Interview, Fireside Chat
How I'm Investing My $230M Fund - AI Investor Tom Tunguz Breaks It Down
- The fund size is set at $230 million, a figure derived from Monte Carlo simulations designed to optimize portfolio construction for a solo general partner.
- The strategy targets 12 to 15 portfolio companies, deviating from traditional diversification norms to allow for significant concentration.
- Historical analysis indicates that 23 companies capture approximately 82–84% of diversification benefits, justifying a smaller, higher-conviction portfolio where the top three holdings may comprise 40–50% of the fund.
- Initial check sizes are estimated at $8 million to $12 million per company.
- While the fund lacks the capacity to lead $50 million Series A rounds common in AI and machine learning, it retains the flexibility to co-lead or participate in $15 million raises alongside other partners.
- The investment philosophy is thesis-driven, with the GP intending to spend six to twelve months researching specific sectors before deploying capital.
- A primary investment theme is "the decade of data," leveraging over a decade of experience and a pre-existing database for benchmarking data companies.
- Exit market analysis is a core component of the diligence process, rejecting projections based on hyper-growth multiples seen during quantitative easing eras (e.g., 40x) in favor of realistic forward multiples closer to 5.5x.
- The GP explicitly labels projecting a 20x forward multiple for a $100 million company growing at 70% as "irresponsible" and unrealistic given current market conditions where multiples are closer to 6x.
- Investment evaluation utilizes a "Fermi estimation" framework (conditional probabilities) inspired by the book Super Forecasters to enumerate and weigh success criteria, such as the ~60% base rate for raising Series A from seed and ~50% for Series B.
- The GP plans to deepen convictions in top holdings by continuously monitoring how conditional probabilities for key success factors (e.g., hiring PhD teams, market acquisition) evolve over a company's lifetime.