Markets

Amazon Is Spending Like It’s a Feature, Not a Bug

Date Published

Amazon Is Spending Like It’s a Feature, Not a Bug

TL;DR

Quick Summary

  • Amazon announced a $12B AI data center investment in Louisiana on Feb. 23, 2026, underscoring how aggressively it’s expanding compute capacity.
  • In Q4 2025 (reported Feb. 5, 2026), AWS grew 24% year over year to $35.6B in revenue, keeping Amazon’s profit engine in the spotlight.
  • The tradeoff is cash intensity: trailing twelve-month free cash flow was $11.2B as Amazon ramped property and equipment purchases in 2025.

#RealTalk

Amazon is choosing scale over comfort: it’s spending heavily to make sure it has “enough internet” for the AI era. That can pay off big—but it also tests patience when free cash flow gets squeezed.

Bottom Line

For AMZN, 2026 is shaping up to be a story about whether massive infrastructure spend turns into sticky, high-value demand. If AWS keeps expanding while retail and ads stay resilient, the company’s “spend now, win later” model remains intact—but investors will keep scrutinizing how long the win takes.

The vibe shift: Amazon is building, not chilling

If you’ve been watching Amazon.com, Inc. (AMZN) lately, you’ve probably felt the same whiplash everyone else has: the company posts huge numbers, then immediately turns around and spends like it’s trying to brute-force the future into existence.

On February 23, 2026, Amazon said it plans to invest $12 billion in new AI-focused data center campuses in northwest Louisiana—complete with 540 full-time data center jobs and major local infrastructure spending. The headline reads like a regional economic development press release (because it is), but the subtext is pure Big Tech: compute is the new land grab.

Amazon’s “too much capex” problem is also its whole point

The market’s current obsession is whether Big Tech is overbuilding for AI. That anxiety isn’t imaginary: when you spend tens of billions, you’re basically asking investors to trust that demand will show up on schedule.

Amazon is leaning into that bet. In its February 5, 2026 earnings release covering Q4 2025, the company reported net sales of $213.4 billion for the quarter (up 14% year over year) and full-year 2025 net sales of $716.9 billion (up 12%). The bigger story, though, was Amazon Web Services.

AWS revenue grew 24% year over year in Q4 2025 to $35.6 billion, while AWS operating income was $12.5 billion in the quarter. For full-year 2025, AWS sales were $128.7 billion (up 20%) and AWS operating income was $45.6 billion. In other words: the cloud engine that funds the rest of Amazon is still very much humming.

But the same release also put a spotlight on the tradeoff: trailing twelve-month free cash flow fell to $11.2 billion, with Amazon pointing to a year-over-year increase of $50.7 billion in purchases of property and equipment. Translation: the cash is getting reinvested into the machine.

Why Louisiana matters (even if you never visit)

A $12 billion data center buildout isn’t about “moving closer to customers” the way an old-school retail expansion might be. It’s about getting more chips online, more power contracted, more cooling figured out, and more physical capacity ready for whatever AI workloads look like in 2027 and beyond.

It’s also a reminder that “Amazon” isn’t a single business so much as a stack:

  • Retail trains consumers to start their shopping journey inside Amazon
  • Ads monetize that traffic
  • AWS sells the infrastructure (and increasingly the AI tools) that other companies use to run their own internet lives

When Amazon adds data center capacity, it’s not just growing AWS. It’s reinforcing the company’s ability to be the default platform—whether you’re buying phone chargers, streaming, training models, or running your startup.

The other moonshot people forget to price in

While AI data centers dominate the conversation in early 2026, Amazon is also still pushing into space-based internet. Its satellite program—now branded Amazon Leo—has been ramping launches, and it faces an FCC milestone that requires roughly 1,600 satellites in orbit by the end of July 2026.

That’s ambitious. It’s also very “Amazon”: build the infrastructure first, then fight for the customer relationship later.

So what’s the real question for investors?

The debate isn’t whether Amazon can spend. It obviously can. The question is whether this spending wave translates into durable demand—enough enterprise AI usage, enough cloud workloads, enough services people won’t churn—even when the AI hype cycle inevitably sobers up.

Because if Amazon is right, today’s buildout looks like foresight. If it’s wrong, it looks like expensive optimism with a power bill.