The increasing demands of AI are driving innovation in hardware and infrastructure, including specialized chips, advanced memory technologies, and new data center architectures. Companies are investing heavily in these areas to support the growing computational needs of AI models.
Memory manufacturer Kioxia has sold out its projected NAND flash memory production capacity for 2026, driven by strong demand from artificial intelligence applications. The news highlights increasing pressures on the memory supply chain.
AMD replicates Nvidia's playbook by financially backing Crusoe, a provider of energy recovery solutions for data centers. The operation involves a US$300 million loan, signaling a growing interest from chip manufacturers towards energy optimization o...
Weekly updates on the world of artificial intelligence, focusing on the memory impacts in AI systems, the evolutions in the infrastructure needed to support AI workloads, and the policy uncertainties affecting the sector.
Dell introduces a custom 16-pin power connector for Nvidia GPUs in its pre-built PCs, addressing overheating and melting issues found with standard connectors. The solution aims to ensure a more stable and secure connection, reducing the risk of malf...
The Linux kernel is preparing for Intel's next-generation Xeon Diamond Rapids processors as the successor to Granite Rapids. Most of the driver support appears to be settled, with the addition of NTB driver support in the Linux 7.0 kernel.
AMD AOMP 23.0-0 has been released as the latest build of this LLVM/Clang downstream. It includes AMD patches focused on delivering GPU offloading support for Radeon/Instinct hardware with OpenMP and OpenACC APIs. AOMP serves as a leading-edge compile...
The increasing demand for computing power for artificial intelligence is driving the evolution of rack infrastructures. Data centers are preparing to support increasingly intensive workloads, with a focus on energy efficiency and compute density.
Swiss company CLIMATEX AG, specializing in circular textile technologies, has closed a €3.5 million funding round led by the Collateral Good Textile & Fashion Innovation Fund. The funds will be used to strengthen sales and marketing activities and to...
UALink, presented at Hot Chips 2024, is an open standard for interconnects in AI data centers. It aims to reduce vendor lock-in while optimizing costs and performance. The initiative promises greater flexibility and control in the infrastructure for ...
The increasing demand for artificial intelligence infrastructure is fueling a global competition, significantly impacting chip and server suppliers based in Taiwan. The article analyzes how this dynamic is influencing the market and the companies inv...
Major tech companies are increasingly investing in physical infrastructure for generative artificial intelligence. This trend reflects the growing importance of specialized hardware and local stacks to manage AI workloads efficiently and securely.
China’s AI expansion is forcing domestic suppliers to rapidly adopt liquid cooling systems for servers, due to the increasing heat dissipation demands of intensive computing clusters.
In blind tests across six titles, Nvidia's DLSS 4.5 significantly outperformed AMD's FSR and native rendering. The results indicate a strong preference among PC gamers for Nvidia's technology, highlighting the importance of upscaling and frame genera...
Meta plans to deploy standalone Nvidia Grace CPUs in production, with the Vera architecture to follow. The company anticipates up to 2X performance-per-watt improvements in some CPU workloads.
Meta has entered into a multiyear agreement to secure millions of Nvidia GPUs and CPUs, bolstering its artificial intelligence infrastructure. The investment underscores the strategic importance of AI for Meta.
Samsung is developing LPDDR5X-PIM (Processing-in-Memory) to accelerate artificial intelligence workloads. This technology aims to reduce bottlenecks by moving some of the processing directly inside the memory chip, improving energy efficiency and ove...
The increasing demand for artificial intelligence infrastructure in Taiwan is driving a surge in the construction sector and land market. The need for data centers and specialized factories to host AI workloads is growing rapidly, fueling new investm...
Meta is deepening its partnership with Nvidia, integrating Nvidia's Grace processors into its CPU-only systems. The company plans to deploy millions of GPUs over the next few years and is already working on adopting Nvidia's upcoming Vera CPUs starti...
Mesh, a startup founded by former SpaceX employees, has raised $50 million in a Series A funding round. The company aims to mass-produce optical transceivers for AI data centers.
A significant speed increase in prompt processing with llama.cpp and ROCm on Strix Halo (Ryzen AI Max) has been reported. Benchmarks show improvements of up to 132% in some models, opening new perspectives for LLM inference on integrated AMD hardware...
