Google is developing a specialized server chip, internally codenamed "Frozen v2," designed to embed the structural blueprint of its Gemini AI model directly into silicon. By hard-wiring the model’s core architectural logic into the hardware, the chip aims to eliminate the compute overhead and data movement required by general-purpose accelerators. This approach is projected to deliver six to ten times the token-per-watt efficiency of Google’s current Tensor Processing Unit (TPU) generation.

Source

Nvidia is aggressively expanding its product portfolio beyond its dominant GPU lineup by introducing the Vera CPU. This strategic shift aims to position the company as a provider of complete, vertically integrated AI data center systems. By offering a unified platform, Nvidia seeks to control the orchestration, networking, and data-flow tasks typically managed by third-party processors.

Source

Google is reportedly developing a server chip called Frozen v2 that would embed parts of Gemini’s architecture directly into silicon. Unlike the original Frozen concept, which would have locked Gemini’s weights into the chip, this version would keep the weights updateable while hardwiring the architecture instead. That approach is meant to preserve usefulness across future Gemini releases, as long as Google keeps the same underlying design.

Source

Microsoft is rolling out Windows 11 build 26200.8942/26100.8942 as KB5101684 in the Release Preview channel ahead of the August 2026 security update. The preview covers Windows 11 24H2 and 25H2 and is being phased in rather than delivered to everyone at once. Alongside a long list of fixes, Microsoft is focusing on quality-of-life improvements rather than major platform changes.

Source

AMD’s Helios is the company’s first rack-scale AI system, built to challenge Nvidia’s dominant Vera Rubin and Grace Blackwell platforms. The system combines 72 GPUs and 18 CPUs per rack, uses liquid cooling, and is designed to compete on performance, efficiency, customization, and total cost of ownership. Early customers including Microsoft, Meta, OpenAI, and Oracle have already committed to deployments.

Source

Microsoft will deploy AMD's Helios rack-scale AI platform on Azure beginning in the second half of 2026 to run large AI models for its own use, customers, and Azure AI services. Helios combines AMD GPUs, CPUs, networking chips, and software as a full-system alternative to Nvidia's dominant rack-scale offerings. AMD CEO Lisa Su called the expanded deal a major milestone, while Microsoft CEO Satya Nadella stressed that customers need more choices in AI infrastructure.

Source

Japan is establishing a state-funded AI factory to serve as the primary compute infrastructure for the national FRONTia program. The facility will utilize 382 Vera Rubin NVL72 racks, integrating 27,500 Nvidia Rubin GPUs and 13,750 Vera CPUs. This massive deployment is designed to draw 140 megawatts of power, placing it among the largest single-site AI training clusters globally.

Source

Microsoft will deploy AMD’s Helios rack-scale AI platform at scale on Azure to handle frontier-model inference for its own services, Azure AI offerings, and customer workloads. Helios integrates 72 Instinct MI455X GPUs providing 31.1 TB of HBM4 memory, up to 1.4 exaFLOPS of FP8 and 2.9 exaFLOPS of FP4 compute, plus 260 TB/s scale-up and 43 TB/s scale-out bandwidth via UALink over Ethernet. The open rack-scale design also incorporates sixth-generation EPYC “Venice” CPUs, Pensando DPUs, and ROCm software for large-scale training and inference.

Source

Microsoft released out-of-band update KB5121767 (OS build 26200.8894) for Windows 11 to resolve compatibility problems with the Intel Innovation Platform Framework driver on a limited set of Dell systems. The Intel IPF driver bundle includes Dynamic Tuning Technology for thermal and power management plus Context Sensing Technology for presence-based features such as Wake on Approach. Recent cumulative updates including the July 2026 security release KB5101650 triggered performance changes, unexpected shutdowns, overheating, elevated power use, and battery drain on the affected hardware, prompting Microsoft to apply a temporary safeguard hold.

Source

Microsoft has temporarily paused Secure Boot certificate updates for Windows 11 on affected devices after the new 2023 certificates failed to install reliably on many systems. The update replaces expired 2011 certificates that underpin Secure Boot, a UEFI feature that verifies only trusted software loads before the OS starts. Failures stem mainly from outdated or faulty UEFI firmware on older PCs that no longer receive OEM updates, leaving Microsoft unable to supply a universal fix.

Source

Nvidia has begun releasing Windows 11 Arm64 GPU drivers for its N1X processor used in upcoming RTX Spark systems, with the initial version 616.00 package optimized specifically for the Surface RTX Dev Box rather than general consumer hardware. The native Arm64 installer includes the nv_surface_woa.inf file that defines supported devices, listing two RTX Spark N1X GPU variants based on Blackwell architecture with 6144 and 5120 CUDA cores plus an unidentified desktop device entry. The same package also registers NVIDIA NPU components via Deep Learning Accelerator device IDs, confirming integrated AI acceleration support in the platform.

Source

Microsoft and major hardware manufacturers recently held a technical session to address challenges regarding the transition from 2011 to 2023 Secure Boot certificates. Engineers confirmed that while some certificates have already expired, the final compliance deadline is October 19, 2026, after which unpatched devices may face boot failures. Administrators were assured that eligible devices, including those currently offline or in storage, will remain capable of receiving the new certificate chain indefinitely via Windows Update or manual firmware packages.

Source

Microsoft has halted the deployment of the July 2026 mandatory security update, KB5101650, for specific Dell computer models. The pause stems from a compatibility conflict between a new Windows USB-C Connection Manager interface and Intel Innovation Platform Framework (IPF) drivers. Affected devices include various Dell XPS, Precision, and Pro Max models that rely on these drivers for power management and thermal regulation.

Source

OpenAI has launched its first physical hardware product, the Codex Micro, a specialized macropad designed to control AI-driven coding workflows. Developed in partnership with boutique manufacturer Work Louder, the $230 device is a modified version of the existing Creator Micro 2 platform. It features 13 low-profile mechanical keys, a rotary dial, and a joystick housed within a CNC-machined aluminum and polycarbonate body.

Source

A regression in the open-source AMDGPU driver is causing significant performance degradation across several major Linux distributions, including Ubuntu, Fedora, and Arch. The issue stems from a bug in the Linux 7.0 kernel series that mismanages GPU power profiles and clock settings. This flaw can allow GPU core frequencies to exceed manufacturer limits, triggering the kernel to throttle the device to prevent damage.

Source

IBM has introduced the Power S1112, a compact one-socket server designed for edge computing and distributed environments. Built on the Power11 architecture, the entry-level system features on-chip Matrix Math Acceleration to support local AI inference and analytics. The hardware is available in rack or tower configurations and allows administrators to run Linux or AIX partitions alongside traditional IBM i environments.

Source

OpenAI is developing a screenless, mobile smart speaker designed to function as a human-like AI companion for the home. The device features mechanical elements that allow it to move autonomously, creating a more lifelike presence than traditional smart speakers. It utilizes integrated cameras and sensors to maintain contextual awareness of its surroundings and the user's activities.

Source

Apple is exploring a partnership with PrismML, a Caltech spinout, to integrate extreme model-compression technology into future iPhone hardware. The startup utilizes ultra-low-bit weight architectures, such as 1-bit and ternary representations, to drastically reduce the memory footprint of large language models. This mathematical approach allows a 27-billion-parameter model like Alibaba’s Qwen 3.6 to shrink from 54 GB to under 4 GB, making it compatible with the 8 GB of RAM found in current Pro-tier iPhones.

Source

Nvidia is transitioning from a GPU manufacturer to a comprehensive AI infrastructure provider by launching the Vera CPU platform. This new processor is specifically engineered to handle agentic AI systems and intensive data center workloads. By moving beyond graphics accelerators, the company aims to capture a larger share of the traditional server market currently held by x86 competitors.

Source