Amazon reported its third-quarter results. Revenue at its cloud unit AWS climbed 39% year over year; total revenue rose 12% to $201.9 billion, and operating profit reached $28.1 billion. Over the same period, the company's trailing 12-month free cash flow fell to negative $42.4 billion — down from positive $14.8 billion a year earlier. Amazon attributed most of the decline to AI spending and said it would ramp up investment even further in 2027.

What happened?

Two opposing pictures sit side by side in a single quarter. On one hand, there's AWS's re-accelerating growth: the 39% increase shows that demand in the cloud market isn't cooling but, on the contrary, is gaining momentum thanks to AI workloads. On the other hand, there's a record level of negative cash flow. The gap comes from the enormous capital expenditure (capex) pouring into data centers, custom chips and GPU infrastructure. Revenue is rising, but cash is going out faster.

Why does it matter? When will the AI capex pay off?

The public debate mostly revolves around the question of whether there's an "AI bubble." Amazon's balance sheet turns that into the wrong question. AWS's 39% growth says demand is real; growth wouldn't be this fast if customers weren't using the capacity. The issue isn't whether demand exists — it's the timing of the cost.

Capital expenditure is paid up front, while the revenue it generates is spread over years. A data center is built today and its chips are bought today; the rental income that infrastructure produces is distributed across later quarters, even years. So negative cash flow on its own isn't a sign of collapse but a phase gap between accounting and physical investment. What's critical is how many quarters that gap lasts. If the spending turns into usage revenue within a few years, the investment makes sense; if the conversion slows or the capacity sits idle, the same numbers suddenly become evidence of overinvestment.

Amazon explicitly saying it will increase capex even further in 2027 is what creates the real tension here. The company is placing a bet between cloud margins and investor patience. AWS has historically been Amazon's most profitable unit; it carries retail's thin margins. If AI investment pressures that margin in the short term, the "when will I see a return?" question from the investor who has held the stock for years grows sharper.

Who does it affect, and how? A comparison with Microsoft and Google

This isn't just Amazon's story. All three major cloud providers are running in the same direction, but with different risk profiles. The table shows the bet isn't the same for everyone:

CompanyCloud unitAI advantageCore risk
AmazonAWS (market leader)Broad customer base, custom chips (Trainium/Inferentia)Heaviest capex burden, thin group margin
MicrosoftAzureOpenAI partnership, enterprise software integrationDependence on an outside model
GoogleCloudIts own models and TPU hardwareThird place in cloud, reliance on ad revenue

All three are reporting record capex, because a provider that can't build capacity loses AI workloads to a rival. This is a "spend until you win" race. But it's unclear who will turn the bill into profit first. Amazon's advantage is the chance to reduce its dependence on Nvidia and lower unit costs by making its own chips. Its disadvantage is that its group-wide margin is already narrow, which makes the capex pressure more visible.

For enterprise customers, this competition is good news for now: an abundance of capacity and a variety of chips could mean price pressure over the long run. For small AI startups, the picture is more mixed; as infrastructure costs fall, access gets easier, but dependence on the three giants deepens.

What's next?

There are a few concrete indicators to watch. The first is AWS's operating margin: is the margin holding as growth continues, or is capex eroding it? The second is the trough of free cash flow — when does the slide into the red slow and turn toward recovery? The third is whether Amazon revises its capex guidance up or down in the coming quarters — the most honest measure of the company's own confidence.

In the end, these figures look less like a bubble bursting and more like the early stage of an expensive, long bet. Demand is real, capacity is real; the only uncertain thing is the timeline of the payoff. Amazon may be making the right bet, but what it's asking of investors is patience — and the price of that patience becomes visible in negative cash flow every quarter.