News

Toto is best known for toilets and bathroom products, but the Japanese company is now getting attention for a very different reason: AI chips. The company’s shares reportedly jumped 18% after it announced plans to increase production of semiconductor components, including parts used in AI chip manufacturing. The link may sound strange at first, but it comes down to ceramics. Toto’s ceramics expertise has

Samsung’s chip making business may be getting a boost from the same AI memory wave that is pressuring the rest of the industry. A new report says Samsung Foundry is seeing strong demand for its 4nm process, with orders extending into 2027. The report says Samsung’s 4nm process has become more stable, which appears to be helping the company win more orders from global

AMD appears to be working on a new Ryzen PRO desktop CPU that could bring 16 cores and 3D V Cache to its business focused lineup for the first time. The chip, listed as Ryzen 9 PRO 9965X3D, has appeared in a PassMark entry, though the details should still be treated as early and unconfirmed. Ryzen PRO chips are made for professional systems, office

Micron says the AI boom is still in its early stage, and memory demand may keep rising as inference workloads grow. CEO Sanjay Mehrotra said AI is in the “first innings,” with faster and denser memory becoming a strategic requirement for companies building large AI systems. The reason is simple. AI models need compute, but that compute depends heavily on memory. GPUs need HBM,

ASUS marketed its ROG Equalizer 12V 2x6 cable as a better solution for balancing power across high end GPUs, but new testing from Der8auer suggests the cable may perform worse than expected. The test used an RTX 5090 and WireView Pro II to monitor current distribution across the connector pins. Instead of showing cleaner balance, the ROG Equalizer showed uneven current loads. In some

xAI may have one of the largest AI GPU fleets in the industry, but a new report claims it is using only a small share of that hardware effectively. The company reportedly has around 550,000 NVIDIA H100 and H200 GPUs across its Memphis and Colossus clusters, but its current utilization is said to be only around 11%. That would mean xAI is effectively using

Intel’s Arc Pro B70 has been tested in games, and the results give a useful look at what a gaming focused Big Battlemage graphics card might have delivered. The Arc Pro B70 is not sold as a normal gaming GPU. It is a professional card with 32GB of GDDR6 memory, a full BMG G31 Battlemage chip, 32 Xe2 HPG cores, 32 ray tracing units,

Intel and SoftBank are working on a new memory technology called ZAM, or Z Angle Memory, and the early claims are ambitious. The goal is to offer a high bandwidth, high capacity, lower power memory option for AI accelerators, with up to twice the bandwidth of HBM4. HBM is currently the key memory technology for high end AI GPUs, but it has limits. As

AMD’s next high end Halo APU has appeared in leaked benchmark results, and it looks like a clear step above the current Ryzen AI MAX Plus 395. The chip is listed as Ryzen AI MAX Plus PRO 495, part of AMD’s upcoming Gorgon Halo family. The leak shows 16 Zen 5 CPU cores, 32 threads, Radeon 8065S graphics, and support for up to 192GB

NVIDIA’s RTX 5050 has finally shown up in Steam’s hardware survey, months after its mid 2025 launch. That means every RTX 50 series GPU now appears in Steam’s database, although the RTX 5050 is still one of the weakest performers in terms of adoption. The desktop RTX 5050 currently sits at only 0.17% share. Its laptop version is more popular, which makes sense because

NVIDIA is reportedly moving faster than expected on co packaged optics, with the technology now planned for its Feynman data center GPUs in 2028. Earlier timelines had placed broader commercialization closer to 2033, so this would bring the shift forward by about five years. Co packaged optics, also called CPO or silicon photonics, replaces more copper based data movement with optical links. The goal

Anthropic is reportedly in early talks with UK startup Fractile, a company working on a new chip architecture designed to speed up AI inference while reducing cost. The talks are still early, and Fractile has not yet produced test chips, so this should be treated as a developing story rather than a finished product. Fractile’s technology is called Memory Compute Fusion Architecture. The basic

AMD’s upcoming Ryzen AI 5 435G desktop APU has appeared in a leaked benchmark, and the early result is interesting. Even while running below its listed boost clock, the chip performed close to the Ryzen 5 8600G in Geekbench. The Ryzen AI 5 435G is part of AMD’s coming Ryzen AI 400 desktop APU lineup for AM5. These chips are expected to bring newer

ASUS is reportedly changing how it allocates its RTX 50 series graphics cards. Instead of pushing as many RTX 5070 Ti models as before, the company may reduce supply and move more production focus toward the RTX 5080. The reason appears to be simple. Both the RTX 5070 Ti and RTX 5080 use 16GB of VRAM, but the RTX 5080 sells at a higher

Xbox is launching a new developer focused video series called Game Dev Update, and the first episode will include more discussion about Project Helix, Microsoft’s next generation Xbox hardware plan. The show is aimed at developers, but many Xbox fans will likely pay attention because Project Helix remains one of the biggest questions around the platform’s future. The first episode will premiere on May

Pearl Abyss has released Crimson Desert update 1.5, and it adds one of the game’s most requested features: boss rematches. After crossing 5 million sales, the studio is clearly still investing in the game with new systems, fixes, and performance features. The update is available across platforms, except the Mac App Store version, which will receive it later. The new Rematch system turns bosses

Intel and AMD are working on a new x86 extension called APX, or Advanced Performance Extensions. It is part of a wider effort by the x86 Ecosystem Advisory Group to modernize x86 while keeping compatibility with older software. APX is different from flashier AI focused updates like ACE. Its goal is more general. It helps normal CPU workloads run faster and more efficiently by

AI memory demand is getting worse, and the pressure may soon move beyond GPUs. A new report says future AI focused CPUs could be equipped with 300GB to 400GB of memory, a major jump from today’s common 96GB to 256GB range per chip. The reason is agentic AI. These systems do more than answer one prompt. They plan, call tools, manage longer context, and

Tenstorrent is making big claims for its new Galaxy Blackhole AI servers. The company says the system can deliver very high inference performance while lowering total cost compared with current GPU based platforms. The hardware is built around Tenstorrent’s Blackhole chips, which use a RISC V based design. Inside each chip are Tensix cores with programmable RISC processors, matrix units, vector units, local SRAM,

TSMC is preparing to show more details about its A16 process at the 2026 VLSI Symposium, and the node looks like an important step beyond 2nm. A16 is part of TSMC’s Angstrom era roadmap and is expected to bring better performance, lower power use, and backside power delivery. The biggest change is TSMC’s Super Power Rail technology. This is the company’s version of backside