Cloud Profit Margins Offer Few Clues on AI Returns

Nearly four years into the AI infrastructure boom, the tech industry's largest players have yet to demonstrate clear profitability from their data center investments, according to recent earnings disclosures.

Amazon, Microsoft, Alphabet and Meta don't separate AI-related revenue and profit from their broader cloud computing operations. Instead, AI results are bundled into Amazon Web Services, Microsoft's Intelligent Cloud division, and Google Cloud. Meta, which lacks a cloud business, uses AI primarily to enhance its existing products.

Why it matters

The companies are collectively spending roughly $800 billion this year on AI infrastructure buildout, including Oracle. Markets and economic growth projections depend on the assumption these investments will eventually generate substantial returns — but earnings data has provided no confirmation that profitability is materializing.

Mixed signals from cloud margins

Amazon Web Services saw operating margins climb 6.5 percentage points to 39% in the second quarter of 2026. However, Amazon executives declined to attribute the increase to AI and suggested the margins might not hold.

Google Cloud's operating margin jumped to 35.6% from 20.7% a year earlier. But Alphabet warned that adding capacity in coming quarters "will put some pressure on operating margins for cloud."

Meta's overall operating margin fell more than 12 percentage points to 31%, largely due to legal expenses and severance costs. Excluding those items, the margin would have been 36.8% — still below the 43% recorded in the same quarter of 2025.

Microsoft's Intelligent Cloud division maintained flat operating margins at 41% despite heavy AI investment costs, suggesting potential for profit growth if capital expenditures moderate.

Analyst concerns mount

Rishi Jaluria, an analyst with RBC Capital Markets, noted that "the margins that Microsoft and Amazon and Google and Oracle are getting off AI are meaningfully less than traditional cloud."

While analysts aren't alarmed by the lack of near-term AI profitability, they grow concerned when companies plan to continue rapid data center expansion without clear returns. "What people are concerned about is: Are you building too much capacity?" said Jason Helfstein, head of internet research at Oppenheimer & Co. "If the world builds too much of it, the price is going to go down."

Customer concentration risk

A separate worry centers on customer concentration. Stephen Bersey, head of U.S. technology research at HSBC, estimates that approximately 50% of AI-related backlogs disclosed by Oracle, Amazon, Microsoft and Alphabet represent orders from just two companies: Anthropic and OpenAI. Both firms depend on continued investor funding to maintain their spending levels.

"Considering both OpenAI and Anthropic are currently private, there is significant uncertainty in assessing customer concentration risk for hyperscalers," Bersey wrote.

These details were first reported by Axios.