The AI Investment Divide: Why the Gap Between the U.S. and Europe Is Widening Fast

In a sweeping new report published in October 2026, Apollo Global Management Chief Economist Torsten Slok, European Economic and Policy Strategist Huw van Steenis, and their colleagues lay out one of the most data-rich assessments yet of the macro implications of the AI investment boom¹. The picture that emerges is not simply one of technological progress. It is a divergence story, with the United States accelerating away from Europe across every dimension the team measures.

A Capital Boom Without Modern Precedent

The team places the AI buildout in a striking historical context. At 1.9 percent of U.S. GDP over the 2022 to 2029 period, AI capital spending trails only the railroad boom of 1865 to 1890 in scale. The telecom and fibre build of the 1990s, the interstate highway system, canals, electrification: all fall below it.

The numbers are accelerating. "Expectations for hyperscaler capex in 2027 have doubled in the last 12 months," the team notes, with combined spend across Amazon, Alphabet, Meta, Microsoft, and Oracle now projected at $1.131 trillion for 2027 alone, up from an estimate of $495 billion as recently as September 2025. S&P 500 capital expenditure, the team observes, "is booming on AI whilst Eurostoxx flattish."

The Europe Gap: Structural, Not Cyclical

Slok and van Steenis are unambiguous about the geographic dimension. "U.S. private investment booming on AI buildout; Europe lags," the team states flatly. American private AI investment reached nearly $300 billion in 2025. Europe barely registers on the same chart. In venture capital, AI now accounts for 90 percent of U.S. VC investment, 79 percent in the United Kingdom, 70 percent in China, and only 65 percent across the EU27.

The infrastructure gap compounds this. The U.S. holds 4,767 data centers compared to 568 in the United Kingdom and 533 in Germany. Installed data center capacity stands at 52 gigawatts in the U.S. versus 17 gigawatts across all of Europe. The team is direct: "EU supercomputer capacity materially lags US and China."

On intellectual property, U.S. firms invest in software and databases at a scale more than double the entire EU27 combined. U.S. firms now direct roughly 8.5 percent of net sales to R&D, compared to 3.5 percent for their European counterparts. In ICT software R&D specifically, the U.S. runs at 13 percent of net sales against Europe's 4.5 percent. "US firms are investing more than 2x the R&D to sales budget," the team confirms. Only one model from Europe appears in the ranking of the world's leading AI models, with the U.S. and China increasingly dominating the 50 most-used AI models globally.

Adoption: A Nuanced Split

The report introduces an important caveat on adoption. Several European countries, notably Norway, Ireland, and France, lead the United States on consumer AI diffusion. However, when it comes to workplace adoption, "American workers adopting AI faster," the team states, with 43 percent of U.S. workers using generative AI for work in 2026, ahead of the United Kingdom at 36 percent, the Netherlands at 35.6 percent, and Germany at 31.5 percent. U.S. firms also use AI tools more systematically across customer service, internal processes, marketing, and human resources.

The Productivity Question Remains Open

Perhaps the most consequential unresolved issue in the report is productivity. "US labor productivity has surged ahead," the team observes, with the U.S. index reaching approximately 122.5 versus roughly 105 for the Euro Area as of mid-2026. "US productivity growth re-accelerating; Europe stalling," the team adds, with the U.S. 10-year moving average on productivity growth climbing back toward 2 percent while Europe drifts toward zero.

Yet total factor productivity data shows no AI-specific signal yet. "No AI productivity boost in the data yet," the team notes, a finding consistent with the broader academic literature, which the report characterizes as "inconclusive about the impact of AI on productivity." Estimates from leading researchers range from Acemoglu's conservative 0.7 percent cumulative gain over 10 years to Bergeaud's 6 to 12 percent, with most studies pointing to time horizons of 10 to 20 years before the full effect materializes.

The Market Verdict

Markets have already delivered a partial verdict. "US has added $40trn in market cap since end-2022, dwarfing all of Europe," the team states, with the United Kingdom, Germany, Italy, and France collectively adding just $2.8 trillion over the same period. The Magnificent Seven alone carry a market capitalization of approximately $23.5 trillion, exceeding the combined public equity markets of the United Kingdom, France, Germany, and Italy. U.S. earnings per share are growing twice as fast as European counterparts, and forward EPS estimates are outpacing Europe "by a widening margin."

Five Key Takeaways for Advisors and Investors

1. The AI capex cycle is larger and more durable than most forecasts assumed. With 2027 hyperscaler capex projections doubling within a single year, advisors should treat the AI infrastructure buildout as a multi-year structural theme, not a near-term trade.
2. The U.S. versus Europe divergence is widening across capital, infrastructure, talent, and output. Portfolios with a structural U.S. equity overweight relative to Europe have macro tailwinds that are, if anything, strengthening.
3. Productivity upside is real but delayed. Academic consensus points to a 10-to-20-year horizon before AI meaningfully lifts total factor productivity. Investors pricing in near-term productivity gains should moderate their assumptions; those with long time horizons should stay positioned for it.
4. The Magnificent Seven concentration risk reflects structural earnings power. With Mag 7 profit margins approaching 25 percent and revenue per employee separating sharply from European peers, the valuation premium carries more fundamental support than prior technology cycles.
5. Watch for crowding out in credit and capital markets. The Apollo team flags AI's growing share of corporate borrowing, noting that 87 percent of VC net issuance and 49 percent of investment-grade issuance is now AI-linked. Advisors should monitor whether AI financing crowds out capital for other sectors.

Footnote:

1 Slok, Torsten, Huw van Steenis, Rajvi Shah, Shruti Galwankar, and Disha Biyani. "The Macro Implications of the AI Investment Boom: US and Europe." Apollo Global Management, Oct. 2026, https://www.apollo.com/content/dam/apolloaem/pdf/daily-spark/2026/oct/12/aiuseuv2.pdf.