NEW YORK, July 21, 2026, 13:09 EDT
- Amazon.com NASDAQ:AMZN, Microsoft NASDAQ:MSFT, Alphabet NASDAQ:GOOGL and Meta Platforms NASDAQ:META are set to allocate between $695 billion and $725 billion for total capital expenditures in 2026.
- On an annualized basis, Meta’s planned Anthropic lease represents up to 3.7% of its capex midpoint, before expenses.
- Alphabet is set to announce its results on Wednesday, with LSEG LON:LSEG analysts expecting a 64% increase in quarterly cloud revenue.
The key challenge for Wall Street in AI is not a single compute lease, but whether four hyperscale companies can turn around $710 billion in yearly capital spending into profit.
The Meta-Anthropic deal could be valued as high as $10 billion across two years, equating to an annual maximum of $5 billion.
This represents 3.7% of Meta’s capex midpoint of $135 billion and 0.7% of the combined midpoint across the four companies. The scale of the deal does not confirm full utilization, nor does it provide evidence of oversupply.
U.S. markets were trading at 13:09 EDT. The S&P 500 climbed 0.8%, and the Nasdaq advanced 1.3% at 11:30 a.m.
Nvidia NASDAQ:NVDA was up 1.0% at about 12:54 EDT. Shares of Amazon, Microsoft, and Alphabet declined between 0.7% and 0.9%, while Meta rose 0.3%.
The table matches company forecasts to share values as of around 12:54 EDT. Ratios are calculated based on the midpoint of spending plans.
| Company | 2026 capex plan | Market value | Capex / market value* |
|---|---|---|---|
| Amazon | $200 billion | $2.70 trillion | 7.4% |
| Microsoft | $190 billion | $2.97 trillion | 6.4% |
| Alphabet | $180 billion-$190 billion | $4.23 trillion | 4.4% |
| Meta | $125 billion-$145 billion | $1.66 trillion | 8.1% |
| Total | $695 billion-$725 billion | $11.56 trillion | 6.1% |
This screen displays scale and is not a forecast of investment returns.
Meta leads the one-year capex-to-equity ranking, reporting 8.1%, with Amazon next at 7.4%. The figure signals urgency but does not guarantee outcomes.
Meta’s leasing of infrastructure may provide monetization opportunities outside of advertising. However, an opinion piece interprets compute leasing as an indicator of oversupply.
Credible reports do not confirm an oversupply across the system. Discussions are still in initial stages, and conditions may shift; both sides retain the option to walk away from a deal.
Chief Executive Mark Zuckerberg stated that entering the cloud market was “definitely on the table.” Zuckerberg noted that companies were reaching out to Meta “almost every week” seeking access to models or extra capacity. Reuters
Alphabet faces a fresh challenge on Wednesday. The anticipated June debut for its core Gemini 3.5 Pro was delayed.
LSEG’s average forecast projects second-quarter revenue at $116.93 billion, marking a 21.3% increase. Cloud revenue is predicted to climb 64%. These numbers are projections, not official results.
UBS Group NYSE:UBS projects hyperscaler capex to climb 76% this year, reaching $673 billion. The company forecasts a 25% increase in the following year and 6% growth in 2028.
Alexis Bossard at Edmond de Rothschild described reduced spending as “a relief for hyperscalers and a negative signal for the semi industry.” This remains the key driver for earnings. Reuters
The high valuation leaves scant margin for postponement. The Financial Times reported the Shiller CAPE close to 41, while its excess CAPE yield stood at 1.4%. Analysts anticipate S&P 500 earnings will climb 24% this year, with a further 17% increase projected for 2027.
According to Goldman Sachs NYSE:GS Research, companies associated with AI have added roughly $27 trillion since late 2022. The firm’s baseline estimate for present value of prospective AI-driven capital revenue stands at $9 trillion.
The company points out that AI was not the sole driver of the increase. Analysts Dominic Wilson and Vickie Chang cautioned against “overestimating the persistence” of supplier earnings. Goldman Sachs
Risks: Quicker uptake by businesses may drive higher utilization and support greater investment. However, delays in model rollout, increased rates, or softer demand could negatively impact cash flow and orders from suppliers.
For investors, the upcoming indicator is not a fresh surge in chip stocks. It is the amount of revenue generated for each dollar of deployed compute.