Elite university endowments boost FY2025 returns through concentrated AI and crypto bets

The fiscal year 2025 saw a remarkable divergence in the performance of elite university endowments, with the University of Michigan, MIT, and Stanford standing out by delivering returns between 14.3% and 15.5%, significantly outpacing their peers and standard benchmarks. While most top-tier university endowments clustered around returns of 11-12%, these three institutions exceeded expectations by 200-300 basis points, raising critical questions about the drivers behind their exceptional performance.

According to a detailed analysis by Markov Processes International (MPI), the outperformance was less about traditional manager selection and more about concentrated thematic investments, particularly in artificial intelligence (AI) and digital assets such as cryptocurrency. This revelation challenges the conventional narrative often presented in endowment reports, which tend to attribute success to diversified portfolios and skilled manager selection. Instead, MPI’s factor-based analysis shows that significant exposures to AI and crypto themes were pivotal in lifting returns well above benchmark levels in FY2025.

Market conditions in FY2025 were generally supportive but not exceptional enough on their own to explain these returns. For example, the S&P 500 returned 15.2%, international developed markets 17.7%, and emerging markets 15.3%. Meanwhile, private equity, venture capital, hedge funds, real estate, and bonds had mixed or modest returns, which under typical endowment allocations would have resulted in overall portfolio performance around 11-12%. The top-performing endowments’ traditional asset mix would actually forecast slightly underperforming a simple 70/30 equity-bond portfolio. However, Michigan’s reported 15.5%, MIT’s 14.8%, and Stanford’s 14.3% returns defy this expectation.

One of the more opaque but critical drivers behind these returns appears to be significant, albeit discreet, exposure to cryptocurrencies and blockchain infrastructure. The University of Michigan has notably invested in Andreessen Horowitz’s crypto fund since 2018, increasing commitments subsequently, while other institutions such as Harvard, Stanford, MIT, Dartmouth, Yale, and Brown are known limited partners in major crypto venture funds. Some universities have also acquired direct Bitcoin holdings or stakes in blockchain firms, Cornell’s equity ownership in Ava Labs, creators of the Avalanche blockchain, being a prime example of institutional ownership in crypto infrastructure rather than just fund investments.

In parallel, exposure to AI-related investments has also been a key factor. Michigan’s involvement as a limited partner in AI-focused venture funds like Alt Capital II and its rumored direct investment in OpenAI represents a shift towards targeted investments in transformative AI enterprise technologies rather than indirect or diversified venture exposure. Similarly, MPI’s analysis shows notable AI-related exposures at Princeton, UPenn, Yale, MIT, Stanford, and especially Michigan, often registered through venture or private equity vehicles.

It is essential to note that much of the reported financial uplift from AI might not be entirely liquid. Many reported gains may reflect paper valuations, mark-to-model valuations on late-stage venture investments, rather than realised cash returns. For instance, Sequoia’s flagship venture fund was marked up significantly in FY24 despite limited exits, indicating that a substantial portion of reported AI-related profits in endowments could be unrealized or illiquid.

Quantifying these thematic exposures, MPI’s attribution analysis suggests that at Michigan, digital assets contributed roughly 2.9% to the total return, while AI-focused investments added about 2.8%, alongside approximately 1% of alpha from manager-specific skill. These concentrated theme exposures combined explain the entire gap between the expected 11-12% returns and Michigan’s actual 15.5%. Similar patterns of thematic concentration are visible at MIT and Stanford, where returns have been strong at 14.8% and 14.3%, respectively, as independently reported in their financial disclosures.

This thematic concentration, while lucrative, raises significant governance and risk management considerations. Boards and stakeholders require transparent attribution of returns beyond broad claims of “manager selection,” especially when returns hinge on volatile sectors such as crypto and AI. The concentrated nature of these themes impacts liquidity planning and risk exposures, as gains largely tied to volatile and sometimes illiquid assets may not contribute to sustainable or spendable returns in the short term. Within this context, Cornell’s quarterly performance disclosures stand out as a model for transparency, allowing more definitive attribution analysis than annual reports typically permit.

Strategically, these FY2025 results underscore a shift in endowment investing philosophy. Large institutional portfolios are increasingly generating meaningful outperformance not through broad diversification but via early, concentrated bets on transformative technologies. The leadership shown by institutions that built significant AI and crypto positions as early as 2018, well before these sectors gained broad acceptance, illustrates the importance of foresight and thematic conviction. For institutional investors grappling with generally compressed returns from traditional asset classes, the lesson is clear: thematic concentration could be the primary engine of future alpha, although it demands rigorous governance, clear risk understanding, and transparency.

As these transformative investment themes evolve, the central question for universities and similar institutional investors will be how to balance risk and reward and whether current AI and crypto valuations will convert into realisable cash gains. The next cycle of endowment outperformance may hinge on identifying new themes beyond these current frontrunners, but the FY2025 experience sets a precedent for concentrated thematic investing as a pathway for superior returns.

In conclusion, the exceptional fiscal year 2025 performance from Michigan, MIT, and Stanford’s endowments was less a stroke of luck or broad strategy execution than a targeted embrace of AI and digital assets. While CIO statements often highlight discipline and diversification, the underlying data reveals a purposeful thematic concentration that drove their outperformance. For other institutional investors assessing their strategies, the path forward involves a critical decision: to commit to concentrated thematic bets at the risk of increased volatility or maintain diversified portfolios with more modest expected returns. Whichever path they choose, transparency and governance must remain paramount in understanding and managing the implications of that choice.