Interview, Fireside Chat, Podcast
Michael Eisenberg: How China Could Overtake the US in the AI Race | E1167
- Foundation models are characterized as the fastest appreciating asset in history, with a prediction that AI will become the most transformational technology of the lifetime, driving a massive financial gold rush and creating a future division between "AI countries" and "non-AI countries."
- The speaker anticipates a massive financial gold rush alongside AI's transformation, noting that while bubbles will generate necessary infrastructure and allow some foundational companies to succeed, many investors will lose significant money and the venture business will remain binary with only one or two model companies generating substantial returns.
- Regulatory threats in Europe are expected to force startups to relocate, with heavy regulation predicted to set Europe back in competitiveness for the next 20 years, while the US, China, and Israel are viewed as key less-regulated markets advantaged by competition and geopolitical strategy.
- Future economic landscapes will differentiate into AI and non-AI nations, with China expected to be highly sophisticated in manipulating minds via algorithms despite potential gaps, and the US needing to maintain technological competitiveness against such threats.
- Market dynamics will shift as large cloud providers like Google and Amazon acquire foundation models, while a battle for dominance is predicted between Anthropic and OpenAI, though current models visually appear similar with moats that are not yet obvious.
- User behavior is expected to evolve so that daily interactions begin on AI screens fetching articles rather than visiting news websites, accompanied by the emergence of AI agents and a shift where most activity is built into hyperscaler stacks.
- Horizontal SaaS processes face disruption as companies increasingly build custom software rather than buying external tools, with Amazon potentially stopping external software purchases; this shift is predicted to occur within the next five years.
- Legacy companies face significant challenges in catching up to AI-native firms, particularly if their data is not correctly configured, making it harder to adapt compared to the Internet era, while per-seat pricing models are expected to struggle against value-based pricing.
- Corporate buyers are expected to prioritize new revenue opportunities over incremental efficiency, leading to a market where companies go public with modest revenues as low as $60 million or $100 million, potentially commanding a 2x premium within 18 months.
- The IPO window is expected to remain wide open despite concerns about pricing, with a belief that companies like Reddit can succeed publicly, whereas Public Equity (PE) funds will buy only a tiny percentage of companies due to higher interest rates and slowing software purchase markets.
- Defense technology is predicted to become a significant investment area as AI helps close the kill chain globally, though selling to government buyers requires specific expertise and years to master, with many investors likely losing money chasing this trend.
- Investment strategies are shifting toward hard tech and physical challenges, including a fabless model in synthetic biology where companies rent early 2000s-built lab infrastructure, while software investors risk losses if they lack the knowledge to navigate hard tech without the traditional SaaS playbook.
- Venture capital returns are described as binary, with a generation of funds expected to die if they lack a top-performing partner, while medium companies may still yield 5 to 7 times returns and top cap table players may recover liquid preferences similar to the 2000-2004 cycle.
- Growth expectations are recalibrated, with 100% growth deemed difficult, 30% to 40% considered great, and 10% to 20% seen as a tricky middle ground for exits, while some slow-growth companies may grind out value or trade in large secondary markets.
- Widespread AI adoption across the broad economy is expected to take longer than currently thought due to regulatory barriers and potential punitive policies, even as the speaker predicts adoption will eventually exceed initial year-over-year overestimations.
- Market consolidation is anticipated where large cloud providers acquire foundation models, and many PE-backed companies are expected to face write-downs, potentially causing funds that took significant hits to lose appetite for new acquisitions.
- Secondary market activity is expected to grow as buyers often seek data rather than actual purchases, with some companies trading in large secondary trades or buying back their own stock to manage liquidity.
- The speaker predicts a movement where some companies become runaways unable to accept more capital, while others receive capital to double down into giant winners, with the best fund performers often originating from the "meh" category rather than the "rocket ship" category.
- Strategic plans for future operations include stopping active deal sourcing in favor of referrals, relying on networking as a primary skill, and acknowledging that management support will be limited due to the speaker's own constraints.
- Risks include the failure of companies with middling growth trajectories, the inability of legacy firms to adapt without proper data infrastructure, and the possibility that China's advancements in AI could render US bans on NVIDIA chips ineffective in slowing them down.