AI
Latest about AI
Chinese AI company's AI model breakthrough highlights limits of US sanctions
By Anton Shilov published
DeepSeek trains DeepSeek-V3 model with 671 billion parameters on a cluster of 2048 GPUs.
Nvidia's next-gen B300 GPUs have 1,400W TDP, 50% more AI horsepower
By Anton Shilov published
With 288 GB of HBM3E memory and 50% more FLOPS than B200, Nvidia's B300 is a major performance upgrade.
Russian firm starts shipments of HPC system based on homegrown CPU
By Anton Shilov published
Russian company Graviton launches first AI and HPC machine with homegrown CPUs that will have to be used with Nvidia's H100.
Elon Musk's xAI raises $6 billion to build more powerful AI supercomputers
By Anton Shilov published
xAI, which is set to build a supercomputer with 200,000 Nvidia GPUs, raises $6 billion.
US gov't set to ban Huawei intermediary Sophgo over AI chip supplies — partnership skirted US chip sanctions
By Anton Shilov published
U.S. to ban Sophgo, which supplied AI chiplets to Huawei, but its affiliate Bitmain avoids sanctions.
Nvidia may postpone volume ramp-up of Blackwell machines: TrendForce [Updated]
By Anton Shilov published
Nvidia may have to delay the volume ramp of Blackwell machines due to overheating and interconnect issues.
Elon Musk reportedly wanted OpenAI to be a for-profit entity but has now sued to block the move
By Anton Shilov published
OpenAI fires back at Elon Musk, saying he advocated transforming organization into a for-profit company for personal financial gain.
Intel tempers expectations for next-gen Falcon Shores AI GPU — Gaudi 3 missed AI wave, Falcon will require fast iterations to be competitive
By Anton Shilov published
The interim CEO says that Intel's Falcon Shores will be the company's first step toward developing AI GPUs.
Synopsys announces Ultra Ethernet and UALink IP to power the next-generation of AI datacenters
By Anton Shilov published
Synopsys introduces Ultra Ethernet IP solution for 1.6T Ethernet networks, UALink IP solution for linking up to 1,024 accelerators per pod.
AMD nearly beats 30X power efficiency goal a year early — AMD's new AI servers are 28.3 times more efficient than 2020 versions
By Anton Shilov last updated
AMD's lofty goal of improved energy efficiency for AI sees it within touching distance of 30 times better performance than a machine from 2020
Stay On the Cutting Edge: Get the Tom's Hardware Newsletter
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.