Andy Jassy used Amazon’s third quarter earnings call to put a number on the end state of the AI buildout. AWS is a 184 billion dollar annualised revenue business after third quarter revenue rose 39.3 percent to just short of 46 billion dollars, and Jassy said it can grow by more than 40 percent in each of 2027 and 2028 and become a one trillion dollar annual revenue business within five to ten years. No company has ever reported a trillion dollars of annual revenue from a single business.
The physical pace behind the claim is the detail that matters. Jassy said AWS installs a new AI accelerator about every 14 seconds, and that customers absorb capacity almost as fast as it is installed. Bedrock, the company’s AI services platform, is already its fastest growing multi billion dollar service, and core compute and storage revenue grew at their fastest rates in several years, which Jassy argued is structural: AI workloads pull core services with them because inference needs to sit near the customer’s data and applications.
The funding requirement is stated just as plainly. Amazon expects about 220 billion dollars of cash capital expenditure in 2026, most of it for AWS, having spent 75.9 billion dollars in the third quarter alone, and expects to spend meaningfully more in 2027. Jassy called AI the biggest transformation of his lifetime and said the company can fund the plan, while conceding it would be more convenient if the opportunity could be spread over 25 years.
Investors did not buy the framing on the day. Amazon shares fell 4.25 percent on Thursday to 243.79 dollars, and were down about 6 percent in after hours trading during the call, even though group revenue reached 180.17 billion dollars and Prime Video was described as profitable with a revenue run rate in the tens of billions of dollars. The contrast with Alphabet, down 12.64 percent the same day, is one of degree rather than kind: in both cases the market is asking when accelerator installation converts into cash, not whether demand exists.