Vanguard Quietly Warning: The AI Builder Bet Is Already Priced In
For the past two years, the investing narrative has been deceptively simple: buy the AI builders — chipmakers, hyperscalers, data center operators — and let the tidal wave of capital expenditure carry you higher. Nvidia, Meta, Microsoft, Alphabet: the names have practically become household investment gospel. But Vanguard’s global chief economist Joe Davis is now sounding a measured alarm that long-term investors would do well to hear: the easy money in AI infrastructure may already be behind us.
In a June 2026 report published through Project Syndicate, Davis drew on a pattern that repeats itself with remarkable consistency across transformative technology cycles. “The companies building transformative technologies rarely capture the greatest long-term value,” he wrote. “Instead, those benefits accrue to the users.” His historical examples are illuminating. Electricity didn’t enrich the power utilities — it enriched the manufacturers who ran assembly lines around the clock. The automobile didn’t primarily reward the automakers — it rewarded suburban developers and retailers who reshaped commerce around the car. Davis argues AI is likely to reproduce this exact pattern. Spending by the hyperscalers will continue for another year or two, he concedes, but the market has already priced in much of that growth. The next phase of value creation belongs to the companies quietly deploying AI to cut costs, personalize services, and automate workflows — not to those building the infrastructure.
With that thesis in mind, Davis identified three asset categories with compelling risk-return profiles over the next five to ten years: value-oriented U.S. stocks, non-U.S. developed markets, and high-quality fixed income. These aren’t exciting picks in the way Nvidia has been exciting. But that’s precisely the point. Healthcare providers automating claims processing, financial services firms offering AI-driven personalized advice, business services companies reducing headcount through intelligent software — these are the kinds of businesses that will harvest AI’s productivity gains without having to spend hundreds of billions to build the underlying models. As of mid-2026, the valuation gap between U.S. growth stocks and international developed-market equities sits near multi-decade extremes, which means the rotation Davis is describing doesn’t require AI to succeed — it only requires valuations to mean-revert even modestly. For long-term investors, that’s an asymmetric opportunity worth sitting with.
What does this mean for the patient investor today? It means being honest about where returns are still available versus where they’ve already been captured. Davis is not telling investors to abandon technology exposure or to time the market. He is telling them that a portfolio overweight to AI infrastructure — trading at 30-40x earnings in many cases — is making a concentrated bet on continued multiple expansion in an already richly priced segment. Spreading exposure toward value stocks with lower starting valuations, dividend-paying international equities trading at 12-15x earnings, and investment-grade bonds yielding 4-5% isn’t pessimism about AI. It’s an acknowledgment that transformative technologies reliably enrich their adopters more than their architects — and that the next decade of compounding may belong to companies you haven’t been watching closely enough.
