Amazon is already spending a record amount on AI data centers. Next year will be even bigger.
The world’s biggest cloud-computing company reported third-quarter results on Thursday and reiterated its plan for $220 billion in capital expenditure for 2026.
Then, CEO Andy Jassy shared his outlook for next year, saying Amazon expects “to spend meaningfully more in 2027.”
Amazon’s spending has become the central question hanging over the company. It’s building massive data centers and buying expensive chips and other gear to meet surging demand for AI.
That’s pushed its free cash flow into negative territory, with third-quarter results showing a hole of $42.4 billion for the trailing twelve months.
Amazon spent $75.9 billion in cash capital expenditures during the third quarter alone, CFO Brian Olsavsky said, primarily on AWS and AI infrastructure.
The results did little to ease investors’ concern about the spending. Amazon’s third-quarter revenue and earnings came in below Wall Street expectations, and its shares fell roughly 6% in after-hours trading.
“The cloud unit’s improving operating margin was one welcome signal, but it may not be sufficient to alleviate concerns about the sharp spike in capex spending and corresponding freefall of its free cash flow,” Sky Canaves, a principal analyst at eMarketer, said in a statement.
Amazon has argued that AI infrastructure takes years to build before it can begin generating revenue, and that AI chip servers typically break even in less than three years once deployed. But the scale of the outlay has raised the stakes for Amazon’s cloud business to sustain rapid growth as competitors make similarly enormous bets.
On Thursday, Jassy reiterated his confidence in the company’s AI business model and outlook, saying there’s a “clear line of sight to attractive financial returns.”
The time it takes Amazon to turn newly delivered servers into revenue-generating capacity has also come under scrutiny.
Last week, Business Insider exclusively reported that Amazon has begun receiving billions of dollars in Nvidia server racks before all of the gear can be deployed at their final data-center sites, as power shortages and construction delays hold back new capacity.
“We can fund this plan”
Jassy argued Thursday that Amazon has visibility into the demand supporting its huge AI investments. Customers that need the largest AI deployments reserve capacity well ahead of time, he said, in blocks ranging from hundreds of megawatts to, in some cases, as much as a gigawatt.
“We have very good visibility into 2027 and 2028 and some of ’29 too with respect to what customers want in those really big chunks of capacity,” Jassy said.
AWS revenue rose 39.3% year over year to $46 billion in the third quarter, its fastest growth in nearly five years. Jassy said he expects AWS revenue to grow more than 40% in both 2027 and 2028, a projection that requires the company to keep rapidly adding capacity.
Jassy described the spending as a once-in-a-lifetime opportunity rather than an optional expansion. It would be more convenient, he said, if Amazon could spread the investment over 25 years. But the speed of the AI market does not allow that.
“We can fund this plan,” he said. “It will forever change our revenue. It will forever change our free cash flow.”
Have a tip? Contact this reporter via email at ekim@businessinsider.com or Signal, Telegram, or WhatsApp at 650-942-3061. Use a personal email address, a nonwork WiFi network, and a nonwork device; here’s our guide to sharing information securely.
