Greg Jensen, co-chief investment officer at Bridgewater Associates, the world’s largest hedge fund, has put forth a straightforward thesis: when a few companies gain control over critical infrastructure to a degree that could destabilize society, regulators inevitably intervene. He believes artificial intelligence compute is the next frontier for such oversight.
Jensen is advocating for regulatory frameworks similar to those applied to systemically important financial institutions (SIFIs) to be implemented for any company that commands more than 5% of global or U.S. AI compute resources.
The SIFI framework emerged in the wake of the 2008 financial crisis, a period when regulators recognized that certain institutions had become so large and interconnected that their collapse posed a systemic risk. This designation subjects these entities to stringent capital requirements, stress testing, and regulatory scrutiny far beyond what typical firms encounter.
Jensen’s remarks, which surfaced around September 17-18, 2026, following an interview with The Information, are strategically timed. He cited projections indicating that OpenAI and Anthropic could collectively control between 35% and 50% of the world’s AI compute capacity within approximately two years. This level of concentration, in his view, mirrors the conditions that led regulators to regret their previously hands-off approach to financial conglomerates before 2008.
Jensen draws an analogy to February 2020, the period just before COVID-19 escalated into a global emergency. He appears to believe that the discourse surrounding AI governance is at a similar critical juncture, with the window for establishing proactive structures closing. He warns that reactive regulation, implemented after a significant incident, will inevitably be more costly and less effective.
A Complex Proposition
Jensen’s call for oversight is complicated by Bridgewater’s own significant involvement in the AI sector. In 2023, the firm launched AIA Labs, a dedicated unit for AI investment and development. This division now employs over 80 staff and manages approximately $4.5 billion in assets dedicated to AI-driven strategies.
Bridgewater has emphasized that AIA Labs incorporates deliberate human oversight into its operational processes. This design choice aligns with Jensen’s broader argument: AI systems wielding substantial influence over critical decisions should not operate without robust accountability mechanisms.
Translating Jensen’s proposed 5% threshold into concrete policy presents considerable challenges. The SIFI designation for banks is supported by a well-established global regulatory architecture, with the Financial Stability Board coordinating international standards and national regulators implementing them domestically. AI compute currently lacks a comparable institutional structure.
Furthermore, defining what constitutes “compute” for regulatory purposes is a complex undertaking. Questions arise as to whether it refers to raw chip capacity, data center square footage, model training runs, or inference throughput. Each definition would yield a different understanding of who controls what and create distinct incentives for companies seeking to remain below a regulatory cap.
The companies most directly implicated by Jensen’s proposal, OpenAI and Anthropic, are both privately held and receive substantial backing from some of the world’s leading technology and venture capital firms. The imposition of regulatory concentration limits could significantly impact their current growth trajectories, particularly at a time when both are making substantial investments in infrastructure.
