Interview
Elon Musk vs Sam Altman | The Implosion of Thinking Machines | Can VC Survive Public Pricing?
- Public market multiples are projected to diverge significantly, devaluing low-growth companies while assigning absurdly high valuations to high-growth entities, with venture returns reinforcing a trend where "hot" companies succeed and "trailing edge" firms fail.
- The market cycle is anticipated to consist of 80% "crap" periods characterized by low liquidity and prices, punctuated by windows of high liquidity, with founders growing at 50% to 70% expected to struggle for capital while those growing at 75% to 100% remain resilient.
- Companies transitioning from high growth to slower rates will face a "long, hard haul" as valuation metrics shift from forward revenue multiples to free cash flow anchors, potentially causing the venture model to fail if the market shifts to an "EPS world" requiring proven earnings.
- Mid-stage SaaS companies growing at 50% to 70% face a "rounding error zero" probability of exploding into massive successes, though founders reaching $200 million in revenue via a "grind" strategy may still sell for five times revenue to generate billion-dollar outcomes.
- Investors are predicted to write off half a trillion dollars in failed guesses to capture a single projected $4 trillion winner, while the venture capital "MO" will likely exclude "grind" growth strategies, causing capital to flee slower-growth firms.
- A $100 billion revenue SaaS company is viewed as an entrepreneurial achievement that may be unfinanceable due to public market return constraints, whereas late-stage investments in entities like OpenAI and Anthropic will offer negligible information rights, effectively making funds function as public market investors.
- Core investors in Thinking Machines may demand redemptions and wind down the firm if team members depart, while investors might accept a 20 cent haircut on failed investments to recycle capital quickly within the current fund cycle.
- AI researchers are expected to leave companies unable to offer chosen intellectual challenges regardless of compensation, with the current political environment expected to vindicate non-US sovereign-based AI plays and the potential for a "transformer paper" equivalent by 2026.
- Non-technical visionaries may struggle to recruit S-tier AI talent, and non-LLM AI ventures are expected to have lower success probabilities than refining existing models, despite offering more interesting upside potential to researchers.
- The Sam Altman vs. Elon Musk litigation is forecast to last two to three years, with Musk potentially claiming $70 billion to $130 billion in shares and OpenAI facing 10% to 20% dilution if a verdict against it prevails or forces a settlement to raise capital.
- OpenAI is expected to introduce ads soon to monetize the free tier, as consumer conversion to paid tiers is predicted to remain under 5% or 10%, with the ads business potentially generating a billion dollars quickly or eventually reaching $25 billion to $100 billion.
- Google and Facebook are expected to lose search and discovery share to OpenAI as users prefer LLMs for product finding, prompting advertisers to shift spend if returns are positive, with the "answer engine optimization" sector becoming a vitally important business.
- Adobe is expected to successfully integrate SEMrush rather than acquiring smaller startups, while ClickHouse is projected to sustain 300% to 400% growth for two to three years, supporting a $30 billion to $40 billion outcome in the OLAP database market.
- Replit is expected to improve product quality by 50 to 100 times and reach $700 million to $900 million in revenue by year-end, with its evolution shifting from code-writing to "describing at the describing level."
- Sequoia and other late-stage funds are expected to continue piling money into pre-IPO winners to capture upside in a new ultra-late stage space, though market forces will eventually push private valuations to a "rational equilibrium" where reduced risk correlates with reduced returns.