AI Could Transform Everything, and Investors Could Still Lose

The S&P 500 and Nasdaq Composite are flirting with all-time highs, but the rally rests on a startlingly narrow foundation:

• Companies linked to artificial intelligence (AI) now account for 45% of S&P 500 market capitalization.
• Alphabet, Amazon and Meta alone explain roughly 70% of the increase in index earnings expectations this year.
• AI-related investment could account for roughly 75% of 2026 U.S. GDP growth, according to some economic estimates.

Strip out the AI complex and the broader market is about 13% below its 52-week high, with non-AI GDP growth running at a negligible 0.1%. Today’s market value is a singular bet on a single technology—a one-trick pony of historic proportions.

The five largest American hyperscalers—Amazon, Alphabet, Meta, Microsoft and Oracle—are expected to spend $775 to $800 billion on AI infrastructure in 2026 alone, a figure equal to about 2.5% of U.S. GDP. That projection rises to an estimated $880 billion in 2027 and crosses $1 trillion in 2028. Projections for AI capital spending from 2028 through 2030 exceed $4.14 trillion—about the size of India’s annual economy.

To justify this commitment at a 15% return on investment, Goldman Sachs calculates these companies need to generate approximately $1 trillion in AI revenue by 2030. Current annualized AI revenue is approximately $100 billion.

The stock market has accepted this thesis. The bond market has not.

AI spending is reshaping global fixed-income markets. In 2025, the five hyperscalers issued $136 billion in corporate debt and financed another $120 billion for data center construction through off-balance-sheet special-purpose vehicles. Meanwhile, Treasury yields posted their sharpest quarterly rise since 1994: the 10-year reached 5.33%, its highest level since 2007, and the 30-year climbed to 5.65%. Private credit funds have also quadrupled their AI-sector lending over five years, which now accounts for 15% of their portfolios. The Bank for International Settlements has classified this shadow-banking concentration as a “macroprudential risk,” warning that it could threaten system-wide financial stability.

These downside projections are stark. If the return on this massive AI investment doesn’t materialize as quickly as investors expect, there could be real market trouble. A delay of 18 to 36 months, for instance, could plausibly bring about a collapse of AI-specific investments, igniting a market correction and a GDP demand shock.

But AI’s upside could be transformative, even life changing. Goldman Sachs projects it could add about 0.4 percentage points to annual U.S. GDP growth beginning in the late 2020s. McKinsey & Company estimates a cumulative 13.7% GDP uplift over a decade. Healthcare, financial services, logistics, legal services and manufacturing each face productivity step-changes of a magnitude not seen since the personal computer was introduced in the 1980s.

History teaches us that promising new technologies, like steam-power, electricity, computers and the internet, may not lead to financial gains as rapidly as initially hoped, but they eventually transform economies and societies. The ultimate winners are individuals and companies that embrace the new technologies and reorganize to best use them to innovate.

That AI reorganization is just starting and hints at spectacular innovations to come.