Miles Dieffenbach - Managing Director, Carnegie Mellon University Endowment. Tracked across 46 mentions in podcasts and expert conversations analyzed by Sonic.

Miles Dieffenbach

Investment professional at Carnegie Mellon University's $4BN Endowment, specializing in venture capital and private markets.

Claims

As speaker

Mentioned in

Key positions and views

Large, multi-stage venture funds are structurally challenged and unlikely to generate the high returns of their smaller predecessor funds due to the mathematical difficulty of achieving necessary market cap outcomes at scale.

The vast majority (90%) of Limited Partners should not be investing in venture capital because the median returns do not adequately compensate for the high risk and illiquidity, and only top-decile managers consistently outperform public markets.

The fee structure for large, growth-stage venture funds is misaligned with the risk profile of their mature assets and should be reformed to mirror public equity funds, with approximately a 1% management fee and 10% carry.

The current AI boom is exhibiting classic signs of a technology bubble that, like historical parallels such as the internet, is likely to experience a significant pop or correction.

GP commitment is one of the most reliable quantitative, forward-looking indicators for predicting a venture fund's future returns, while actions like selling a stake in the management company are a 'massive red flag' for institutional LPs.

Timeline

2002-2004

Cites this period as having a more robust IPO market in terms of capital raised than the recent 2022-2024 period, highlighting a current liquidity crisis in venture capital.

2012 & 2014

References the IPOs of Facebook and Alibaba as the two largest in venture-backed history, setting a high watermark for exit valuations.

c. 2017

Identifies the launch of SoftBank's Vision Fund 1 as the catalyst that triggered a wave of fund size increases across the venture capital industry.

2021

Describes this year as a market peak characterized by extremely high software valuation multiples (20x median, 40x top-quartile) and a cohort of IPOs that are now trading significantly below their offering price.

2024

Observes that venture fundraising is on track to hit its lowest levels since 2017 in the US and 2016 in Europe, indicating a significant market contraction.

2026 (Predicted)

Predicts this will be a significant year for venture capital liquidity, citing pending large-scale M&A deals for companies like Wiz and Figma.

Podcast consensus on Dieffenbach

Points of consensus

▶Miles Dieffenbach consistently argues that the venture capital industry, particularly at the multi-stage growth level, faces significant structural challenges. He believes large funds will struggle to generate historical returns due to the mathematical difficulty of deploying massive capital effectively.Mar 2026

▶He maintains a strong, pro-LP (Limited Partner) stance, repeatedly asserting that LPs are not adequately compensated for the risk and illiquidity of the venture asset class and that most LPs should avoid it altogether.Mar 2026

▶Dieffenbach views markets through a cyclical lens, believing that the current AI boom will likely follow historical patterns of other technological shifts (railroads, internet) by experiencing a bubble that eventually pops.Mar 2026

▶He emphasizes the importance of manager selection and specific fund attributes, highlighting GP commitment as a key forward-looking indicator of success and praising firms like Index Ventures and Union Square Ventures for their discipline and performance.Mar 2026

Points of debate

▶Dieffenbach's assertion that 90% of LPs should not invest in venture capital is a stark counterpoint to the industry's continuous fundraising efforts and the widespread pursuit of venture allocation by institutional investors.Mar 2026

▶His proposal that large growth-stage venture funds should adopt lower, public-equity-style fees (1% management, 10% carry) directly challenges the standard '2 and 20' model that has dominated the venture industry for decades.Mar 2026

▶While the market has been extremely bullish on AI infrastructure companies, Dieffenbach presents a contrarian, cyclical view of NVIDIA, suggesting a plausible scenario of a 70% stock drawdown, which contrasts with prevailing market sentiment.

▶He expresses skepticism about the long-term independence of major AI companies like OpenAI and Anthropic, questioning their viability in a way that challenges the narrative of their massive venture capital funding rounds and high valuations.Mar 2026

Key themes

▶Venture Capital Fund Mechanics and PerformanceMar 2026

Dieffenbach provides a detailed critique of the venture capital fund model, focusing on the inverse relationship between fund size and returns. He argues that as funds scale into the billions, the mathematical requirement to generate massive market cap outcomes makes outsized returns exceedingly difficult, a reality he believes many LPs and GPs ignore.

Investors should be highly skeptical of 'fund size creep' and scrutinize the strategy changes required to deploy larger pools of capital, as historical performance from smaller funds is not a reliable indicator of future success at scale.

▶The Limited Partner (LP) Dilemma

A core theme is the precarious position of LPs in the venture ecosystem. Dieffenbach contends that LPs are not being paid for the risks they take, citing mediocre median returns and extreme illiquidity, leading to his provocative claim that 90% of them should not be in the asset class at all.

This perspective serves as a crucial reminder for allocators to rigorously assess whether they have the access and selection skill to reach the top-decile of managers, as returns below that threshold often fail to justify the risk compared to public markets.

▶Market Cycles and Macro HeadwindsMar 2026

Dieffenbach consistently analyzes the current environment through the lens of historical market cycles. He points to the sluggish IPO market, a venture fundraising winter, and the AI boom as phenomena that have historical parallels, suggesting a major correction or 'bubble pop' for AI is an anomaly if it *doesn't* happen.

Analysts should consider the cyclical nature of hardware and infrastructure spending when modeling long-term growth for companies like NVIDIA, as historical patterns suggest periods of negative growth are common even for market leaders.

▶Carnegie Mellon's Endowment Strategy

The claims provide a window into the sophisticated strategy of the Carnegie Mellon endowment, which has a higher-than-average allocation to venture capital. This strategy is predicated on disciplined manager selection, a focus on fund sizes between $80M and $1B, and a portfolio construction that has become self-funding through distributions.

The CMU endowment's approach demonstrates that a successful venture program requires not just high allocation but also strict underwriting criteria, including identifying red flags like GPs selling management company stakes and a preference for managers with significant personal capital commitment.

Source episodes

Suggested prompts

How does Miles Dieffenbach's framework for access vs. picking skill apply to emerging venture markets outside the US and Europe? &nearr;What are the potential second-order effects on the startup ecosystem if large growth funds adopt the 1% management and 10% carry fee structure Dieffenbach advocates for? &nearr;Given his cyclical view of NVIDIA and the prediction of an AI bubble pop, what specific indicators would validate or invalidate his thesis over the next 24 months? &nearr;How does Carnegie Mellon's higher-than-average allocation to venture capital square with Dieffenbach's assertion that 90% of LPs should avoid the asset class? &nearr;

Key concepts

Venture Capital Performance & Returns 1 ep Limited Partner (LP) Strategy 1 ep Market Cycles & Bubbles 1 ep Fund Management & Fees 1 ep Carnegie Mellon Endowment 1 ep AI Industry Analysis 1 ep IPO & M&A Markets 1 ep Specific Company Analysis (NVIDIA, OpenAI, Index) 1 ep

Notable quotes

“Do you think LPs are getting paid for the risks that they are taking in Vastian Venture? Absolutely not.”

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

“90% of LPs shouldn't be investing in venture.”

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

“But those growth funds that are really for scaled businesses that are mature assets, like you should be charging long-only public equity fees, which are one in 10.”

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

“More capital was raised through IPOs in the public markets from 2002 to 2004 than from 2022 to 2024, despite the venture asset class being ten times larger now.”

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

Report last updated: Jul 28, 2026

Create a free account to see Miles Dieffenbach's full intelligence report - every claim, the relationship network, and AI Q&A across all sources. No card needed.

Get started free

Back to Entities Intelligence Report