Headline & Intro
Amazon’s latest earnings show something unusual: one of the world’s most profitable cloud businesses is willingly crushing its own free cash flow to ride the AI wave. AWS is growing at a pace it hasn’t seen in years, but the bill for that growth is arriving now in the form of eye‑watering capital expenditure. This isn’t just another strong quarter; it’s Amazon nailing its colours to the mast in the AI infrastructure arms race. In this piece we’ll unpack what the numbers really say, who wins and loses, and why European companies in particular should be paying close attention.
The News in Brief
According to TechCrunch, Amazon reported a blowout first quarter for 2026, with Amazon Web Services (AWS) again the star of the show. AWS net sales hit $37.6 billion, up 28% year over year — the fastest growth rate for the unit in roughly four years. Amazon CEO Andy Jassy linked this acceleration directly to demand for AI workloads and said AWS’s AI-related revenue run rate over the first three years of this AI wave has already reached about $15 billion.
That growth is being fuelled by a surge in capital spending. TechCrunch reports that Amazon’s purchases of property and equipment, largely tied to AI infrastructure like data centres and chips, increased by $59.3 billion year over year. As a result, free cash flow for the trailing 12 months dropped to $1.2 billion, down from $25.9 billion a year earlier — a roughly 95% decline. Overall Amazon sales grew 17% to $181.5 billion, with solid growth in both North America and international markets.
Why This Matters
AWS is effectively choosing to behave less like a mature cash cow and more like a hyper‑growth startup again. For investors who fell in love with Amazon’s free cash flow story over the last decade, that’s a jarring pivot. But strategically, it makes sense: AI infrastructure is quickly becoming the new operating system for the digital economy, and missing this build‑out would be far more expensive than a few years of weaker cash generation.
The immediate winners are obvious. Chipmakers, data‑centre landlords, and power utilities tied into AWS regions can expect years of sustained demand. Enterprise customers that standardise on AWS for AI get earlier access to cutting‑edge compute, managed models, and tooling that smaller providers simply can’t match at this scale.
The losers are more subtle. Second‑tier cloud providers will find it even harder to keep up with the pace of investment and hardware sophistication. Startups that hoped to compete on raw infrastructure will be squeezed into niches or forced to build on top of the hyperscalers. Customers may enjoy low prices and generous credits today, but as AWS’s AI stack becomes more indispensable, Amazon’s long‑term pricing power and lock‑in will quietly increase.
There is also execution risk. Spending tens of billions upfront assumes that AI demand will remain on a near‑vertical trajectory and that AWS can fill new capacity fast enough. If enterprise AI adoption slows, or if open standards make switching providers easier than expected, today’s heroic capex could turn into tomorrow’s overcapacity.
The Bigger Picture
Amazon’s move fits neatly into a pattern we’ve seen across Big Tech: massive, front‑loaded investment to secure a long‑term AI advantage. Microsoft has poured billions into data centres and GPUs tied to its OpenAI partnership. Google has been expanding its TPU‑based infrastructure and rebuilding its cloud around AI‑centric services. Meta has publicly reframed itself as an “AI company”, justifying huge capex to investors on the promise of more efficient ads and new consumer experiences.
What’s different with AWS is the starting point. Unlike Microsoft or Google, whose cloud businesses were still chasing AWS’s market share, Amazon is defending an incumbent position. It is spending not just to grow, but to make sure the next generation of AI workloads doesn’t shift permanently to a rival ecosystem.
Historically, we’ve seen similar cycles. Early AWS growth required heavy capex that temporarily depressed cash flow but ultimately created a near‑unassailable moat. Telecom operators did something comparable in the 3G/4G build‑out, though many misjudged demand and returns. The bet this time is that AI workloads — training, inference, and the new category of always‑on AI agents — will be far more compute‑hungry and sticky than previous waves.
This investment race also has a geopolitical dimension. Cloud infrastructure has become strategic infrastructure. Nations without hyperscale capacity increasingly depend on a small club of US‑based providers for their AI ambitions. Amazon’s willingness to sacrifice short‑term finances to entrench AWS in this role says a lot about where it believes economic power will sit in the 2030s.
The European / Regional Angle
For European companies, AWS’s capex binge is a blessing and a warning. On the positive side, more investment usually means more European regions, better latency, and faster access to specialised AI hardware that would be uneconomical to build locally. For a mid‑sized manufacturer in Bavaria or a fintech in Barcelona, renting GPU‑rich clusters from AWS is the only realistic way to play in the frontier‑model league.
But the dependency worry doesn’t go away — it deepens. The EU is in the middle of implementing the AI Act, on top of existing frameworks like GDPR, the Digital Services Act (DSA) and the Digital Markets Act (DMA). As AI workloads move into US‑controlled hyperscale clouds, European regulators will have to grapple with how to enforce rules on transparency, data protection and systemic risk when the underlying infrastructure is concentrated in a few foreign platforms.
There are regional alternatives — OVHcloud, Scaleway, Deutsche Telekom’s offerings, various GAIA‑X‑aligned initiatives — but none can match Amazon’s AI capex firehose. That raises a strategic question for Europe: double down on regulatory leverage over non‑European providers, or actively co‑fund and protect local champions despite their smaller scale?
For digital‑sovereignty‑minded governments, AWS’s latest numbers will strengthen the argument that Europe needs its own high‑end AI infrastructure, even if it can’t be fully competitive on global price. For startups and enterprises, though, the gravitational pull of AWS’s rapidly expanding AI stack will be hard to resist.
Looking Ahead
Expect Amazon to keep signalling two things at once: discipline on the retail side of the house, and aggressive, almost impatient spending in AWS. As long as AI‑driven growth stays near current levels, management can credibly tell investors that today’s free‑cash‑flow pain is tomorrow’s margin expansion.
Key metrics to watch: the AI revenue run rate within AWS, the utilisation of new data‑centre capacity, and any change in segment operating margins. If AWS can fill new capacity quickly while keeping margins healthy, the market will tolerate heavy capex. If margins compress sharply, the narrative could flip from “strategic investment” to “capital misallocation” surprisingly fast.
For customers, the next 12–24 months will likely bring more AI‑specific pricing tiers, longer‑term GPU reservation contracts and tighter integration between AWS’s own models and customer workloads. The more value‑added AI services you adopt, the harder it becomes to move away later — something CTOs should factor into their risk planning.
Regulators in Brussels, London and national capitals will also sharpen their focus. As AI infrastructure centralises, questions about fair access, interoperability, and systemic dependency will become political issues, not just technical ones.
The wild card is macroeconomics. A meaningful slowdown or a backlash against expensive AI projects that fail to deliver ROI could slow AWS’s growth just as new capacity comes online. That’s unlikely in the immediate term but cannot be ruled out over a five‑year horizon.
The Bottom Line
Amazon is making a conscious trade: weaken free cash flow now to hard‑wire AWS into the foundations of the AI economy. For investors, that’s uncomfortable but rational; for customers and regulators, it’s both an opportunity and a warning about deepening dependence on a single US‑based platform. The real question for Europe and other regions isn’t whether AWS will win big on AI infrastructure — it almost certainly will — but whether they are comfortable letting so much of their future digital capacity sit in Amazon’s hands.
