Amazon Already Has 2028 Booked: AWS 37% Surge Puts Big AI Demand on Display

Andy Jassy says 2028 capacity demand is the real signal

Amazon is not treating AI demand as a future scenario. On the latest earnings call, Andy Jassy said demand for 2028 capacity is demand we already have for 2028 is striking, turning AI from a long-dated narrative into a present-day capacity constraint at one of the world's largest infrastructure platforms.

AWS growth came back fast

The operating numbers reinforce that shift. AWS revenue jumped to $42.2 billion, up 37%, while total revenue rose to $200.61 billion. Management also raised 2026 cash capex to roughly $220 billion. Even that increase still leaves Amazon short of capacity for 2026, let alone the demand extending into 2027 and 2028 we now believe we will spend approximately $220 billion in cash capex in 2026.

Why this matters beyond one strong quarter

This is not just a good quarter for AWS. It is a sign that capacity planning is becoming a multi-year issue. When a company of Amazon's scale is already talking about striking demand three years out, the risk for investors is underestimating the buildout required before that demand turns into sustained revenue and cash flow.

Why AI demand looks like a multi-year capacity squeeze

Enterprise adoption is still early

Andy Jassy said enterprises are still very early in using inference at scale in production. That matters because demand does not have to arrive all at once to keep straining capacity. Jassy also said about 85% of global IT spending remains on-premises, a reminder that the migration to cloud infrastructure can remain a long runway rather than a short burst.

Infrastructure takes time to build, even if demand arrives quickly

Amazon's buildout is tied to data centers with 30-plus-year useful lives, while servers inside them can cycle through five or six generations. In practical terms, the facility is a long-term asset, but the equipment needs regular refreshes. That is why supply cannot simply flip on when orders show up: the physical plant has to be planned and built well in advance.

Cash flow can lag before utilization improves

Jassy was direct that heavy construction creates free-cash-flow headwinds until data centers come online and servers spend a few years in use. So the spending curve can look tougher than the revenue curve in the short run, even if the long-term demand case is intact.

What to watch: - Whether spending remains concentrated around the 2026 plan - Whether leadership continues to describe demand stretching into 2027 and 2028 - Whether utilization improves enough to show that newer server generations are earning better economics on older infrastructure

If enterprise adoption stalls or capacity comes online faster than expected, the payback period could compress. If not, Amazon's capex profile may keep looking uneven even while underlying demand compounds.

The real investor debate is scale versus timing

AWS is compounding again, and the gap to Azure is narrowing

AWS has accelerated for five straight quarters. That matters because AWS is growing from a very large base, so sustained acceleration is harder to achieve than at smaller peers. Bulls see a business that is not merely stabilizing but gaining momentum at a time when demand for AI infrastructure remains tight.

If that momentum continues while Amazon keeps its long-term ceiling in view, investors may be underestimating how much revenue and cash flow AWS could generate over time.

The countercase: capex is front-loaded, and cash flow may disappoint

The bear case is less about whether demand exists and more about when the money comes back. Amazon expects approximately $220 billion in cash capex in 2026, and trailing free cash flow has been negative. Jassy has said the heavy spending phase creates near-term free-cash-flow pressure until new capacity is monetized.

That leaves the core debate on AWS simple: can the business widen its moat and grow fast enough, fast enough to convince investors the payback period is worth the strain?

What would strengthen or weaken the thesis

The bullish case gets stronger if: - AWS keeps posting stronger growth while margins hold up - Management continues to describe capacity constraints into 2027 and 2028 - New capacity comes online and shows healthy utilization

The thesis weakens if: - Leadership no longer sounds constrained into 2027 and beyond - Spending stays high, but AWS does not convert that spend into stronger revenue or better returns - The market signals that demand is less broad than current capacity plans assume

Amazon's results cleared the first hurdle; the next test is follow-through

After results that came in higher than analysts' expectations, the next step is practical. Investors do not need more proof that AI is important. They need evidence that AWS demand keeps translating into revenue strength, utilization, and eventually cash generation that catches up with the capex plan.