Cloud giants plan $595 billion for AI infrastructure

Amazon, Google, and Microsoft are collectively forecasting about $595 billion in 2026 capital spending, as demand for AI training and inference pushes cloud infrastructure expansion to unprecedented levels. The total is not a like-for-like comparison: Amazon reports cash capex across several businesses, Alphabet’s figure covers the entire company, and Microsoft’s number includes the effect of lease-accounting changes.

Even with those qualifications, the spending surge is clear. Cloud revenue is rising alongside investment, yet the companies say they still cannot add capacity fast enough to satisfy customers.

Amazon raises 2026 capex forecast to $220 billion

Amazon Web Services generated $42.2 billion in revenue during the second quarter ended June 30, 2026, up 36.7% year over year. It was AWS’s fifth consecutive quarter of accelerating growth.

CEO Andy Jassy said AWS now has a $169 billion annualized revenue run rate, which would rank it 24th on the Fortune 500 if AWS were a standalone company.

Recommended reading

AI data centers are becoming integrated power campuses

Amazon has now increased its 2026 cash-capex forecast from approximately $200 billion to $220 billion. The company says most of that spending will support AI and AWS, while higher memory costs are also pushing the estimate upward.

“We now believe we will spend approximately $220 billion in cash capex in 2026, with the higher cost of memory pushing this number up from our prior estimate of about $200 billion. But even at that amount, we will still not have enough capacity to meet all the demand we have in 2026, and I believe this dynamic will also be true in 2027 too.”

Jassy compared the current build-out with the first era of cloud computing, but said demand has accumulated much faster with AI. He also argued that hardware shortages are encouraging companies with on-premises infrastructure to use cloud services instead, because suppliers are prioritizing the largest buyers — including cloud providers.

The shortages extend beyond accelerators. Memory, GPUs, and hard drives are all constrained, a trend that has already contributed to declining PC and smartphone shipments when manufacturers cannot secure enough memory. Seagate’s enterprise HDD capacity squeeze illustrates how storage demand is also being pulled into the infrastructure expansion.

Google says demand still outpaces capacity

Alphabet has raised its full-year 2026 capex guidance to $195 billion–$205 billion, up from an earlier range of $180 billion–$190 billion. CFO Anat Ashkenazi said the increase is mainly due to accelerating capacity delivery to meet demand.

Google Cloud revenue rose 82% to $24.8 billion in the quarter. Ashkenazi identified Core GCP, AI solutions, and AI infrastructure as important growth drivers, with GCP growing faster than cloud services overall.

Google expects investment to rise significantly again next year, although it did not provide a more specific figure. The spending will also increase depreciation and data-center operating costs, including energy, while Google continues hiring in AI and cloud and invests in marketing for its AI products.

“While we have increased our capacity quite significantly over the past three years, the demand still outpaces that investment. And we are, just like the rest of the industry, working in a supply-constrained environment, so we’re working hard to do this.”

That constraint is not just a matter of data-center construction. The AI build-out depends on a broader supply chain that includes memory and chipmaking equipment; SK Hynix’s planned DRAM expansion has already drawn reported interest from major technology companies willing to help fund new production capacity.

Microsoft’s $175 billion forecast reflects accounting changes

Microsoft reported $59.3 billion in commercial cloud revenue for the quarter ended June 30, its fourth quarter of fiscal 2026. That was up 27% year over year, while Azure and other cloud services grew 43%.

Microsoft expects approximately $175 billion in calendar-year 2026 capex, around $15 billion below its earlier figure. The reduction does not represent a cut to the infrastructure build-out. Instead, Microsoft is changing how it classifies some future data-center leases.

Starting with fiscal 2027, the company will extend the estimated useful lives of data centers and office buildings from 15 years to 25 years. More leases will also shift from finance leases to operating leases, changing how the spending appears in the forecast.

CFO Amy Hood said Microsoft’s underlying 2026 investment expectations are unchanged and that capex in the next quarter, Q1 FY27, will exceed $50 billion, including the impact of the lease reclassification.

Nearly $600 billion, with capacity still rationed

The three forecasts add up to roughly $595 billion, but the total should be treated as an indicator of scale rather than a precise industry comparison. Amazon’s number spans multiple businesses, Alphabet’s covers the whole company, and Microsoft’s includes lease-accounting effects.

The spending is nevertheless large enough to affect hardware availability beyond the cloud. Synergy Research reported that enterprise spending on cloud infrastructure services exceeded $143 billion in the second quarter, up 43% year over year, and reached $500 billion over the previous 12 months.

The sharper point is that spending at this level has not eliminated scarcity. Amazon says it will not have enough capacity to meet all 2026 demand and expects the problem to continue into 2027; Google says demand still exceeds its investment. For now, the cloud giants are not merely building for AI growth — they are spending hundreds of billions of dollars and still rationing access to the infrastructure customers want.