How AI Is Rewriting the Power Law of Venture Capital

Market Dynamics and Power Law Extremes

• Only 20 out of 3,000 US venture capital firms achieved consistent 3x net returns over the last two decades.
• The power law in technology investing is currently more extreme than in the previous 10 to 20 years.
• Three frontier model companies (SpaceX, OpenAI, Anthropic) represent a combined potential enterprise value of $3.5 to $5 trillion.
• AI has reached $100 billion in aggregate industry revenue in four years, compared to 15 years for SaaS to reach the same milestone.
• AI is uniquely impacting all facets of the GDP, including transportation, labor, services, capital, and coordination.
• The total addressable market (TAM) for AI in sectors like healthcare is estimated to be 10x larger than traditional SaaS or healthcare IT markets due to the capture of labor value.

Venture Capital Structure and "Death of the Middle"

• Venture-like returns are now possible in the late stage if a single category-defining company represents 5% to 10% of a fund.
• The "middle" of the venture ecosystem is contracting, favoring either highly specialized early-stage firms or massive full-lifecycle platforms.
• Late-stage success is increasingly dependent on early-stage franchise access, which provides deal flow, information, and pro rata participation rights.
• The average venture return over the last 10 years is 1x to 2x net, which is outperformed by private equity returns without the same liquidity risk.
• Early-stage investment loss rates in top funds are approximately 60%, whereas late-stage loss rates range between 10% and 20%.
• Large venture firms are increasingly co-investing or leading pre-seed/seed rounds to maintain "ball control" over category winners, despite traditional preferences for later entry.

Valuation Challenges and LP Incentives

• Evaluating traction is difficult due to rapid scaling (e.g., zero to $5M ARR in a month) and inflated valuation multiples based on non-renewal metrics.
• There is a structural misalignment between GPs and LPs: GPs are fired for missing category winners (error of omission), while LPs are rarely fired for missing them or investing in safe, low-growth assets like IBM.
• LP portfolio construction often suffers from over-diversification; holding 50 to 70 venture funds typically yields average returns rather than the top-tier performance needed to justify illiquidity.
• Private equity valuations for legacy software assets have contracted significantly; 21–22 software deals valued at 25–32x EBITDA are now worth approximately half that.
• Many PE-backed software companies face distress because they cannot rapidly transform into AI-native workflows while servicing existing debt loads.

Liquidity and Capital Allocation Strategies

• Private equity exit values are increasingly being surpassed by venture M&A deals (e.g., Cursor's acquisition by SpaceX at $300M ARR vs. PE deals of ~$500M enterprise value).
• Top-tier VC funds can achieve fund-returning liquidity earlier than private equity through M&A exits of category leaders.
• LPs with long horizons (e.g., endowments, family offices) often prefer allowing winners to compound rather than taking early liquidity to avoid tax events or reinvestment friction.
• CalPERS has aggressively shifted allocation from 9% to 43% in venture and growth equity to correct past underperformance in backend tech.
• Public markets now value AI-native growth acceleration significantly higher; 1% growth in public markets is viewed as equivalent to 3% EBITDA growth.

Future Outlook: The $10 Trillion Market Cap

• Future trillion-dollar valuations will likely emerge from new product categories rather than iterating on current chatbot interfaces.
• Key growth vectors identified for the next decade include robotics, AI autonomy (e.g., robo-taxis), and deep healthcare innovation (drug discovery, care delivery).
• Consumer AI applications will evolve from skeuomorphic chat interfaces to "native" agents that perform work on behalf of users.
• The primary bottleneck for AI growth is not demand but supply-side constraints, specifically energy grid capacity, permitting speed, and data center density.
• New investment vehicles are targeting the "left side" of the AI stack (energy generation, transmission, and data center infrastructure) to unlock 100x value creation opportunities.
• Forward-looking statements suggest the next SpaceX-level company will likely arise from the convergence of AI with physical world domains like defense, manufacturing, or energy.