Higher usage limits for Claude and a compute deal with SpaceXAnthropic News / May 6, 2026Claude Code上限を2倍にClaude Opus APIのレート増加SpaceXで300MW超の容量確保claudeapirate-limitscomputespacexgpudata-residency
Building the compute infrastructure for the Intelligence AgeOpenAI News / Apr 29, 2026Stargate surpassed its 10 GW U.S. goal ahead of scheduleGPT‑5.5 trained at Abilene on OCI using NVIDIA GB200 systemsClosed-loop cooling minimizes ongoing water use; community and workforce investments underwaystargatedatacentergpuinfrastructuregpt-5.5oracle-cloudenergy-efficiency
Unweight: how we compressed an LLM 22% without sacrificing qualityCloudflare / Apr 17, 2026Lossless BF16 exponent Huffman codingReconstructive matmul decompresses in shared memory15–22% model size reduction (~3GB VRAM saved)compressionbf16huffmangpuh100inferencerust