Microsoft just outbid everyone at the AI capex table with a $190 billion bet, and Amazon's free cash flow collapsed 95% to make theirs. The competition isn't about who's winning. It's about who can afford to keep playing.
The Summary
- Microsoft announced $190 billion in capital expenditures for this year, the highest disclosed AI infrastructure spend among Big Tech, while Amazon held steady at $200 billion.
- Amazon's AWS posted its best growth since 2022 with 28% sales growth, but free cash flow dropped 95% year-over-year as capex outpaced revenue.
- Amazon signaled it may sell its Trainium AI chips to external customers within two years, a potential new revenue stream that puts it in direct competition with Nvidia.
- The earnings signal where Apple, Nvidia, and other AI infrastructure plays stand as hyperscalers race to build out compute capacity.
The Signal
Four companies worth over $12 trillion reported earnings in a single day, and the story wasn't about profit margins or user growth. It was about who can spend the most money the fastest on AI infrastructure. Microsoft's $190 billion capex plan represents the highest single-year commitment to AI buildout disclosed by any tech giant. That's not a budget. That's a declaration of intent.
Amazon didn't blink at its $200 billion target, but CEO Andy Jassy had to explain why the company's free cash flow cratered. The 95% drop compared to the same quarter last year is the cost of building the compute layer for Web4. Jassy acknowledged the short-term pain, saying it takes time for these investments to manifest value. Translation: we're spending faster than we're earning, and we're fine with that because the alternative is falling behind.
"In times of very high growth, where capex growth meaningfully outpaces revenue growth, free cash flow is challenged."
AWS is the bright spot in Amazon's story. The 28% sales jump last quarter marks its best growth since 2022, proof that enterprise demand for cloud AI infrastructure isn't theoretical anymore. Companies are signing contracts. But the more interesting signal is what Amazon isn't saying yet: it plans to sell Trainium chips externally within two years. That's a direct shot at Nvidia's dominance in AI accelerators. If AWS can supply both the cloud layer and the silicon, it becomes a full-stack AI infrastructure provider.
Key implications across the sector:
- Microsoft and Amazon are now spending $390 billion combined on AI infrastructure this year
- AWS growth acceleration suggests enterprise AI adoption is hitting inflection
- Amazon's chip play threatens Nvidia's pricing power and supply chain control
The earnings create a roadmap for what's next at Apple, Nvidia, and Broadcom. If hyperscalers are spending nearly $400 billion on compute infrastructure, someone has to supply it. Nvidia's next earnings will show whether demand is keeping pace with this capex wave. Broadcom's networking chips become critical bottlenecks. And Apple's AI strategy, whatever it is, needs to justify why it isn't spending at hyperscale levels.
The Implication
The AI infrastructure race is no longer about who has the best models. It's about who can build and operate the most compute at the lowest cost per token. Microsoft and Amazon are betting they can outspend their way to structural advantage. If you're building AI agents or tokenized asset platforms, watch where these companies are placing compute capacity. That's where the Web4 rails are getting laid. And if you're Nvidia, the clock just started on how long you can hold pricing power before hyperscalers verticalize into their own silicon.