Amazon Just Took a Swing at Nvidia's Real Moat: CUDA

Everyone thinks Nvidia's dominance is about chips. It's not — it's about software.

On Monday, Amazon's chip unit Annapurna Labs introduced TorchNeuron, a free and open-source tool that lets developers run code written in PyTorch on Amazon's Trainium AI chips with a single configuration change and no rewriting, according to Reuters. Amazon plans to release it publicly as open source.

Why This Is a Direct Shot at CUDA

PyTorch is the framework most AI developers build on, and Nvidia spent more than a decade making its CUDA software the best way to run PyTorch code on its hardware. That software lock-in — not just the silicon — is what keeps most teams on Nvidia chips.

TorchNeuron is Amazon's attempt to break that habit. "This has been about meeting customers where they're at," Colin Brace, vice president of software for Annapurna Labs, told Reuters. In other words: no rewrites, no relearning, no excuses not to try Trainium.

Who's Already Using It

Amazon says more than 30 early customers — spanning robotics, world models, and voice AI — are already using the software, and that switching to its chips can result in steep cost savings.

That number is worth watching rather than taking at face value. But if even a fraction of large inference workloads move, it proves the pitch: Amazon isn't trying to out-chip Nvidia, it's trying to make switching cheap.

It's Not Just Amazon

Reuters notes that other cloud players with their own custom AI chips — Google included — are also working to lure developers away from Nvidia with better PyTorch tooling. The pattern is clear: nobody thinks they can beat Nvidia on raw hardware alone, so everyone's attacking the software moat instead.

Why It Matters

AI computing costs are the quiet tax on every AI product you use. When switching chips gets easier, cloud providers have to compete harder — and cheaper compute eventually flows down to the tools, apps, and AI AI agents we all pay for.

The AI chip war just became a software war. That's the one worth watching.