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 Riot Platforms has signed a 15-year lease agreement with an artificial intelligence lab for capacity at its data center in Corsicana, Texas. Riot Platforms states that the 910MW contract is valued at $28 billion and could expand to $54 billion if extension options are exercised.
AI-generated summary Digital Realty launched a liquid cooling lab in São Paulo, Brazil, to help customers test and validate multi-vendor artificial intelligence infrastructure. The facility is designed for high-density artificial intelligence workloads and features liquid-cooled racks supporting cabinet densities of 150kW and above.
AI-generated summary Photonic chip company Q.ANT has partnered with Ionos to launch a pilot program that lets participants test AI models on Q.ANT hardware. Approved applicants can access the second-generation Native Processing Unit via the Ionos website to run their own workloads.
AI-generated summary Running local AI models requires varying amounts of hardware memory and processing power depending on model size. OpenAI says its gpt-oss-20b model runs with 16 GB of memory, while larger models need significantly more resources.
AI-generated summary A lawyer built a private artificial intelligence server using eight used Nvidia V100 graphics processing units to keep case files offline. The system runs the DeepSeek V4.1 Flash model at up to 215 tokens per second while costing about $2,400 per four-board setup.
AI-generated summary AMD announced an expanded collaboration with Google Cloud utilizing 5th Gen AMD EPYC processors. AMD says these processors power host infrastructure for Google Cloud TPU 8tx and help protect sensitive AI models.
AI-generated summary CoreWeave announced a multi-year agreement with Wayve to power its artificial intelligence driver training using CoreWeave Cloud with NVIDIA Blackwell GPUs. Wayve says the cloud infrastructure helps accelerate the commercial deployment of its artificial intelligence driver by providing reliability and observability.
AI-generated summary Google Cloud says dynamic slicing is generally available on Google Kubernetes Engine to run large-scale AI workloads on TPU infrastructure. Google Cloud says the feature reduces AI workload startup times by up to 5x and workload recovery by 4.5x.
AI-generated summary The Nvidia GeForce RTX 3090 offers 24 GB of VRAM, making it a capable hardware choice for running local large language models and other AI workloads. While the Nvidia GeForce RTX 3090 is no longer in production, it provides an alternative for users seeking memory capacity comparable to newer cards like the RTX 5070 Ti, RTX 5080, and RTX 5090.
AI-generated summary Alphabet Inc.'s Google is releasing new versions of its custom tensor processing units to compete with Nvidia Corp. in the artificial intelligence market. The latest chips are designed to support confidential computing and help models access a wider range of information for handling user queries.
AI-generated summary Microsoft is reportedly exiting a 2023 compute capacity agreement with Oracle that utilized older Nvidia hardware to power an AI-assisted version of the Bing search engine. According to Bloomberg, Microsoft declined to renew the contract because the older hardware did not meet its current specifications.
AI-generated summary InfoQ reports on optimizing distributed AI inference using llm-d and vLLM on Kubernetes to manage cache locality and batch inference. Edoardo Vacchi discusses running Java workloads on GPUs using TornadoVM and the evolving landscape of WebAssembly on the backend.
AI-generated summary Amazon introduced TorchNeuron, a native PyTorch backend that allows developers to run existing code on AWS Trainium chips without modifications. Amazon says a single configuration change switches hardware, and the backend supports distributed training and PyTorch parallelism APIs.
AI-generated summary Israel plans to build its own artificial intelligence infrastructure to reduce its dependence on foreign technology companies. The strategy includes establishing a base of around 300,000 graphics processing units and developing a domestic foundation model.
AI-generated summary The author argues that Apple designed the 2,899 dollar MacBook PROLED as a luxury device with a unified memory bottleneck rather than a computer built for local artificial intelligence. The reviewer suggests that power users focused on local workloads should instead consider Windows alternatives featuring Nvidia RTX Spark and AMD Strix Halo chips.
AI-generated summary SpaceX plans to build a twenty-four billion dollar data center and natural gas power plant in Starbase, Texas. The proposed facility will reportedly operate with thirty-five turbines to produce seven hundred forty-eight megawatts of power.
AI-generated summary NVIDIA announced a collaboration with Rebellions and SK Telecom to connect Rebellions processors to NVLink Fusion for Korean sovereign AI infrastructure. The resulting systems from NVIDIA, Rebellions, and SK Telecom are expected to be ready for deployment in 2028.
AI-generated summary NVIDIA released Dynamo-optimized deployment recipes for open models including A.X K2, K-EXAONE 2.0, Solar Open 2, and Motif-3. NVIDIA says the recipes optimized for B200 GPUs deliver up to two times performance and help developers bring open models into production faster.
The top AI news in your email delivered each morning. Choose categories