AMD has begun production of Helios, an AI server rack competing with Nvidia's data center systems, with Samsung Electronics supplying the high-bandwidth memory. AMD CEO Lisa Su announced the milestone…
AMD and Cerebras Systems announced a partnership to build a disaggregated AI inference system combining AMD's Helios rackscale solution with Cerebras's Wafer-Scale Engine, claiming up to 5x higher tok…
AMD announced the Instinct MI455X, its first GPU designed for rack-scale AI deployments, based on the new CDNA5 architecture and featuring 256 Work Group Processors, 432 GB of HBM4 memory at 23.3 TB/s…
AMD launched the Instinct MI400 Series GPUs for AI workloads, including the MI455X for frontier AI and the MI430X for sovereign AI and HPC. Built on CDNA 5 architecture and a 2nm process, the MI455X f…
Samsung Electronics is forecasting a 19-fold increase in operating profit for Q2 2026 compared to the same quarter a year earlier, driven by a projected tripling of high bandwidth memory (HBM) sales. …
NVIDIA detailed its new "Rubin" GPU architecture, which delivers 50 PetaFLOPS at sparse NVFP4 operations, features 224 streaming multiprocessors, 288 GB of HBM4 memory, 336 billion transistors, and 89…
NVIDIA's Rubin GPU, the core of the Vera Rubin platform, delivers up to 10x more agentic throughput per unit of energy than the Blackwell architecture, according to NVIDIA. The GPU features 336 billio…
AMD announced Helios, a single rack containing 72 Instinct MI455X GPUs, 31 terabytes of HBM4 memory, and 2.9 exaflops of FP4 inference compute, as its direct answer to Nvidia's Vera Rubin NVL72. Engin…
Researchers at UNIST and Hanbat National University in South Korea, and engineers in Japan, presented alternative memory chip designs at the IEEE VLSI Symposium that stack DRAM dies side by side inste…
High Bandwidth Flash (HBF), a NAND-based memory stack with read bandwidth comparable to HBM4 but 10x the capacity, is set to sample from Sandisk in late 2026 and appear in AI inference devices by earl…
Samsung Electronics reported a second-quarter operating profit of 89.4 trillion won ($58.4bn), a 19-fold increase year-over-year, driven by surging demand for AI memory chips. Revenue more than double…
UBS advises investors to buy SK Hynix's upcoming American Depositary Receipts on Nasdaq and sell its Korean-listed shares ahead of a historic $29 billion ADR launch on July 10, 2026. The Swiss bank ex…
PhantaField's Sophon PFG-1 AI ASIC integrates 330 GB of on-die DRAM using a monolithic 3D architecture, eliminating the need for high-bandwidth memory (HBM) to overcome the memory wall in large langua…
Sophon PFG-1, a monolithic-3D AI ASIC with 330 GB of on-die DRAM and no HBM, delivers 2,100 TFLOPS BF16 and 4,200 TFLOPS FP8 on a 750 mm² die, achieving up to 174× the tokens-per-watt of NVIDIA Rubin …
Micron Technology's HBM4 memory chip is ramping production at twice the speed of its predecessor HBM3E, with full allocation for 2026. The company's fiscal Q3 2026 results crushed Wall Street estimate…
Micron Technology reported fiscal Q3 2026 revenue of $41.46 billion, crushing Wall Street estimates by nearly $6 billion, driven by AI-fueled demand for memory chips. Adjusted EPS of $25.11 also beat …
Micron Technology reported fiscal third-quarter revenue of nearly $42bn, quadrupling year-over-year, driven by surging demand for high-bandwidth memory used in AI accelerators. Gross margins exceeded …
SK Hynix briefly surpassed Samsung Electronics to become South Korea's most valuable company, reaching a $1.35 trillion market cap on June 22, 2026, driven by surging demand for its high-bandwidth mem…
Samsung Electronics reviewed strategies to expand sales of high-bandwidth memory (HBM) chips and discussed long-term supply agreements with major customers at a global strategy meeting, amid surging d…
Nvidia's Vera Rubin NVL72, announced at CES and detailed at GTC, promises up to 10x lower cost per token than Blackwell, but the headline figure depends on ideal conditions: mixture-of-experts models,…