Energy consumption in large language models extends beyond AI chips, with Alphabet Inc.'s recent research revealing that measuring only the chip underestimates the total environmental...
Taiwan Semiconductor Manufacturing Company (TSMC) plans to begin volume production of its 2nm chips in the fourth quarter of 2025, despite foundry prices soaring to a record US$30,000...
At the OCP APAC Summit, Ethernet and UALink advocates clashed over data center interconnect architectures. Kurtis Bowman, UALink board chair and AMD's director of architecture and...
Saudi Arabia's artificial intelligence startup Mumain has started building its first data centers in Riyadh and Dammam, with operations expected to commence in early 2026, reports...
Investment bank Jefferies Group LLC has warned in a memo that China has imposed a sweeping ban on Nvidia's H20 AI GPUs, blocking domestic companies from placing any new orders for...
Malaysia has launched its first locally designed edge AI processor, the MARS1000, developed by SkyeChip Sdn Bhd. Unveiled at the Malaysia Semiconductor Industry Association Merdeka...
The H20 chip faces resistance from the Chinese government, with reports indicating that Nvidia has requested its supply chain partners to halt production of the H20. However, Nvidia...
DeepSeek has introduced its new large language model, DeepSeek-V3.1, featuring a key upgrade: the "UE8M0 FP8 Scale" precision format. The company emphasized that the format was tailored...
The spotlight will be on Nvidia as it prepares to release its financial results for the second quarter of fiscal year 2026 (2QFY26), with the earnings announcement scheduled for August...
Samsung Electronics and SK Hynix are reportedly pulling back on investments in advanced NAND flash memory as weak demand and rising costs push them to concentrate on higher-margin...
Chennai-based Mindgrove Technologies is charting a path from prototype to mass production of secure IoT chips, betting on cost efficiency, regulatory compliance, and deep partnerships...
China's push to challenge Nvidia in AI chips is accelerating, with Baidu's Kunlunxin winning a US$140 million order from China Mobile. Alongside Huawei's Ascend processors and Cambricon's...
Nvidia executives have reportedly urged Samsung Electronics to prioritize production of sixth-generation high-bandwidth memory (HBM4) rather than fifth-generation 12-layer HBM3 during...