AWS CFO says AI spending is sustainable, citing ~$500B backlog and $25B+ AI run rate

In an exclusive interview with Axios, AWS chief financial officer John Felton argued that the AI infrastructure boom is sustainable because it is producing real innovation and revenue. He said he is "comfortable about what we're doing" on capital expenditures, citing a nearly $500 billion order backlog as of Q2. That backlog comes from a mix of "big and small customers, large enterprises, large AI labs, but also small and medium-sized businesses." AWS's AI revenue run rate is above $25 billion and growing at triple digits.

Felton's argument is that AI is best seen as a productivity tool rather than a cost-cutting one. It is about "more interesting customer experiences, more interesting products, more revenue-driving features." Inside his own finance team, he called AI "an amazing muck reduction tool" that lets professionals spend time on higher-value work. AWS is also building products to turn infrastructure into cash flow, including the preview of Bedrock Managed Agents powered by OpenAI and an expanded AgentCore with web search and consumption-based memory pricing.

The context is a market getting nervous. Gartner analysts rattled investors this week by estimating that more than $3.5 trillion in AI infrastructure spending "could be exposed to impairment" by the end of 2030 if revenue fails to keep up with buildout costs. Gartner also projected that more than 80% of specialized AI cloud providers will stop operating independently by 2030. Debt levels, sky-high valuations and circular financing between chipmakers and labs are all under scrutiny.

Axios notes AWS is exposed to the same risks but better positioned than most, as a hyperscaler with diversified revenue. Skeptics will note that backlog is not revenue and that much of it depends on a small number of large AI labs whose own finances are still being tested, including Anthropic's upcoming IPO.