Interview, Fireside Chat, Podcast
Michael Eisenberg: How China Could Overtake the US in the AI Race | E1167
- Foundation models are described as the "fastest appreciating asset in history," yet their depreciation velocity is equally high, leading to concerns about business models where core assets (talent/models) can "walk out at night."
- Lena Connell's aggressive stance on M&A is characterized as a threat to American capitalism, with a prediction that she must be removed from her position regardless of the outcome of the US presidential election.
- Harry Stebbings disputes the notion that the IPO window is closed, citing Reddit's successful public market entry and arguing that companies with significant revenue (or even lower, like Lemonade's $60M) should go public early to build value in the public market.
- Michael Moritz identifies three specific categories of investors likely to lose money in the current AI cycle:
- LPs with false portfolio diversification due to heavy exposure to the same trends or companies across multiple funds.
- "Logo chasers" who invest in hot names (e.g., OpenAI, Character AI) solely to bolster their marketing credentials.
- Investors attempting to replicate US-scale foundation models in other geographies without the necessary academic or data infrastructure.
- The prevailing view among VCs is that only one or two AI companies will generate outsized returns, while the majority will result in significant losses for LPs and GPs.
- A structural shift in value creation is occurring from the model itself to the "collection of specific people" (teams) capable of integrating models; Perplexity's valuation is attributed to its team rather than the underlying AI model.
- Future AI infrastructure will likely consolidate around two layers: "head-on" services (e.g., Anthropic) and underlying API layers accessed via hyperscaler stacks (Google, Amazon), with consumer interaction shifting to personalized AI agents rather than direct website visits.
- Vertical SaaS is predicted to face existential threats as AI enables companies to build custom software internally or via consultants (e.g., McKinsey, Accenture), reducing the need for third-party off-the-shelf vertical software.
- Harry Stebbings contrasts the "SaaS bubble" with the emerging "hard tech" era, noting that synthetic biology and chemistry are now "fabless" due to accessible lab infrastructure, though this requires a shift in VC investment playbooks.
- Michael Moritz expresses concern that European regulatory environments (specifically data access and AI policy) will stifle innovation and competitiveness compared to the US and Israel, advising founders in Europe to "get out."
- Geopolitical strategy is redefining AI as a critical national security asset; Moritz argues against "closed systems" as a viable long-term strategy for the US, noting that China can still access semi-degraded chips or open-source designs.
- A shift in military strategy involves using AI to "close the kill chain," with examples of European companies generating hundreds of millions in revenue for defense applications.
- Moritz critiques the current excitement around defense stocks among MBA students as intellectually shallow, contrasting it with the deep expertise required to navigate government defense sales and regulations.
- Private Equity liquidity is constrained by high interest rates and cost of capital, leading to write-downs (e.g., Pluralsight) and a reduction in the number of successful liquidity events.
- Moritz advocates for a "binary" investment approach: taking full positions in companies with high conviction rather than diluting capital across many bets, contrasting this with the public market strategy of averaging in and out.
- In the context of LP returns, Moritz emphasizes that DPI (Distributions to Paid-In) is the primary metric, though TVPI (Total Value to Paid-In) remains relevant if underlying portfolio companies possess deep, sustainable competitive advantages.
- Michael Moritz identifies his own primary areas for improvement as managing board meeting efficiency (aiming to reduce them to 45 minutes) and tempering his direct communication style to avoid rattling first-time founders.
- Moritz expresses a preference for "net new naive" founders, citing the success of Lemonade's founders who lacked insurance industry experience and thus broke traditional rules.
- Moritz has changed his view on the younger generation, now seeing them as a defining, responsible cohort forced to stand up for their values, contrasting this with previous concerns about TikTok and Instagram addiction.
- The biggest misconception about the Israeli startup ecosystem is that Israelis inherently know how to scale; Moritz notes that foreign investors must actively help them think in terms of global scale.
- Moritz views his own "biggest sin" of the zero-interest rate era as not having sold enough of his portfolio during the bubble, warning that challenges are essential for the satisfaction of success.
- Moritz plans to transition away from proactive deal sourcing in the next decade, relying instead on his 30-year referral network to identify two or three high-conviction deals annually.