Nvidia is opening its NVLink technology to work with Arm-based custom chips from hyperscalers like Microsoft, Amazon and Google. The move comes after Nvidia’s failed $40 billion Arm acquisition in 2022 and follows a $5 billion investment in Intel. This represents a strategic shift toward platform op
Intel is pushing early Linux graphics support for its upcoming Nova Lake processors while addressing security vulnerabilities. The company published 30 new security advisories alongside microcode updates for current CPUs.
Chinese tech giant Baidu is making a major play in the AI hardware space with custom chips designed to compete with Nvidia. The company plans rack-scale systems with 256 and 512 accelerators starting in 2026, alongside a new training chip for trillion-parameter models.
The Netherlands is sending officials to China early next week to resolve the escalating Nexperia dispute. The conflict involves export controls on chips crucial to the automotive industry. Both sides are seeking concessions in this geopolitical tech battle.
Tower Semiconductor is expanding its proven 300mm wafer bonding technology beyond image sensors to silicon photonics and SiGe BiCMOS processes. The move enables heterogeneous 3D-IC integration for data center applications. Cadence design tools now support the multi-technology platform.
Samsung’s upcoming Exynos 1680 chipset for the Galaxy A57 has appeared on Geekbench with surprising specifications. The chip features AMD’s latest RDNA 3.5 graphics architecture, which could significantly boost gaming performance in mid-range devices.
AI is fundamentally reshaping chip design, with advanced packaging becoming critical for performance. Foundries like TSMC, Intel and Samsung are expanding into packaging while traditional assemblers scale up. The technology could triple package sizes by 2030 to accommodate more memory and computing
The Chinese government is reportedly intervening in how SMIC allocates its chip production, prioritizing Huawei amid ongoing US export restrictions. This represents a significant escalation in the tech cold war between the two superpowers.
NVIDIA’s Jensen Huang admits it would be “foolish” to underestimate Huawei’s rapid AI advancements. The Chinese tech giant plans competitive AI chips through 2027 and claims its Atlas SuperPoDs can match NVIDIA’s Rubin performance. With NVIDIA locked out of China, Huawei is gaining ground fast.
SoftBank has completely exited its Nvidia position, selling 32.1 million shares for $5.83 billion. The Japanese conglomerate is redirecting these funds toward what it calls a “large” investment in OpenAI, potentially exceeding $30 billion.
