB projects, with B-side investments comprising a relatively high proportion of the total.

4. Challenges Facing Consumer Application Startups

Consumer applications are more affected by shifts in base model capabilities. Every major model update from OpenAI, Google, and others squeezes innovation space for consumer apps. It's evident that consumer applications like Midjourney and Stable Diffusion, which accumulated first-mover advantages at the dawn of the AIGC wave, have gradually faded with each GPT iteration. After passing through one phase, how to confront the challenges posed by base model iteration becomes an unavoidable question. This is one reason why both entrepreneurs and investors have grown more cautious about committing to consumer applications.

Changes Among Institutional Investors

Under Macro Cycles, Investment Institutions Have Become More Prudent

Despite the friendliness and optimism of Silicon Valley's venture ecosystem carrying forward, the impact of macro cycles has nonetheless transmitted to institutional investment strategies.

1. Declining Institutional Activity, Rising Investment Standards

Since late 2022, Silicon Valley funds' investment strategies and attitudes toward AI projects have in fact shifted. Generally, once projects reach Series A and beyond, institutional investors become the primary backers. Against the backdrop of dollar interest rate hikes and tightened IPO channels, institutional investors have become less active and raised their investment standards accordingly. For example, at Series A, institutions may have raised their ARR benchmark from the previous $1 million requirement by 1.5-2x.

Thus, if a startup remains in a cash-burning mode without an established business model, it will face considerable pressure raising Series A or B rounds.

2. Institutional Capital Highly Concentrated in Headline Projects

Reflected in investment data: while overall AI investment has grown over 40% year-on-year, approximately 80% of capital is concentrated in large headline projects. Many investors believe that rather than investing in 100 small projects, it's better to concentrate on 10 particularly promising large ones. For example, veteran research teams out of DeepMind, or well-known figures like GitHub's co-founder — these projects might command valuations reaching hundreds of millions in their first round. This trend makes fundraising harder for mid- and long-tail projects, while individuals or teams with brand effects find capital more accessible.

3. Entrepreneurs Pay Greater Attention to Efficiency and Costs

This shift has also transmitted to startups' operating styles. First, many entrepreneurs have begun focusing more on commercialization, hoping to achieve self-sufficiency quickly and avoid dependence on continuous fundraising. Second, they exercise more cautious cost control — for instance, in hiring, they avoid being particularly "loose with money," instead emphasizing efficiency and cost discipline.

This also shows up in team configurations. Given the high labor costs in the Bay Area, we observe that many startup teams may be based across the US, or even around the world — Eastern Europe, Indonesia, Pakistan, and elsewhere.

Big Tech's Offensive

Multiple Advantages Highlighted, Startup Pressure Mounts Sharply

This AI wave was ignited by ChatGPT's mainstream breakthrough. Compared to emerging startups like OpenAI, tech giants like Google and Microsoft initially appeared somewhat behind in AI innovation momentum, but they quickly regained confidence — not only manifested in the series of large models and AI applications they launched, but also in landmark moves like major AI startups Inflection and Character.ai being acquired by large companies.

1. Big Tech Advantages Reinforced: Compute, Data, Traffic, Talent

Like alchemy requiring furnaces and raw materials, the giants have more furnaces and more materials. Their advantages in compute, data, traffic, and talent are precisely the bottlenecks constraining many startups. Most intuitively, traffic advantages: Apple's iPhone, Google's Chrome browser, Meta's Facebook — these are already mature traffic entry points. Big tech AI features don't require massive customer acquisition spending; they simply embed AI functionality into existing products. Startups face the opposite situation.

The same holds for talent. While leaving big companies to start ventures has been a recent trend, founders flowing back to big tech is becoming a new trend. For example, following recent acquisitions, Character.AI's people returned to Google, Inflection's to Microsoft. Combined with already deep talent reserves, this further solidifies tech giants' advantages.

2. Tests for Startup Direction: More Niche, More Precise

Looking across the overall landscape of US AI startups, the vast majority remain in early stages before achieving PMF. This ecosystem resembles China's. Big tech's catch-up and inherent advantages mean increasingly narrow breakout space for startups. How to excavate niche scenarios and find growth vacuums beyond big companies' strongholds has become critical.

This partly explains why AI hardware has become one of the hottest startup directions — because Silicon Valley is a "very soft" place, and Apple is virtually the only big company with hardware success, making it seen by many entrepreneurs as an opportunity.

3. Tests for Founding Teams: More Versatile, More "Hexagonal Warriors"

Beyond technical expertise, exceptional product and business acumen are also core traits bringing founding teams closer to success. In Silicon Valley's relatively mature and top-tier entrepreneurs, we see such "hexagonal warrior" qualities.

Particularly impressive was Glean, an AI enterprise search and knowledge management platform. Founded in 2019 by former Google search engineer and Rubrik co-founder Arvind Jain, the company has raised through Series D at a valuation exceeding $2.2 billion, with projected annual revenue reaching $100 million this year. As a serial founder who has built two unicorns, Jain appeared at a booth at a Silicon Valley event. In conversation with us, he mentioned that he was manning the booth himself to directly engage with product users and people genuinely interested in the product — he wanted to chat more with them. Additionally, Jain's erudition and curiosity across many topics were refreshing.

Conclusion

Decade after decade, Silicon Valley as a landmark of technological innovation has always surged with a spirit of free and courageous innovation. An important lesson for us is "truth emerges from practice." When exploring an unprecedented technology and its path to commercialization, forgiving soil for trial and error is important and precious. Silicon Valley's currently active, rich, complete, and positive AI venture ecosystem is a vivid embodiment of this.

Though the future brims with uncertainty, when trends become clear, proactive posture and mindset may help us reach answers faster.

Feel welcome to leave comments and exchange good perspectives and new discoveries about Silicon Valley with us (＾－＾)V