Amazon Trims Its AI Superteam Even as It Spends Billions Chasing AGI

Amazon confirmed Wednesday that it's cutting jobs inside its artificial general intelligence unit, the internal group tasked with building AI systems capable of matching or beating human performance across most tasks. The company hasn't said how many people are losing their jobs or which parts of the sprawling AGI organization — which also houses silicon development and quantum computing work — are taking the hit.

Reuters broke the news first. Amazon's official line, delivered through a spokesperson, leaned on familiar corporate language about speed and focus: "This is a fast-moving space, and we're sharpening our focus on the initiatives that matter most for customers, so we can move faster on what counts." The statement went on to acknowledge that means "eliminating some roles" even as investment continues elsewhere in the unit.

That framing is doing a lot of work to soften what is, in practice, another round of cuts inside a company that has laid off more than 30,000 employees since last October. Amazon has been trimming headcount steadily since its pandemic-era hiring surge gave way to a leaner, AI-first posture, and these cuts land squarely inside the part of the business meant to define its AI future.

A team under pressure to catch up

The AGI unit isn't some side project. It's Amazon's answer to OpenAI, Anthropic and Google in the race to build frontier AI models — systems that generalize across tasks rather than just excelling at narrow ones. In 2024, the group shipped Nova, its family of foundation models. Leadership of the unit changed hands last December, when longtime cloud executive Peter DeSantis took over from Rohit Prasad and was given a broader mandate.

The unit has also seen turnover at senior levels. David Luan, who joined Amazon in 2024 when the company acquired his startup Adept in what's known as an acquihire, left as head of Amazon's AGI lab in February. Losing a lab chief that quickly, followed months later by layoffs across the group, paints a picture of an organization still finding its footing.

DeSantis himself has been candid about where Amazon stands. In an interview with CNBC last month, he acknowledged that Amazon's models "haven't been at the very frontier for the very largest, most demanding workloads" — a notable admission given how much money and attention the company has thrown at the effort. He said Amazon is working to close that gap and wants to field one of the "most capable intelligent models out there."

Amazon's spokesperson reinforced that the company still sees this as central to its future, saying it has been building large models for years and that the work "remains one of the most important things we're working on."

Cutting costs while spending like never before

What makes this round of layoffs notable isn't just where they're happening, it's the contrast with how much Amazon is spending on AI overall. The company has forecast capital expenditures of $200 billion for the year, more than 50% higher than 2025, and is taking on tens of billions of dollars in debt to help finance that buildout. Data centers, custom chips and AI infrastructure are absorbing enormous sums even as the team building the actual models gets smaller.

That's not necessarily a contradiction. Big tech companies have increasingly separated infrastructure investment from headcount, arguing that AI tools let smaller teams do more with less. But for a unit whose entire purpose is to keep Amazon competitive against rivals that are moving fast and spending heavily on talent, losing people right as the pressure to catch up intensifies is a hard needle to thread.

Amazon reports second-quarter earnings next week, which should offer a clearer picture of how the company is balancing its AI ambitions against its bottom line, and whether investors are satisfied with a strategy that pairs record capital spending with continued job cuts in the very group meant to deliver on that spending.

What happens next inside the AGI unit will say a lot about Amazon's real position in the AI race. DeSantis has set an explicit goal of reaching frontier-level performance. Doing that with a smaller team, while competitors like OpenAI, Anthropic and Google keep expanding their own research staffs, will test whether Amazon's approach of throwing capital at infrastructure while trimming the humans behind the models actually works.