Advancements in semiconductor technology, GPU development, and data center infrastructure that power AI computing capabilities.
Google announced a $40 billion investment in Texas to build three new AI data centers in Armstrong and Haskell counties, expanding its presence in the state and supporting workforce training and energy initiatives.
Anthropic announced a $50 billion investment to build AI data centers in Texas and New York, partnering with Fluidstack to expand computing capacity for its Claude models. The first facilities are expected to go online in 2026.
Baidu has introduced two new AI chips, the M100 and M300, aimed at providing powerful and low-cost computing solutions for Chinese firms. The chips, developed by Baidu’s Kunlunxin unit, are scheduled for release in 2026 and 2027 respectively.
Microsoft has opened its Atlanta Fairwater AI data center, connecting it to a Wisconsin facility via a dedicated AI network to form part of a distributed supercomputer system.
Microsoft has opened its new Fairwater AI datacenter in Atlanta, designed as part of its Azure AI superfactory network. The facility connects to other Fairwater sites and integrates hundreds of thousands of NVIDIA GPUs for large-scale AI training.
Google announced plans to invest €5.5 billion through 2029 in new and expanded data centers and offices across Germany, including a new facility in Dietzenbach and continued development at its Hanau campus.
POET Technologies and Quantum Computing Inc. announced a collaboration to co-develop 3.2Tbps optical engines using thin-film lithium niobate modulators for next-generation AI and data center networks, with completion targeted for 2026.
Alif Semiconductor has announced support for the ExecuTorch Runtime across its Ensemble E4, E6, and E8 microcontrollers, enabling developers to run generative AI models locally on edge devices.
MAINGEAR has announced its aiDAPTIV+ package, co-developed with Phison Electronics, to bring on-premise AI fine-tuning and large-model inference to its Pro RS and Pro WS workstations.
Israeli startup RAAAM Memory Technologies has raised $17.5 million in Series A funding led by NXP Semiconductors to develop its low-power Gain-Cell RAM technology for AI chips.
Video platform and cloud provider Rumble has agreed to acquire German AI infrastructure firm Northern Data in an all-stock transaction valued at approximately $767 million, with the deal expected to close in the second quarter of 2026.
AI startup Tsavorite Scalable Intelligence has received more than $100 million in pre-orders for its Omni Processing Unit (OPU), a chip that integrates CPU, GPU, memory, and connectivity in one architecture to support scalable AI workloads.
The Fidelis Partnership has launched a new consortium to insure risks associated with the construction of AI data centers, establishing an initial capacity of $250 million with plans to expand in 2026.
Calpine Corporation and CyrusOne have finalized the second phase of their power supply agreement for a hyperscale data center in Bosque County, Texas, increasing total capacity to 400 megawatts. The facility, located next to Calpine’s Thad Hill Energy Center, is expected to begin operations by late 2026.
US hyperscalers leased a record 7.4GW of data center capacity in the third quarter of 2025, surpassing the total for all of 2024, according to TD Cowen. Oracle accounted for most of the activity, securing 5.4GW of capacity to support OpenAI workloads.
Scala Data Centers has received approval for the master plan of its Tamboré Campus in Barueri, São Paulo, establishing it as the largest and most advanced data center campus in Latin America with a total capacity of 600 MW.
Digital Realty announced a new phase in its collaboration with NVIDIA to advance AI infrastructure design and deployment at its Manassas, Virginia campus, home to NVIDIA's AI Factory Research Center.
The Open Compute Project Foundation announced new contributions from Arm and Eliyan to its Open Chiplet Economy, introducing the Foundation Chiplet System Architecture and BoW 2.0 memory interconnect enhancements for AI and HPC workloads.
Pure Storage and Cisco have unveiled a new FlashStack Cisco Validated Design integrating NVIDIA AI Enterprise software, unifying compute, storage, and networking to accelerate enterprise AI deployment from pilot to production.
SuperX Digital Power, a joint venture between SuperX AI Technology Limited and Zhonhen Electric, has introduced two 800VDC power systems — Panama and Aurora — designed to enhance efficiency and scalability in AI data centers.
Weekly coverage of AI hardware developments including chips, GPUs, cloud platforms, and data center technology.
