Venture Capital Reform II: The Hidden Risks of Traditional Investment Screening (Zenodo)

Summary

• Median fund performance comparison shows 2017-vintage funds with a median TVPI of 1.87x and net IRR of 14.77%, compared to 1.12x TVPI and 2.75% net IRR for the 2024 vintage.p.2
• SoftBank Vision Fund 2 experienced a significant mark-to-market shift from a cumulative gross loss of $23.8bn in March 2025 to a $21.8bn cumulative gross gain in March 2026.p.2
• Tiger Global's $12.7 billion PIP 15 megafund was reported to remain in a loss-making position at the end of 2024.p.3
• The HAS-CORE early-stage AI safety infrastructure project incorporates nine core modules to evaluate how structural technological risks differ from commercial readiness markers.p.4

Key insights

1. Common screening filters such as initial revenue and early customers confirm current transaction activity but fail to evaluate underlying unit economics, technical durability, and structural fragility.p.3
2. Fund-level median net IRR declined from 14.77% in the 2017 vintage to 2.75% in the 2024 vintage, though differing maturity stages prevent interpreting this as a final failure rate.p.2
3. Megafund marks are heavily driven by unrealized valuation swings from key holdings such as OpenAI rather than realized net cash distributions.p.2
4. A structural weakness in risk judgment arises when commercial maturity indicators are treated as substitutes for deep technological assessment.p.1
5. Deep technology projects face a structural dilemma where external validation requires capital and engineering resources that investors condition on existing validation.p.4
6. Investment screening must assess how capital deployment itself alters and mitigates a project's intrinsic risk profile rather than relying solely on present-state readiness snapshots.p.4