Key Takeaways:
- Big Tech's combined AI capex is projected to hit $730 billion by 2026
- Every $1 of cash generation now requires $1.57 in infrastructure spending
- Rich valuations leave AI stocks exposed as Treasury yields climb to 4.7%
返回
Big Tech spends $730B on AI as debt costs test patience
Big Tech's $730 billion AI infrastructure bet is colliding with rising debt costs, testing investor tolerance for spending without immediate returns.
Big Tech's combined AI infrastructure spending is on track to hit $730 billion by 2026, but every $1 of additional cash generation now requires $1.57 in investment, raising questions about return on capital.
"The market is shifting from 'spend whatever it takes' to 'show me the returns,' and that transition is going to be painful for companies that over-invested without clear monetization paths," said Rachel Kim, AI infrastructure analyst at Edgen.
Combined 2026 capital expenditure forecasts for Microsoft Corp., Alphabet Inc., Amazon.com Inc., Meta Platforms Inc. and Oracle Corp. jumped to $730 billion in July from about $485 billion in January, according to Reuters analysis of LSEG consensus estimates. By 2027, their cumulative capex is projected to rise $534 billion from 2025 levels, while operating cash flow is expected to increase by just $340 billion — a ratio of $1.57 in spending for every $1 of cash generation.
The spending gap leaves the five hyperscalers exposed to rising financing costs. The 10-year Treasury yield touched 4.7%, and markets are pricing a 38% chance of a 25-basis-point rate hike, up from 12.8% a week earlier, according to CNBC. At the S&P 500's 20.1-times forward earnings multiple, the implied earnings yield of roughly 5% offers just 0.3 percentage points of compensation over Treasuries — a spread that leaves richly valued tech stocks vulnerable to any repricing of risk.
Investors have already begun punishing companies that spend heavily on AI without demonstrating clear returns. Alphabet shares fell nearly 7% and Tesla lost more than 14% after their most recent quarterly results, with both companies facing scrutiny over rising AI capital expenditure and uncertainty about when those investments will translate into meaningful revenue growth. Meta Platforms also saw its stock slide sharply as investors reacted to the scale of its AI spending plans.
The valuation math is particularly acute for the most expensive names in the AI trade. Tesla Inc. trades at 174.1 times forward earnings, a 994% premium to its five-year average of 15.9 times, according to Seeking Alpha data cited by Bank of America. Arm Holdings commands 124.4 times forward earnings, a 415% premium to its historical average. Advanced Micro Devices Inc. trades at 70.1 times forward earnings, a 197% premium. Even Meta, at 18.8 times, trades at a 49% premium to its five-year average of 12.6 times.
Bank of America technical analyst Paul Ciana warned that the August-through-October period has historically been the S&P 500's weakest three-month stretch, with the index rising only 55% of the time since 1928 and suffering an average maximum drawdown of 7.35%. For a market carrying elevated tech valuations and heavy concentration in a handful of mega-cap names, a pullback of that magnitude could prove more disruptive than the near-flat average return suggests.
The earnings season now under way will test whether the economics of the AI buildout can justify the spending. Microsoft and Meta report on Wednesday, followed by Amazon and Apple on Thursday. FactSet data shows S&P 500 earnings rose 23.3% year over year in the second quarter, with 57% of companies issuing positive guidance. But investors are demanding more than top-line beats — they want evidence that cloud growth, advertising gains and AI products are converting infrastructure spending into free cash flow.
This article is for informational purposes only and does not constitute investment advice.
