Amazon’s AI spending surge wins Wall Street’s approval — for now

Amazon’s second-quarter results showed that investors still reward AI-heavy growth when it comes with fast-rising cloud revenue. The company beat earnings expectations, lifted sales 20% year over year, and sent its shares up nearly 10% after hours on Thursday, even as it committed to far more data center spending.

That matters because Amazon is one of the clearest examples of the current AI market split: cloud providers with visible demand are being applauded, while AI developers and capital-intensive platforms without a clear revenue path are facing sharper scrutiny.

Amazon’s quarter reassures investors on AI demand

Amazon Web Services was the main reason the market reacted so positively. AWS revenue climbed 37% from a year earlier to $42 billion in the quarter, signaling that customers are still ramping up cloud use as they build and deploy AI systems.

The numbers did not magically erase Amazon’s rising infrastructure bill, but they gave investors a reason to believe the company’s spending will be matched by future sales. That distinction is important because data centers take years to build, outfit and bring into service. The market is effectively betting that today’s outlays will translate into tomorrow’s capacity — and tomorrow’s revenue.

Why did Amazon stock jump after earnings?

Because investors saw strong cloud growth paired with confidence about future demand. Revenue rose faster than expected, AWS posted robust growth, and management signaled that it intends to keep building for the AI boom rather than pull back.

Amazon executives argued on the earnings call that the AI opportunity can be very large even without a single dominant model, and that AWS and Bedrock can thrive by serving a broad mix of customers and workloads.

That message appears to have resonated. Markets have become more selective this earnings season, rewarding companies that can show tangible AI monetization and punishing those that cannot. Amazon fit the former category.

How much is Amazon spending on AI infrastructure?

Amazon is still spending aggressively. The company said it invested $173 billion in property and equipment over the fiscal year ended June 30, up from $107.65 billion a year earlier. That line includes items such as GPUs, land, and natural gas turbines — the kind of physical assets required to power large-scale cloud and AI operations.

It also raised its 2026 capital expenditure outlook from $200 billion to $220 billion, underscoring that management expects demand to justify even more buildout. For the quarter, Amazon’s cash balance fell by $7.6 billion compared with a year earlier, and the company recorded its first negative free cash flow period of 2026.

For many businesses, that combination would raise alarms. In Amazon’s case, it is being read as a sign of conviction. The company is choosing to spend through the cycle in order to secure future AI capacity and protect AWS’s long-term competitive position.

What is Amazon betting on beyond data centers?

Amazon’s AI strategy is broader than warehouses full of servers. It is also investing in custom silicon, including the Trainium chip family and the Arm-based Graviton processors, both of which are designed to improve economics inside AWS.

Those chips are strategically important even if they are less visible than a new data center campus. Custom hardware can lower compute costs, increase efficiency and improve margins over time. In other words, Amazon is not only buying capacity — it is trying to make that capacity cheaper to run.

That approach reflects a familiar Amazon pattern: spend heavily up front, then use scale, tooling and proprietary infrastructure to improve profitability later.

Why custom chips matter in the cloud race

Custom chips can help AWS avoid relying entirely on third-party suppliers and can give customers cheaper ways to train and serve models. They also strengthen Amazon’s position against rivals that are pursuing similar economics through their own hardware and platform investments.

In a market where AI workloads are expensive and competitive, every small improvement in efficiency matters. For cloud providers, a few percentage points of margin can shape whether a platform wins or loses enterprise customers over time.

How does Amazon compare with Microsoft and Google?

Amazon is not alone in seeing the market reward cloud-led AI growth. Microsoft and Google also saw share-price gains after posting strong cloud results, reinforcing the idea that investors currently view the cloud layer as the safest place to make money from AI.

That pattern stands in contrast with Meta, where investors are growing more uncomfortable with big spending and limited near-term revenue clarity. Meta’s shares dropped 8% after earnings this week as traders focused on cash flow pressure and ongoing capital outlays.

The difference is not simply about who is spending the most. It is about whether investors can see a credible near-term business model attached to that spending. Amazon, Microsoft and Google can point to cloud revenue today. Meta’s AI ambitions are more diffuse and less directly monetized.

Why investors are still nervous about the wider AI economy

Amazon’s results were reassuring, but they do not settle the larger question hanging over the AI boom: is the industry building enough real demand to justify the unprecedented amount of capital being deployed?

The current market structure is circular in a way that can look healthy until it does not. Amazon earns cloud revenue when AI companies spend on infrastructure. Anthropic and other model builders spend heavily on compute to develop products. Their spending becomes Amazon’s revenue, and the same dynamic applies across much of the cloud stack.

That arrangement works only if usage continues to grow and if customers eventually make enough money from AI products to keep paying for more compute. If the demand curve softens, the weakest links in the chain could feel the pressure first — but the cloud providers would not be immune for long.

In practical terms, Amazon’s hosting revenue depends on someone else’s AI budget. If those budgets tighten, the cloud revenue stream can weaken too.

The concern is especially visible among AI startups and model labs, where many businesses are still chasing product-market fit. Investors appear more willing to fund infrastructure with visible demand than frontier-model companies with enormous costs and uncertain payback periods.

What is David Cahn’s $3 trillion question?

It is the market’s central unresolved bet: whether AI demand will ultimately be large enough to support the trillions of dollars of infrastructure now being planned and built. If the answer is yes, today’s spending may look prudent in hindsight. If the answer is no, the industry may be overbuilding.

Amazon’s earnings did not answer that question, but they did sharpen it. AWS is acting as a proxy for broader AI adoption because it sits close to the money flow. When cloud demand rises, it suggests that companies are still willing to pay for AI experimentation and deployment.

Still, close does not mean safe. Cloud providers may be a step removed from AI product risk, but they are not insulated from it. Their fortunes rise or fall with the same underlying demand that powers the rest of the ecosystem.

The bigger picture: cloud hosts are the market’s preferred AI trade

The latest earnings cycle suggests a simple rule of thumb for investors: AI is easiest to like when it shows up as cloud revenue. That is why Amazon, Microsoft and Google are being treated as relatively reliable bets, while many AI labs and startups are asked to prove their economics.

This does not mean the current wave of spending is irrational. It means the market is distinguishing between providers selling infrastructure today and companies still asking it to believe in future usage. Amazon falls into the first group, which helps explain both the enthusiasm and the scrutiny around its results.

The next few quarters will show whether AWS can continue growing quickly enough to absorb the company’s massive spending plans. If it can, Amazon’s AI strategy may look prescient. If growth slows, the market’s current confidence could fade just as fast as it arrived.

Timeline of Amazon’s latest AI and cloud story

Here is how the key developments line up around the company’s latest quarter and investment outlook.

What to watch next

Investors will now be watching three things closely: whether AWS growth can stay elevated, whether Amazon’s cash generation can catch up with the scale of spending, and whether AI demand across the broader market remains strong enough to support the current infrastructure boom.

If those pieces hold, Amazon’s aggressive posture may look like discipline rather than excess. If they do not, the current enthusiasm could unwind into a broader reassessment of how much AI capacity the market really needs.

For now, though, Amazon has managed to do what few AI-linked companies have accomplished this year: spend more, promise even more, and still win over investors.