A $650 billion spending spree on AI infrastructure is forcing investors to rethink the decade-old "asset-light" thesis for big tech.
Back
A $650 billion spending spree on AI infrastructure is forcing investors to rethink the decade-old "asset-light" thesis for big tech.

The tech-heavy Nasdaq Composite has fallen 5.9% in 2026, driven by geopolitical uncertainty and a fundamental shift in the sector’s financial structure, as giants like Amazon, Meta, and Alphabet commit a combined $650 billion to AI capital expenditures this year alone.
"Tech stocks were once viewed as solid investments because they were ‘asset light, revenue heavy, cash flow generating machines,’" Steve Sosnick, chief strategist at Interactive Brokers, told Barron's. The massive spending commitments have changed that narrative.
The spending surge is reshaping balance sheets. Amazon.com, Meta Platforms, and Alphabet are building out data centers to house the Nvidia-powered hardware required for AI, a move that pressures free cash flow in the short term. This comes as the Nasdaq’s decline was exacerbated by a broader market shift to risk-off assets following the start of the Iran war.
This capital-intensive era challenges the high-margin, low-asset valuations that propelled tech stocks for a decade. For investors, the key question is whether future AI-driven revenue can justify the upfront cost, a dynamic that benefits "picks and shovels" suppliers while creating new risks for the hyperscalers themselves.
"If this kind of pull back really made you nervous, you’re probably carrying too much risk," Sosnick said, suggesting an opportunity to look toward low-beta, high-dividend stocks. He advises focusing on companies that can fund dividends from free cash flow, a metric now under pressure at the big spenders.
Conversely, the selloff has reduced valuations for some of the sector's biggest names. John Belton, portfolio manager at Gabelli Funds, notes that four of the Magnificent Seven stocks—Microsoft, Nvidia, Amazon, and Meta—are now trading below their five-year forward price-to-earnings averages, potentially creating long-term entry points.
For those wary of the hyperscalers' spending, the beneficiaries are the suppliers. "While the hyperscalars’ cash flow is going to zero, it’s a bonanza if you look at memory and optical and other names like that," said Jonathan Cofsky, portfolio manager at Janus Henderson Investors. This points to an investment thesis centered on semiconductor, memory, and hardware companies that directly receive the $650 billion in capital expenditure.
This article is for informational purposes only and does not constitute investment advice.