The AI Repricing: Hedge Funds Double Down and Goldman Sachs Maps the $3 Trillion Token Economy

The scale of the repositioning is significant. Goldman Sachs analyzed the holdings of 1,059 hedge funds managing $4.6 trillion in gross equity positions and found that funds lifted their net tilt toward the Information Technology sector . Major players, including Point72 Asset Management, Bridgewater Associates, and D. E. Shaw Group, have been actively rebuilding exposure to compute infrastructure and AI-enabling semiconductor names after a brief period of profit-taking .

This is not just a tactical trade. AI-driven strategies have crossed a clear performance threshold. According to BarclayHedge data, hedge funds that systematically integrate machine learning across their investment process have outperformed traditional systematic strategies by 3 to 4 percentage points annually since 2023, and that gap is widening . What was once an experimental edge is now described by analysts as a "structural necessity" for competitive returns .

The capital flowing through the sector is immense. Morgan Stanley Research estimates that nearly $3 trillion in AI-related infrastructure investment will move through the global economy by 2028, with more than 80% of that spending still ahead . Morgan Stanley characterizes this build-out as an "industrial" shift rather than speculative tech spending, with adoption progressing from pilots to tangible productivity solutions .

The 2030 Token Economy: 120 Quadrillion Tokens Per Month

While capital rushes into the hardware that runs AI, Goldman Sachs Research is providing a quantitative map of what the software layer will consume. In a May 2026 report, senior equity analyst Jim Schneider projected that agentic AI will lift global token consumption to approximately 120 quadrillion tokens per month by 2030, up from roughly 5 quadrillion per month in 2026 .

The growth is split across two major fronts:

• Consumer-grade agents are forecast to consume approximately 60 quadrillion tokens per month by 2030, a 12-fold increase. These agents will operate across online shopping, travel booking, email, and personal productivity tools .
• Enterprise-grade agents are expected to consume roughly 56 quadrillion tokens per month, a 24-fold increase over the same period .

The underlying engine of this demand is a projected rise in total AI queries. Goldman Sachs anticipates daily AI queries will climb from about 5 billion in 2025 to 23 billion by 2030, with up to 30% of those queries—roughly 6.9 billion per day—being handled by non-human agents operating autonomously .

The 2040 Horizon and the Cost Trap

Goldman Sachs views the 2030 numbers as only a waypoint. The bank's longer-term analysis suggests that enterprise-grade agents will be the largest multiplier in the AI economy, potentially lifting token consumption 55-fold by 2040 if enterprise adoption hits peak velocity .

However, the report is not universally bullish. Goldman Sachs explicitly cautions that data quality issues could undercut the expected payoff from agentic AI . There is also a looming cost trap: even as the per-token price of AI inference continues to decline, the sheer volume of tokens consumed by autonomous agents operating around the clock could cause aggregate AI costs to rise sharply for businesses .

This duality—massive potential paired with significant execution risk—mirrors the outlook from other major institutions. In its 2026 outlook for AI markets, Morgan Stanley acknowledged the transformative potential of AI while warning that "signs of excess are appearing" and that the market may be "ripe for a period of creative destruction" . For hedge funds, this environment creates the volatility and dispersion that active managers rely on to generate alpha .