AI Weekly Malaysia

Google open-sourced TorchTPU, a PyTorch version that lets AI apps run on TPUs and move to Google Cloud with a configuration change rather than a rewrite; Amazon released TorchNeuron for Trainium/AWS a day earlier and is submitting PyTorch code to decouple features from Nvidia chips. PyTorch, originally built by Meta in 2016 and moved to the Linux Foundation in 2022, is used by some 85% of AI apps, according to the article. Google says it worked with Meta, Hugging Face, vLLM maker Inferact, and SGLang maker RadixArk on TorchTPU.

Why: If the 'no rewriting' claim holds for your model, TPU and Trainium become a second deployment target and a negotiating lever on GPU cloud pricing; if it does not, you will burn time on portability work. Before committing, ask for benchmarks on your exact model and confirm the vLLM/SGLang/Hugging Face paths you rely on are supported, because the article gives no performance or price numbers.

Google's google-pytorch org has published torch_tpu, an open-source PyTorch backend that runs PyTorch models on TPUs via custom ATen kernels, an XLA compilation cache, torch.compile() graph-mode integration, and distributed training support (data and tensor parallelism). The repo ships examples for Llama, Qwen, ResNet, DLRM, minGPT and UNet, plus Marimo notebooks that the README says are easiest to run on a v6e GCE instance. It has ~502 stars, 25 forks and 4,470 commits, and drew 354 points / 74 comments on Hacker News.

Why: If your training stack is PyTorch and you have (or can get) GCP TPU quota, this is the first-party path to keep your model code and torch.compile() workflow instead of rewriting in JAX — check the ATen kernel coverage for your ops and the compile-cache behaviour before assuming parity, and budget for v6e quota rather than assuming a drop-in pip install.

Tom's Hardware's headline states that Google TPUs running AI workloads are now deployed in Meta's data centers, framed as the first time TPUs have been sold to external customers and as a challenge to Nvidia's market dominance. The article body as retrieved contains only site navigation, membership prompts, and newsletter sign-up boilerplate — no pricing, capacity, contract terms, deployment timeline, or named sources.

Why: There is nothing to act on yet: without a price, volume, or availability date, you cannot plan to buy TPU capacity outside Google Cloud or re-cost inference against Nvidia GPUs. If you are setting a 2026 compute budget, treat the headline as an unconfirmed claim and wait for terms before changing any procurement or architecture assumption.