From: allin
A significant “AI FOMO frenzy” is currently underway, characterized by massive fundraising in the generative AI space [00:06:50]. Recent reports indicate there are now 500 generative AI startups, excluding Microsoft’s 11 billion [00:07:19]. This activity spans across audio, image, and text generation, signaling a multimedia revolution akin to the rise of personal computers in the 1990s [00:07:29].
AI as a Platform Shift
Many prominent investors view AI as the next major platform shift.
“I actually think that AI is the next platform shift in the same way that mobile was the one before internet was the one before. So I think AI is real, but I said earlier we’re going to overestimate it in a short term, we’re going to invest in everything in the same way that in 1999 we invested in everything. But then Google came out of that or Facebook came out of that.” [00:07:54] — Doug Leone, Sequoia Capital
Brad Gerstner concurs, suggesting AI could be even bigger than both the internet and mobile platform shifts [00:11:45]. He highlights that the 20 billion annual spending on AR/VR [00:11:59]. David Sacks echoes this, stating that AI is definitely on par with major platform shifts like mobile, social, or cloud [00:15:04].
Investment Strategies and Market Dynamics
The current investment climate for AI is driven by a complex interplay of incentives, value capture considerations, and evolving market conditions.
Investor Incentives and Capital Deployment
Chamath Palihapitiya emphasizes the importance of understanding incentives in investment, citing Charlie Munger’s dictum: “Show me the incentive and I’ll show you the outcome” [00:08:51]. With Credit Suisse offering 6.5% for a three-month T-bill, venture investors are under immense pressure from their Limited Partners (LPs) to deploy capital [00:09:07]. For long-term startup investments (10-15 years), returns need to be 20-25% to make sense against a 6.5% short-term risk-free rate [00:09:35]. This pressure often leads to capital being “torched” as many companies won’t amount to much, creating excessive correlation and zero time diversity [00:10:37].
Brad Gerstner, however, points out that LPs typically have a 10-year view and understand that current rates won’t last forever [00:27:08]. Instead of pausing investments, they are narrowing their focus to the top 10% of venture firms that demonstrate a strong track record and selectivity [00:28:04]. This means sub-scale funds with no distributions (DPI) will face significant challenges [00:28:52].
Value Capture in the AI Ecosystem
The question of where value will be captured in the AI ecosystem remains uncertain. Brad Gerstner draws a parallel to the internet boom: investing in early search engines like Yahoo or Infoseek led to losses, while Google emerged as the dominant player [00:12:46]. He suggests that value could accrue to major players like Microsoft and Google at the foundation model level, similar to iOS and Android [00:13:20].
Chamath Palihapitiya offers a “pick and shovel” framework, akin to the 1849 Gold Rush:
- The “Gold Panners”: Companies building large language models [00:18:46].
- The “Pick and Shovel Providers”: Companies at the silicon layer, re-architecting compute for AI [00:19:01]. Firms like AMD and Nvidia are seen as clear winners [00:19:07].
- “White Truffles”: Unique, singular sources of data that, when used in reinforcement learning, significantly enhance AI output. Examples include Facebook and Quora, but startups gathering sufficient data could also become “white truffles” [00:19:26]. Chamath notes that most current AI investments are in the “bologna in the middle” (commoditized models), rather than silicon or unique data [00:20:06].
OpenAI’s API and Developer Ecosystem
OpenAI recently announced a 90% reduction in the metered rate for its ChatGPT 3.5 API, with a 4.0 version planned later in the year [00:20:26]. An API (Application Programming Interface) allows developers to build on top of a service, meaning companies like Notion can integrate large language models without needing in-house AI expertise [00:20:51].
Chamath suggests that OpenAI’s strategy might be to cut costs so drastically that it becomes unfeasible for others to compete, allowing them to take a “minuscule take rate” and become a pervasive, very large company, similar to cloud computing [00:23:54]. This potentially explains their significant deal with Microsoft, as they may need substantial capital to subsidize their developer platform with negative gross margins for a period [00:24:51].
David Sacks points out that many founders are strategically marketing themselves as AI companies by building even one feature on top of the OpenAI API, capitalizing on VCs’ desire to invest in the “next big wave” [00:22:43]. This leads to a rapid increase in the number of purported AI startups [00:23:11].
The Current Investment Climate
Jason Calacanis notes that the current environment, with more measured AI investments and milestone-based funding, feels like the “best it’s been for me as an angel investor seed investor or seed fund for a long time” [00:48:20]. Deals are taking longer to close (six weeks), and founders are more thoughtful about deploying capital, building models, and managing salaries and hiring [00:48:41].
Sacks differentiates between the tech ecosystem and the broader economy, arguing that tech has already experienced its bubble and crash (2021-2022), and new companies are starting with a clean slate [00:49:14]. He believes that innovation will continue regardless of macroeconomic conditions [00:50:06].