Elon Musk's Terafab semiconductor project could cost $5 trillion, Bernstein claims — herculean effort would cost more than 70% of the total yearly US government budget By Anton Shilov published 26 March 26 But for TeraFab, money is not the only limitation.
Price of Nvidia's Vera Rubin NVL72 racks skyrockets to as much as $8.8 million apiece, but server makers' margins will be tight — Nvidia is moving closer to shipping entire full-scale systems By Anton Shilov published 24 March 26 While quotes for NVL72 VR200 rack-scale system will increase to $5 million – $7 million, margins of ODMs will likely decline.
AMD's Enterprise CPU and GPU roadmap: Venice, Verano, Zen 6, Helios, and CDNA By Anton Shilov published 23 March 26 Premium We take an in-depth look at AMD's 2026 and 2027 roadmaps.
Planned 10-gigawatt Softbank data center in Ohio might be the largest in the world — will require a $33 billion natural gas plant, equivalent to nine nuclear reactors By Anton Shilov published 20 March 26 SoftBank is looking forward to building a $60 billion - $70 billion AI data center spanning multiple phases in Ohio.
China develops new ultra-cold alloy that can reach -273°C without helium — could enable compact cooling for superconducting quantum chips, military equipment, and beyond By Anton Shilov published 20 March 26 Chinese scientists develop europium, cobalt, and aluminum alloy that can be used to cool down quantum computers and military equipment.
Tesla hiring semiconductor fabs construction manager — Elon Musk's ambitious Terafab project begins By Anton Shilov published 19 March 26 Tesla begins to form a core team behind its Terafab project, which will define the fab and spearhead its construction and tamp.
Researchers reach superconductivity at ambient pressure, record high temperature — milestone of -122°C reached by using pressure quenching, still 140 degrees off room temperature target By Anton Shilov published 19 March 26 A step closer to practical applications, but still 140°C away the target.
AMD and Samsung ink memory supply memorandum for EPYC and Instinct products — unprecedented deal also includes scope for foundry partnership By Anton Shilov published 19 March 26 Samsung to remain primary HBM memory supplier for AMD's AI accelerators as the companies look into possible foundry relationship.
Nvidia will only produce one 88-core Vera CPU model — Jensen says the company will make billions of dollars from a single SKU By Anton Shilov published 18 March 26 Nvidia intends to build a multi-billion-dollar CPU business with its Vera processors, but only plans to offer one SKU.
Nvidia updates data center roadmap with Rosa CPU and stacked Feynman GPUs — optical NVLink, Groq LPUs with NVFP4, and NVLink also on deck By Anton Shilov published 17 March 26 Nvidia publishes 2026 – 2028 data center roadmap with Rosa CPU, Feynman GPU, optical NVLinks and Groq LPUs with NVFP4 and NVLink.
Nvidia demonstrates Rubin Ultra tray, the world's first AI GPU with 1TB of HBM4E memory — new chips will slot into Kyber racks By Anton Shilov published 17 March 26 Nvidia shows off its next-generation Kyber rack-scale solution to be powered by Rubin Ultra GPUs with four compute chiplets and 1 TB of HBM4E memory per package.
Nvidia removes Rubin CPX accelerators from its roadmap — Groq 3 LPUs take center stage as CPX is removed By Anton Shilov published 17 March 26 Nvidia's slides at GTC lack any mentions of Rubin CPX, but praise Groq LPUs instead.
Jensen Huang expects Nvidia to sell $1 trillion of AI hardware through 2027 — AI buildout intensifies as Agentic AI takes hold By Anton Shilov published 16 March 26 Nvidia believes it earn $1 trillion by selling AI hardware, says co-founder and chief executive Jensen Huang.
STMicro to deploy humanoid robots to its legacy fabs in Europe — over 100 humanoid robots to be used for routine and physically demanding tasks in fight for efficiency By Anton Shilov published 16 March 26 As cheap Chinese chips threaten to invade various sectors of the economy in Europe, STMicroelectronics is on track to improve the efficiency of its own fabs
Elon Musk says his chipmaking 'Terafab Project' venture will launch in seven days — Musk's latest moonshot multi-billion project launches on a Saturday By Anton Shilov published 15 March 26 Elon Musk plans to start his 'Terafab Project' semiconductor production venture in a week.
US gov't revokes controversial AI hardware export rule that would mandate investments from foreign companies — new export rules are still in the works, though By Anton Shilov published 14 March 26 AMD, Nvidia, other suppliers of AI accelerators may now celebrate.
11-month old Russian outfit claims it has developed 16-core and 32-core chips, flaunts Cyrillic-badged processors — chips appear to be sanctions-swerving rebadged Chinese Loongson processors By Anton Shilov last updated 13 March 26 Russia-based Tramplin Electronics obtains samples of Loongson's LS3C6000 processors with Cyrillic inscriptions, claims these are its own CPUs.
China's ByteDance to access 36,000 Blackwell GPU cluster through Malaysia cloud operator — Nvidia confirms no objections, deal is in line with US export controls By Anton Shilov published 13 March 26 Nvidia says it is permissible for ByteDance to use AI clusters outside of China to develop its AI prowess as long as these clusters are built in compliance with the U.S. export controls.
2026 PC shipment forecast slashed amid memory shortages — IDC says total PC market value to nonetheless increase to $274 billion due to ongoing price hikes By Anton Shilov published 13 March 26 IDC expects unit shipments of PCs in 2026 to be down 11.3% year-over-year, but the whole market value will increase 1.6% due to higher prices.
AMD, Broadcom, and Nvidia join hyperscalers to define optical scale-up interconnect of the future for AI clusters — Meta, Microsoft, and OpenAI to benefit as speeds eventually scale to 3.2 Tb/s By Anton Shilov published 12 March 26 AMD, Broadcom, and Nvidia team up with Microsoft, Meta, and OpenAI to develop protocol-agnostic optical scale-up interconnects for AI clusters.
Intel's Heracles chip computes fully-encrypted data without decrypting it — chip is 1,074 to 5,547 times faster than a 24-core Intel Xeon in FHE math operations By Anton Shilov published 11 March 26 No decryption occurs inside the processor, eliminating entire classes of attacks.
Intel Panther Lake-H high-res die shot emerges — image show 18A compute tile, Xe3 GPU tile in new X-series processors By Anton Shilov published 9 March 26 An enthusiast blogger published annotated die shots of Intel Panther Lake-H CPU: 16-core mobile processor with 12 Xe3 clusters and two Thunderbolt 5 ports examined.
OpenAI's massive Stargate data center canceled as firm can't reach terms with Oracle, operator struggles with reliability issues — Meta said to be interested in snatching excess capacity By Anton Shilov published 8 March 26 Financing terms and swinging OpenAI capacity forecast collapse a major expansion of a Stargate data center. Yet, Meta could take the yet-to-be-expanded space.
Engineer receives $30,000 for exposing a vulnerability affecting 7,000 robot vacuum cleaners — tinkerer just wanted to drive his robot vacuum with a PS5 controller By Anton Shilov published 7 March 26 $4.2 per unit? Or something to say about the reputation?
Nvidia dominates gaming GPU market with 95 percent share as sales of AMD Radeon graphics plummet to a historical low of 5 percent By Anton Shilov published 7 March 26 Nvidia strengthens its position on the market of discrete GPUs as AMD's market share continues to dive, whereas Intel fails to gain traction.
