Open-source scheduler turns PyTorch work mappings into IBM Spyre pipelines
A new open-source Dataflow Scheduler converts PyTorch compiler work mappings into execution pipelines for IBM's Spyre accelerator, according to a PyTorch technical post. The scheduler takes KTIR, a representation of which cores pe...
Amazon commits more than €16 billion to German operations in 2027
Amazon announced plans to invest more than €16 billion ($18 billion) in Germany in 2027, including €4.8 billion to expand its AWS data centre region around Frankfurt. The company also plans a second robotics and AI research lab in...
Google makes Gemini 3.5 Flash and B300 compute available on air-gapped cloud
Google says Gemini 3.5 Flash and A4 Ultra machines powered by NVIDIA Blackwell B300 GPUs are now generally available on Google Distributed Cloud air-gapped deployments. The release lets organizations run the model within environme...
Daytona raises $60M Series B at $1B valuation
Daytona raised a $60M Series B led by FirstMark at a $1B valuation, Fortune reported. The company is building cloud infrastructure for AI agents. The round brings Daytona's total funding to $91M, adding capital for its infrastruct...
Modal adopts Nvidia Vera CPUs after reporting twice the throughput in tests
AI cloud provider Modal will add Nvidia Vera CPUs after preliminary internal evaluations showed twice the throughput of its existing AMD and Intel processors, Tom's Hardware reported. Modal will retain its current suppliers and ch...
Google open-sources TorchTPU for native PyTorch workloads on TPUs
Google released TorchTPU as open source on GitHub, saying the stack lets developers train and serve existing PyTorch models on its TPU accelerators with minimal code changes. Co-developed with Meta's PyTorch team, it supports fami...
