TurboQuant compresses the KV cache in AI applications, improving efficiency without sacrificing accuracy. This innovation addresses the challenges of HBM density penalties and DRAM price pressures in the AI memory landscape.

Arm has announced the Arm AGI CPU, a new class of production-ready silicon built on the Arm Neoverse platform, aimed at enhancing AI infrastructure. This marks Arm's first foray into delivering its own silicon products, expanding the options for customers deploying Arm compute.

NVIDIA launched the Vera CPU, which operates with twice the efficiency and 50% faster than traditional CPUs. Collaborating customers include Alibaba, ByteDance, Meta, and Oracle Cloud Infrastructure, with manufacturing partners such as Dell Technologies, HPE, and Lenovo adopting the new CPU.

ByteDance is planning a significant AI hardware deployment in Malaysia, including approximately 500 Nvidia Blackwell computing systems, equivalent to around 36,000 B200 chips. The initiative, facilitated by Aolani Cloud, is expected to exceed $2.5 billion in costs.

Intel's Heracles chip accelerates fully homomorphic encryption (FHE) tasks up to 5,000 times faster than leading Intel server CPUs. Utilizing 3-nanometer FinFET technology and high-bandwidth memory, Heracles enhances efficient encrypted computing at scale, with significant implications for AI and secure data processing.

nCPU is a CPU architecture that operates entirely on GPU, utilizing tensors for registers, memory, flags, and the program counter. All arithmetic operations, including addition, multiplication, bitwise operations, and shifts, are performed through trained neural networks, with specific methods like Kogge-Stone carry-lookahead for addition and learned byte-pair lookup tables for multiplication.

Talos is a custom FPGA-based hardware accelerator designed specifically for executing Convolutional Neural Networks with high efficiency. It reimagines deep learning inference at the circuit level rather than merely reimplementing existing software logic in hardware.

OpenAI plans to release an AI-powered smart speaker in 2027, which will include a camera. The company is also developing smart glasses and a smart lamp with a dedicated team of over 200 employees.

The RTX 5090 delivers 104.8 TFLOPS of FP32 compute and only 1.64 TFLOPS of FP64 compute, resulting in a 64:1 FP64 to FP32 ratio. Over the past fifteen years, this ratio has widened on consumer GPUs, reflecting a growing divide between consumer and enterprise silicon, which is now being challenged by advancements in AI.

Western Digital has sold out its hard drive storage capacity for 2026, with over 10 months remaining in the year. CEO Irving Tan announced that the company is effectively sold out for the calendar year.