AI Infrastructure SoftwareGoogle (TensorFlow)Google

Google releases XLA

Released XLA — JIT-compiles computation graphs to fuse operations and generate native code, for up to 50% speedups on GPUs

Why it matters

The starting point of the ML compiler race. The layer that made TPUs practical and the foundation of all of JAX

Source — Primary sources first; secondary reports are labelled “Press”

Officialdevelopers.googleblog.comhttps://developers.googleblog.com/2017/03/xla-tensorflow-compiled.html

Added in Week 35, 2026 · See that week

Same week

Finn, Abbeel and Levine publish Model-Agnostic Meta-Learning (MAML) on arXivPaper

Nearby on the same axis

NVIDIA presents Mixed Precision TrainingPaperNVIDIA releases a preview of CUTLASSOfficialJAX goes open sourceOfficialNVIDIA releases NCCLOfficial