Qatar Helium Shutdown: Two-Week Clock for Chip Fabs

Shared entity: GPU / Same source domain / What happened next / Tension

Intel Arc Pro B70: 32GB VRAM for $949.

Both cover GPU; reported by the same outlet (tomshardware.com); picks up the GPU thread on 2026-04-21.

Shared entity: GPU / Same source domain / What happened next

GLM-5.2 is the hedge, and the benchmarks say it's a real one

Both cover GPU; reported by the same outlet (tomshardware.com); picks up the GPU thread on 2026-06-21.

Shared entity: HBM / Shared topic / What happened next

Context Drought: 1M Token Windows for 5–10 Years.

Both cover HBM; overlapping topics (demand, production); picks up the HBM thread on 2026-03-15.

Shared entity: HBM / Shared topic / Earlier coverage

Meta's 4-Generation Custom Silicon Roadmap.

Both cover HBM; overlapping topics (chip, compute); earlier HBM coverage from 2026-03-12.

Shared entity: GPU / What happened next / Tension

GPU vector indexing is now the default in OpenSearch Serverless, 10x faster at a quarter the cost.

Both cover GPU; picks up the GPU thread on 2026-06-25; pushes against this story (versus).

Gate speculative decoding on batch size or it'll make you slower.

Both cover GPU; picks up the GPU thread on 2026-06-24; pushes against this story (but).

Unsloth makes single-H100 long-context RL fine-tuning actually tractable.

Both cover GPU; picks up the GPU thread on 2026-06-20; pushes against this story (versus).

LLM-generated GPU kernels pass weak benchmarks and ship real bugs.

Both cover GPU; picks up the GPU thread on 2026-06-20; pushes against this story (against).