How Hedge Funds Are Ramping Up AI: Strategies, Risks & Key Infrastructure Bets

Operational Integration: Leveraging AI Internally

Leading hedge funds have rapidly expanded their internal use of AI tools to enhance efficiency and analytical depth. For instance, Citadel’s CTO Umesh Subramanian has described the firm’s data consumption as measured in petabytes, and AI has become essential to parsing and extracting actionable insights. Balyasny has deployed an internal chatbot, BAMChatGPT, now in use by ~80% of staff, alongside hiring ex-CIA AI talent [1]. Two Sigma, by 2024, had used generative AI in workflows for over five years, and Bridgewater in that same period launched a fully machine learning–run fund (~US$2 billion) aimed at generating alpha uncorrelated with human-driven decisions [1]. This trend reflects a broader shift: AI is increasingly applied not just to augment research and data processing but to power investment strategy directly.

Dedicated AI-Focused Investment Strategies

Some hedge funds have gone further, creating new strategies or sub-funds explicitly focused on AI. Point72’s Turion, launched in October 2024 under portfolio manager Eric Sanchez, is one example; it has outperformed Point72’s main fund in 2025 [1]. Tiger Cubs prioritize AI exposure through public and private equity, notably in semiconductor and infrastructure companies. Maverick’s private fund, Maverick Silicon, targets chip-ecosystem investments [1]. Such dedicated vehicles help investors isolate exposure to what many believe is a multidecade secular trend.

Where Funds Are Placing Their Capital

Portfolio tilts are clearly toward AI infrastructure. Major holdings among these funds include Nvidia, AMD, SK Hynix, Alphabet, Microsoft, Meta—names dominating chipmakers, cloud providers, and core AI model developers [1]. Meanwhile, those in the Tiger Cub sphere are channeling significant venture capital into startups building out AI infrastructure, including compute, semiconductors, data centers, and ancillary enabling technologies. This reflects a dual strategy: capturing upside in public markets while gaining earliest access via private investments.

Risks, Constraints, and Strategic Implications

• Hype vs. performance: Some fund leaders remain skeptical that AI alone can produce alpha. Citadel’s Ken Griffin, for example, argues AI hasn’t yet shown it can reliably beat markets without support [1].
• Bubble risk: Bridgewater warns of a dangerous phase where Big Tech is relying heavily on external capital for AI growth; UBS estimated AI infrastructure investments rose from ~US$15B in 2024 to US$125B by November 2025 [2].
• Talent competition: Attracting top machine learning and AI engineering talent has become a critical competitive edge, with hedge funds often paying at levels rivaling or exceeding those in tech companies [1].
• Governance and human oversight: Even highly automated funds retain human control over strategic decisions and risk. Firms like Man Group’s Numeric use AI only under human supervision; output traceability and oversight remain vital [1].

Strategic Implications for Stakeholders

For institutional investors and LPs, understanding which hedge funds have truly embedded AI in strategy versus those engaging in marketing is crucial. Those with dedicated sub-strategies, such as Turion or funds like Maverick Silicon, may offer differentiated exposure. For regulators, the ballooning capital flow into AI infrastructure and external funding models raises questions about systemic risk and valuation sustainability. For talent markets, this boom promises opportunities but also raises retention challenges. Cross-industry partnerships may further accelerate innovation but also intensify spillover and competitive risks.

Open Questions Going Forward

• Can AI enhance risk-adjusted returns enough to consistently outperform in both bull and bear cycles without overfitting or overreliance on backtests?
• What are the governance, bias, and transparency risks when models are increasingly black boxes or when many funds use similar datasets?
• How sustainable are valuations in AI hardware and infrastructure, and where might overcapacity or overinvestment lead to corrections?
• What regulatory frameworks or standards should emerge around AI usage in finance—around data privacy, model explainability, liquidity, and systemic spillovers?