Amazon and Google are winning the AI capex race — but what’s the prize?

At times, the artificial intelligence industry can look like a contest to see which company can spend the most money on data centres. The prevailing belief is straightforward: build the most data centres, secure the most computing power, and you'll be best positioned to create the strongest AI products — setting yourself up to dominate the future. There are obvious limits to this logic, since businesses ultimately succeed by earning more than they spend, but it has nevertheless proven deeply persuasive among the largest technology companies.

If that's the competition, Amazon appears to be out in front.

In its earnings report on Thursday, Amazon said it expects to spend roughly $200 billion on capital expenditures through 2026, covering investments in "AI, chips, robotics, and low Earth orbit satellites." That figure is up sharply from $131.8 billion in capex in 2025. It's tempting to assume that most of that spending is aimed squarely at AI. But unlike many of its rivals, Amazon operates a vast physical infrastructure — including warehouses increasingly outfitted with expensive robots — making it harder to dismiss the company's non-AI investments as marginal.

Google is not far behind. In its earnings release on Wednesday, Google projected capital expenditures of between $175 billion and $185 billion for 2026, up from $91.4 billion the year before. That represents a dramatic increase in fixed-asset spending and far exceeds what most competitors plan to spend.

Other major tech firms trail the leaders by a noticeable margin. Meta, which reported earnings last week, forecasted between $115 billion and $135 billion in capex for 2026. Oracle, once viewed as a flagship name in AI infrastructure, projects revenue of approximately $50 billion. Microsoft has not yet released an official 2026 capital expenditure (capex) projection, but its most recent quarterly spending was $37.5 billion. Annualised, that would amount to roughly $150 billion, assuming the pace continues. While that marks a significant increase—and has already triggered investor pressure on CEO Satya Nadella—it still places Microsoft behind Amazon and Google.

Within the tech industry, the rationale behind this spending spree is clear. If AI fulfils its promise, advanced computing power will become one of the scarcest and most valuable resources in the global economy. In that future, only companies that control their own compute supply will be able to compete at the highest level. Acting on that belief, Amazon, Google, Microsoft, Meta, Oracle, and others are racing to prepare for what they see as a coming shortage of compute.

Wall Street, however, is far less enthusiastic. Investors have reacted sceptically to the scale of these commitments, with each company's stock price falling after announcing major spending plans. Notably, the companies pledging the most money tended to experience the sharpest declines.

Importantly, this investor pushback is not limited to firms still searching for a clear AI monetisation strategy, such as Meta. It extends to companies such as Amazon and Microsoft, which have well-established cloud businesses and relatively straightforward paths to generating revenue from AI. The sheer size of the numbers involved has proven unsettling for investors.

Investor sentiment is not the sole factor shaping corporate strategy—and in this case, it may do little to slow the industry's momentum. If AI is genuinely on the verge of reshaping the global economy, many executives would argue that retreating simply because markets are nervous would be a mistake. Still, as spending continues to rise, Big Tech companies are likely to face growing pressure to soften how they discuss the actual cost of their AI ambitions.