Miles Dieffenbach: Inside Carnegie Mellon’s $4BN Endowment & The Math Behind DPI, TVPI, Illiquidity

Miles Dieffenbach of CMU’s $4B endowment argues 90% of LPs shouldn’t be in venture, median net IRR is only 8%, and calls for GPs to IPO portfolio companies now.

• CMU allocates ~25% of its $4B endowment to venture globally, overweight peers by 5-10 points; private book has been self-funding since 2021 for the first time.
• Median venture IRR from 1998-2016 vintages is ~8% net; top-quartile DPI over 15 years is only 1.8x — below the NASDAQ 100 PME benchmark.
• $50-100M seed funds are structurally broken: to own 3-3.5% across 30 companies you need $3M+ checks, impossible at that fund size without subscale ownership or diversification.
• Multi-stage firms deploying $5-10M seed checks are destroying traditional seed fund economics by offering cheaper cost of capital.
• OpenAI burning $5-10B/year and could be a zero if capital markets freeze; unlike Google/Meta at IPO or SpaceX today, it does not control its own destiny.
• Google, Microsoft, Amazon, and Meta combined generate $600B in annual operating cash flow — Wiz approval would green-light a new wave of large acquisitions.
• A manager holding OpenC at a $13B valuation in 2023 — later revised — was the worst GP behavior Dieffenbach encountered; best firms like Sequoia/Excel mark at 20-30% discounts even on strong assets.
• Yale sold a strip of its venture portfolio at ~10% discount (Dieffenbach expected 20%); CalPERS bought it and saw ~$100M writeup from Circle alone within two months.

2025-08-04 · Watch on YouTube