The Twenty Minute VCMiles Dieffenbach: Inside Carnegie Mellon’s $4BN Endowment & The Math Behind DPI, TVPI, Illiquidity
CHAPTERS
- 0:00 – 1:01
Cold open: Why VCs should take companies public (and why access matters)
Miles opens with a blunt message to venture capitalists: the market needs IPOs again. He frames venture as an access-and-selection business where only top-decile managers reliably beat public market equivalents (PME).
- •Direct call for venture-backed companies to go public to restore liquidity
- •Venture GP economics as unusually high-margin business models
- •Top-decile access is the key threshold for outperformance vs PME
- •Even top-quartile managers may fail to beat the public-market alternative
- 1:01 – 4:28
Cancer at 26: perspective, discipline, and redefining adversity
Miles recounts being diagnosed with lymphoma at 26 and how it reshaped his mindset. He describes choosing action and discipline during treatment and how the experience permanently changed his perspective on stress and resilience.
- •Shock of diagnosis and the decision to “attack” the situation
- •Using routine and fitness as a coping and control mechanism
- •The emotional moment of being told he was cancer-free
- •How mortality shifts perspective: fewer things feel existential afterward
- 4:28 – 5:49
Inside CMU’s $4B endowment: top-down portfolio construction
Miles lays out Carnegie Mellon’s endowment structure and how it’s managed. He explains the high-level split between equity and fixed income, and then the 50/50 split between privates and hedge funds/liquids.
- •CMU manages ~$4B with an 85% equity / 15% fixed income target
- •50% of the portfolio in private assets; 50% in hedge funds + public markets
- •Privates include VC, PE, real estate, natural resources, and private credit
- •“Best athlete” approach: allocate to the best risk-adjusted opportunities across private sub-asset classes
- 5:49 – 7:15
Where CMU is overweight: venture exposure, liquidity, and self-funding dynamics
The conversation moves into how CMU’s private book has behaved through recent liquidity conditions. Miles highlights that buyout has driven distributions while venture has been a drag, and he explains CMU’s relatively high venture allocation versus peers.
- •Private equity book has been self-funding for ~3 years (distributions covering calls)
- •Buyout drove most distributions; venture has detracted due to slowed exits and markdowns
- •Venture allocation is just under ~25% of the total endowment
- •CMU is overweight venture by ~5–10 points versus similar-sized endowments
- 7:15 – 8:28
Risk-first allocation thinking: why venture must be compensated differently
Miles explains how CMU evaluates opportunity cost across private asset classes by starting with risk. He contrasts stable assets like real estate with the extreme uncertainty of early-stage venture and argues that return expectations must reflect that difference.
- •Opportunity cost framed as return per unit of risk
- •Real estate example: stable cash flows and replacement cost support lower risk
- •Early-stage venture described as among the riskiest asset classes (idea-stage investing)
- •Core principle: LPs must be paid for the risk they’re underwriting
- 8:28 – 10:04
Are LPs getting screwed in venture? The DPI/TVPI/PME reality check
Miles argues LPs are not being adequately compensated for venture risk based on long-run fund data. He compares venture outcomes to the Nasdaq-100 PME and emphasizes that DPI (cash returned) is the metric that reveals the true experience for LPs.
- •Median net IRR around ~8% over long horizons (per cited dataset)
- •Top-quartile TVPI framed as ~2.5x, but top-quartile DPI closer to ~1.8x
- •PME comparison for venture: QQQ/Nasdaq-100 as the benchmark
- •Core claim: below top-decile managers, venture may not beat public market alternatives
- 10:04 – 15:49
“90% of LPs shouldn’t be in venture”: access vs picking and the seed-fund trap
Miles and Harry debate who should allocate to venture and why many new entrants are structurally disadvantaged. They dig into seed fund sizing math, multi-stage firms moving earlier, and whether early-stage advantage is access-driven or picking-driven.
- •Key allocator question: do you truly have access to top-decile managers?
- •First-time/micro funds are hyper-competitive and operationally risky to underwrite
- •Seed fund sizing math: ownership vs diversification trade-offs get ugly at $50–$100M
- •Access vs picking framing: multi-stage skews toward access; small early-stage skews toward picking
- 15:49 – 29:07
The five pillars of venture—and how CMU diligences people and partnerships
Miles introduces a venture framework—sourcing, picking, winning, helping, and selling—calling selling an underdeveloped muscle. He then explains how CMU references managers, hunts for interpersonal/partnership risk, and avoids common diligence traps like misleading attribution.
- •Five pillars: sourcing, picking, winning, helping, selling (selling becomes crucial in illiquid markets)
- •Preference for cash distributions over stock to avoid timing/pricing slippage
- •Deep reference process (targeting ~20 calls, prioritizing off-sheet references)
- •Partnership failure drivers: incentives and mismatched work ethic; CMU focuses on “partnership risk”
- 29:07 – 35:30
Fundraising mechanics and GP alignment: closes, LP churn, and selling the management company
The discussion turns to fundraising realities: LP churn, concentration risk, and how managers should structure raises. Miles is explicit that selling pieces of the management company is a major red flag because it dilutes the incentive engine—carry.
- •LP churn is rising; diversified LP bases reduce fragility
- •Concentration guidance: >30% from one LP is risky (10% ideal if possible)
- •Fundraising advice: minimize time fundraising; set crisp close timelines
- •Selling part of the management company is viewed as a major misalignment and institutional red flag
- 35:30 – 46:49
The $140B multi-stage math: why mega-funds struggle to generate alpha
Miles walks through a simple but sobering underwriting method: dollar-weighted entry ownership across early/growth/opportunity vehicles. He argues that when average ownership is low, required exit market cap to hit target returns becomes implausibly large—creating structural return compression.
- •Pari passu commitments force most LP dollars into lower-ownership growth/opportunity funds
- •Example underwriting: blended ~5% entry ownership on a ~$7B platform implies ~$140B of enterprise value exposure
- •To target ~4x net may require ~6x gross, implying ~hundreds of billions in needed exit value
- •Debate on future outcome sizes: Miles cites rarity of $50B+ and $100B+ venture IPOs historically
- 46:49 – 57:11
Fees, scaling, and the GP–LP bond: why growth should price like public long-only
Miles argues that as venture platforms become late-stage, they begin to resemble long-only public equity investing—but still charge venture-style fees. He predicts fee pressure over time (similar to hedge funds post-GFC) and advocates differentiated fees by strategy maturity.
- •At scale, alignment can break: fee streams become enormous even without commensurate alpha
- •Late-stage/growth investing often looks “passive” relative to early-stage sourcing effort
- •Proposed fee split: early-stage can justify 2.5/20 (or 2.5/30 if exceptional), growth should be closer to 1/10
- •Analogy to hedge fund fee compression after performance and competition cycles changed
- 57:11 – 1:01:03
Liquidity drought, IPO markets, and why ‘closed’ is mostly a pricing issue
Miles diagnoses the venture fundraising slump as primarily a liquidity problem, not merely sentiment. He argues IPO markets aren’t truly closed—companies are simply unwilling to accept market-clearing prices—and contrasts private valuation expectations with public comps like Microsoft.
- •US/Europe fundraising headed toward lowest levels since ~2016–2017 (per their discussion)
- •Liquidity post-2021 has been unusually weak relative to the size of the asset class
- •IPO market ‘closed’ is framed as a price problem, not a binary availability problem
- •Public comp discipline: slower-growth, low-margin SaaS struggles versus mega-cap quality alternatives
- 1:01:03 – 1:11:02
Secondaries and fat tails: Yale/Harvard sales, Circle’s surprise, and timing risk
Miles explains the hidden danger of selling old fund positions: right-tail outcomes can show up very late. Using Circle as an example, he describes how a small, discounted position in a 13-year-old fund can suddenly create multiple extra turns of performance, illustrating the cost of liquidity-driven selling.
- •Harvard’s $1B sale contextualized as small relative to a ~$50B endowment
- •Circle example: position held at a discount in a tail fund later becomes a major winner post-S-1/IPO
- •Lesson: fat-tail outcomes can materialize in years 8–13+, making strip sales perilous
- •View that reported ~10% discount on a Yale secondary sale seemed very attractive for the buyer
- 1:11:02 – 1:20:32
Cash-rich giants vs liquidity-starved venture—and the AI risk stack (OpenAI, NVIDIA)
Miles and Harry explore the paradox of huge public-company cash flows alongside venture’s liquidity drought. They discuss how regulation/timing risk shapes M&A behavior in AI, why OpenAI’s capital intensity creates existential financing risk, and why NVIDIA’s cyclicality matters even in a powerful trend.
- •If mega-deals clear (e.g., regulatory approval), big tech may re-accelerate strategic acquisitions
- •AI M&A constrained by review timelines and fast-changing tech; “acqui-hire + IP licensing” becomes a workaround
- •OpenAI risk framing: massive burn and large pref stack could be vulnerable if capital markets tighten
- •NVIDIA framed as strong today but still cyclical; peak earnings + peak multiple can create large drawdowns
- 1:20:32 – 1:30:35
Founder-friendly backlash + quick-fire: GP lies, red flags, and the ping-pong diligence story
Miles challenges the industry’s obsession with “founder-friendly,” arguing hard coaching and tough conversations can be healthier for outcomes. The episode closes with a quick-fire covering fee practices, fundraising misrepresentations, GP commitment, and memorable diligence moments—including a ping-pong match that preceded a check.
- •Founder-friendly is over-weighted; constructive pressure and “hard coaching” can be beneficial
- •Common fundraising lie: “we’ll never raise bigger than $300–$400M”
- •GP commitment treated as a meaningful forward-looking alignment signal (context matters more than the nominal amount)
- •Anecdotes: ping-pong match with Long Journey; shock at a manager marking OpenSea at $13B in 2023
