The music is still playing. Investors are dancing. David Rubenstein is looking for the exit. The Carlyle Group co-founder recently voiced a sentiment that many in the private equity world have whispered for months. Fear of missing out is driving the current market cycle. It is no longer about fundamentals. It is about the terror of being left behind while the machines take over the tape.
The Rubenstein Warning
Capital is cheap for the few and expensive for the many. Rubenstein noted on Yahoo Finance that time will eventually reveal whether the current aggressive entry into AI-linked equities was a strategic masterstroke or a generational error. This is not merely a hedge fund manager being cautious. This is a structural critique of how capital is being misallocated in the pursuit of a productivity miracle that has yet to appear on a balance sheet. The premium paid for anything with an ‘AI’ suffix has reached levels that defy traditional discounted cash flow models. We are seeing valuations based on 2030 projections being traded at 2026 prices.
The Infrastructure Debt
Data centers are the new oil wells. They are also incredibly expensive to maintain. The cost of the latest Blackwell-class chips and their successors has forced a massive capital expenditure cycle among the hyperscalers. According to recent Reuters reporting on the semiconductor supply chain, the lead times for high-bandwidth memory have not improved despite increased production capacity. This creates a bottleneck. If the hardware cannot be deployed, the software cannot be monetized. Yet, the stock prices reflect an assumption of seamless deployment. The gap between physical reality and digital speculation is widening.
AI Infrastructure Spend vs Net Revenue Growth Q1 2026
The chart above illustrates the disconnect. Blue bars represent the percentage increase in capital expenditure for AI infrastructure. Red bars represent the actual net revenue growth attributed to AI services. The divergence is staggering. Companies are spending four to five times more on the capacity to provide AI than they are currently earning from it. This is the ‘mistake’ Rubenstein is referencing. It is a bet on a future that must be perfect to justify the present.
The Fed Factor and Liquidity Traps
Interest rates remain the ultimate arbiter of value. Following the Federal Reserve meeting on May 6, it is clear that the ‘higher for longer’ mantra is not a suggestion. It is a policy. With the federal funds rate hovering near 5 percent, the cost of carry for these massive AI bets is significant. Institutional investors are using leverage to chase these returns, betting that the AI-driven productivity gains will eventually lower inflation and force the Fed’s hand. It is a circular logic. They are buying the cure for the high rates they are currently paying.
| Metric | May 2025 Value | May 2026 Value | Change (%) |
|---|---|---|---|
| S&P 500 AI Sub-Index P/E | 32.4 | 44.1 | +36.1% |
| Average Data Center Utility Cost (kWh) | $0.07 | $0.11 | +57.1% |
| Enterprise AI Adoption Rate | 18% | 24% | +33.3% |
The table reveals the friction. Utility costs for data centers have spiked by over 50 percent in the last twelve months. This is a direct hit to margins. While enterprise adoption is growing, it is not keeping pace with the valuation of the companies providing the tools. The market is pricing in a 100 percent adoption rate while the reality is stuck in the low twenties. This is the definition of a crowded trade. When everyone is on one side of the boat, it only takes a small wave to capsize the vessel.
The Technical Mechanism of the Squeeze
Liquidity is drying up in the secondary markets. We are seeing a shift where large-cap tech is used as a source of funds to cover losses in other sectors. This creates a ‘volatility dampening’ effect that is artificial. When the selling starts, it will not be a slow decline. It will be an algorithmic cascade. The same AI models that are driving the buying frenzy are also programmed to hit the sell button at the same technical triggers. We are building a market where the speed of the crash will be dictated by the speed of the light through the fiber optic cables.
The next major data point arrives on June 12 with the release of the May Consumer Price Index. If energy costs associated with the tech boom continue to prop up the headline number, the Fed will have no choice but to maintain its restrictive stance. Watch the 10-year Treasury yield. If it crosses the 4.8 percent threshold, the AI FOMO trade will face its first true stress test of the year.
