The morning after Amazon corporate employees went on a bender sending emails to Chinese suppliers to cancel orders that were already being manufactured… CEO Andy Jassy rolled into HQ with a 4,000-word shareholder letter locked and loaded… ready to explain why Amazon is preparing to drop $100 billion on artificial intelligence. (Which, for context, is more than NASA’s entire budget. Or, if you prefer your comparisons geopolitical… about the same as Slovakia’s GDP.)

If that number sounds outrageous, Jassy doesn’t just want to justify it… he's calling it essential. In his annual letter, he framed AI as a once-in-a-generation transformation. A shift so massive, it will reshape customer experiences across the board: how we shop, how we code, how we get healthcare, how we entertain ourselves, and yes, how Alexa responds when you ask it for the weather (again).

“If your mission is to make customers’ lives better and easier every day, and you believe every customer experience will be reinvented by AI, you're going to invest deeply and broadly in AI,” Jassy wrote.

He’s not joking. Amazon’s set to spend $25 billion more than Google, $50 billion more than Microsoft, and more than twice the amount that Meta’s planning on spending. The company is pouring capital into massive new data centers, designing custom silicon like Trainium2, and ramping up its AI offerings inside AWS (Amazon’s real profit engine).

To really understand why Jassy is swinging so hard at AI, you have to understand what AWS did for Amazon. Back in 2006, AWS started as a side project to rent out Amazon’s excess computing power. Nearly two decades later, it’s become the backbone of the internet and one of the most profitable business lines in all of tech. In 2024, AWS generated $108 billion in revenue (up 19% year over year) and accounted for most of Amazon’s operating income.

While Amazon’s retail business runs on razor-thin margins, AWS consistently operates above 25%, printing billions in quarterly profit that bankroll everything from Prime Video to one-day delivery to risky side projects. Jassy sees AI following the same trajectory… spend heavily now on infrastructure, then lease it out at scale.

And that infrastructure buildout is already well underway. Amazon’s homegrown Trainium and Inferentia chips are built to cut AI costs by as much as 40% over traditional GPUs… a huge deal considering that inference (the part where AI actually does something useful) is about to become one of the biggest cost drivers in tech. To sit atop the full stack, Amazon is also rolling out its own large language models under the “Nova” brand, and offering them through a platform called Bedrock… essentially a digital spread where developers can choose from different AI models.

Even Alexa is getting a facelift. Powered by Claude (the chatbot built by Anthropic, which Amazon invested $8 billion into) the new “Alexa+” is supposed to remember your preferences, take real-world actions, and stop sounding like it’s stuck in 2015 (in theory).

But while the AI spending has reached interstellar levels, the rest of the company is on a pretty strict diet. Since 2022, Amazon has laid off over 27,000 employees. Several experimental projects (like a TikTok-style video feed and a “Try Before You Buy” clothing service) have been shelved. And in a move straight out of a startup playbook, Jassy even launched a “red tape inbox,” asking employees to flag unnecessary bureaucracy. Over 1,000 submissions later, Amazon made 375 changes based on the feedback.

If you believe they can make their AI ambitions strike like AWS, then look out, with the tariffs weighing the stock down 19% since february (even counting yesterday’s 11% jump)... it could be a big buying opportunity.

Stock.News has positions in Amazon, Microsoft, Meta, and Google.