Infrastructure and hardware comprise the physical compute systems, specialized chips, and data center components that power artificial intelligence workloads. Truncated covers chip architectures, supply chains, energy requirements, and the hardware deployments that enable large scale computing.
AI-generated summary ZincFive states that its nickel-zinc technology brings high power capability and usable energy storage together as rack power requirements rise in data centers. ZincFive says its nickel-zinc batteries use a nonflammable, aqueous electrolyte and have no risk of thermal runaway at the cell level.
AI-generated summary Acurast says it has onboarded more than 335,000 smartphones to its network, with around 40,000 fully active on an average day. A Google-backed project at the University of California, San Diego found that 20 retired smartphones handled certain workloads.
AI-generated summary The RTX PRO 5500 Blackwell has been listed by SHI Government Solutions with an MSRP of $17,999 and a discounted price of $17,408.63. This retail pricing exceeds the $16,000 cost of the higher-end RTX PRO 6000 Blackwell Workstation Edition listed on NVIDIA’s US Marketplace.
AI-generated summary About 600,000 Nvidia H200 chips remain stored in Hong Kong because Beijing has not approved large-scale imports, according to a Reuters report. CNBC reported that Moonshot AI has sought to rent Nvidia Blackwell chips in Europe, while Washington rules restrict direct export to China.
AI-generated summary Analysts say the real test for the artificial intelligence build-out will happen in 2028. Democratic gubernatorial winners who pledged to restrict data centers include Gina Hinojosa in Texas, Rob Sand in Iowa and Keisha Lance Bottoms in Georgia.
AI-generated summary About 600000 Nvidia artificial intelligence chips are stored in Hong Kong waiting for shipment to mainland China according to a Reuters report citing people familiar with the matter. Beijing has not approved large-scale imports of the H200 chips despite demand from technology companies while several major Chinese internet companies have each received about 10000 H200 chips in recent months.
AI-generated summary Huawei states that the next stage of enterprise artificial intelligence will depend less on purchasing graphics processing units and more on rebuilding data infrastructure. Huawei introduced the OceanStor A800 and OceanStor Pacific products to serve as an artificial intelligence data platform and data lake for knowledge and memory generation.
AI-generated summary DeepSeek V4.1 Flash 1017 matches GLM-5.3 intelligence at about one third the cost and four times the speed according to Baseten. Baseten details three optimizations used to make Flash models faster including kernel fusion Programmatic Dependent Launch and host loop minimization.
AI-generated summary Microsoft decided not to double its compute capacity agreement with Nscale in Texas, according to reporting from the Financial Times. The decision involved capacity at an Nscale-owned site adjacent to its Ionic Digital data center in Barstow, Texas.
AI-generated summary China has implemented export controls on critical metals that have disrupted supplies of micro drill bits for the artificial intelligence chip supply chain. Nikkei Asia reports that these restrictions demonstrate how geopolitical tensions are affecting the global technology industry.
AI-generated summary Octomind states that Flash Attention computes attention results in tiles without keeping the full score matrix in GPU memory. Octomind notes that the method does not reduce model weights or the persistent KV cache.
AI-generated summary IREN shares dropped 18 percent Thursday after a report detailed Microsoft raising concerns regarding data center delays and execution risks. Microsoft officials met with IREN last week to negotiate a revised timeline that could delay completion by several weeks or months.
AI-generated summary AMD Quark uses mixed-precision quantization to evaluate transformer blocks for video generation on AMD Instinct GPUs. The calibration policy promotes sensitive blocks to W4A8 while keeping others on the W4A4 path.
AI-generated summary Figma says its users are moving faster with AI tools like the agent and Figma Make, which require secure sandboxes with stable infrastructure to run AI-generated code quickly. Figma says it built a centralized platform called Nimbus to manage sandboxes at scale across multiple cloud providers and handle resource allocation and network security.
AI-generated summary AMD announced day-zero support and optimization for Kimi K3.1 Pro on AMD Instinct MI355X GPUs using the MXFP4 checkpoint and DSpark speculative decoding. AMD says the implementation combines memory-efficient sparse attention with decode context parallelism for long-context serving and low-latency interactive workloads.
AI-generated summary Blackfuel is emerging from stealth with more than $250 million in contracted revenue under multi-year customer agreements to build an inference-native neocloud. The company plans an inference-only deployment in Barcelona combining AMD Instinct MI355X GPUs and AMD ROCm open software with Dell, NTT DATA and Digital Realty.
AI-generated summary According to Wasabi Technologies, the Wasabi 2026 Global Cloud Storage Index found that UK organizations allocate 62 percent of their AI budgets to data, storage, and compute. The research from Wasabi Technologies also indicates that 48 percent of UK respondents exceeded their cloud storage budgets last year.
AI-generated summary The author benchmarked a local AI model on an RTX 4070 Ti Super PC using llama-bench and logged power draw with HWiNFO and nvidia-smi to calculate electricity costs. The measurements showed that running the Qwen3.6-35B-A3B model locally costs less than a single monthly subscription for ChatGPT, Claude, or Gemini.
AI-generated summary Elon Musk posted on X that data centers in space are looking appealing following recent political developments. Democratic election victories raised concerns about potential tighter restrictions on data centre construction in the country.
AI-generated summary PsiQuantum is sharing optical circuit switch samples with prospective customers as it adapts silicon photonics developed for quantum computing to the networks connecting AI infrastructure. The company reported typical insertion loss of 1.2 decibels, crosstalk performance of 40 decibels, and reconfiguration in less than three milliseconds in its own laboratory tests.
The top AI news in your email delivered each morning. Choose categories