Industry
Semiconductors, datacenter infrastructure, and AI compute bottlenecks.
Monitor
Intel reported robust Q1 results driven by AI demand in data centers, sending shares soaring in after-hours trading.
Google announced new Trillium-based TPUs optimized for agentic AI, expanding beyond Nvidia with improved efficiency for training and inference. Part of Google Cloud Next announcements.
NVIDIA's GB200 NVL72 systems enable dramatic efficiency gains for OpenAI's new model, scaling agentic AI.
OpenAI's upgraded Codex, now running GPT-5.5, leverages NVIDIA's latest GB200 rack-scale AI infrastructure for superior coding agents.
SpaceX is developing its own GPUs to address substantial supply constraints in AI compute hardware.
AI models like Morpheus are processing massive datasets from telescopes like James Webb, driving up GPU demand and straining academic resources amid NSF budget cuts.
35K Samsung workers plan 18-day strike over bonuses amid AI-driven HBM demand surge, where data centers gobble 70% of high-end memory.
Tesla revealed in its Q1 10-Q a deal to acquire an unnamed AI hardware firm for up to $2B in stock and equity, with most value tied to deployment milestones. Not mentioned in earnings call or letter.
Google unveiled TPU 8t for training and TPU 8i for inference at Cloud Next '26, aiming to scale AI infrastructure amid surging demand.
At Cloud Next '26, Google launched TPU v8t for training (scales to 9,600 chips, 3x prior perf) and v8i for inference (low-latency multi-agent), integrated with Axion CPUs and liquid cooling for better price/performance.
Researchers unveil a neuromorphic chip mimicking brain efficiency, potentially reducing AI energy consumption by 70% while enhancing adaptability. The breakthrough targets sustainable scaling of AI compute.
TSMC launches A13 process node optimized for AI accelerators and mobile SoCs, promising higher performance and efficiency. Positions TSMC to capture more AI chip demand.
SK Hynix posted a fivefold profit surge to 37.6 trillion won ($25B), driven by high-bandwidth memory (HBM) shortages for AI chips, with operating margin at 71.5%.