Google has announced the AI Hypercomputer, which brings together TPUv8 series, NVIDIA Rubin, & Axion CPUs to power the Agentic AI era. Google Cloud Next 26: AI Hypercomputer Announcement Gives Agentic AI The Next Push, Leverages In-House TPUs, CPUs & Scales Beyond With NVIDIA Rubin Gone are the days of supercomputers; the Agentic AI era will be all about hypercomputers, which will combine various compute options to deliver customers the most flexible and performant AI architecture ever built. Today, at Google's Cloud Next 26 event, the company formally announced its AI Hypercomputer. The new high-performance computing datacenter for Agentic AI […]

Read full article at https://wccftech.com/google-unveils-the-heart-of-agentic-ai-the-ai-hypercomputer-8th-gen-tpus-nvidia-rubin-axion-cpus/

Samsung's 1d DRAM (7th Gen 10nm) for next-generation HBM solutions might not undergo production soon due to failure to meet yields. Samsung's Next-Gen DRAM Tech For Future HBM5E Memory May Not Be Ready For Production A report published by Korean outlet, IT Chosun, suggests that due to less than ideal yields of its 1d "D1d" DRAM based on the 10nm process technology, Samsung might be pulling the plug on mass producing its next-gen HBM solutions. The DRAM technology had already received a pre-production approval (PRA), but concerns have been raised regarding the ROI of initiating a trial run, let alone […]

Read full article at https://wccftech.com/samsung-postpones-hbm5e-memory-production-indefinitely-d1d-dram-yields-fall-short/

NVIDIA CEO has said that while their company produces expensive AI hardware, they also produce the lowest cost tokens in the world. NVIDIA's Leadership in AI Is Not Only Because of Hardware, But It's full-stack approach that makes the "Lowest Cost Token" Possible Talking at Cadence Live 2026, NVIDIA CEO stated that they are the leaders of low-cost tokens because they produce the world's lowest-cost tokens. A token is the fundamental unit of AI; think of them like the ABCD of the AI language that AI models process to generate responses. The speed at which tokens are generated depends upon […]

Read full article at https://wccftech.com/i-produce-the-lowest-cost-tokens-in-the-world-says-nvidia-ceo/

Nscale will be adding 30,000 NVIDIA Rubin GPUs in addition to the 100,000 chips that are already being deployed at its facilities in 2027. Nscale Scales Up Its AI Datacenters With 30,000 Additional NVIDIA Rubin GPUs, Bringing The Total Count To 130,000 NVIDIA's Vera Rubin AI chips are hot in demand, and the reason behind this is simple: they are aiming to be the fastest solution for AI inferencing, fueling the race towards Agentic AI. Josh Payne, Founder and CEO at Nscale said: “Customer demand for advanced AI infrastructure continues to accelerate across markets, and our focus is on bringing the […]

Read full article at https://wccftech.com/130000-nvidia-rubin-gpus-deployed-at-nscale-for-microsoft-showing-massive-interest-in-next-gen-ai-chips/

China's domestic memory manufacturers are rushing to produce HBM chips, but the biggest DRAM maker has run into roadblocks with HBM3 tech. CXMT Is Reportedly Struggling With Its HBM3 Development & Has Delayed The Project To 2H 2026, A Cause of Concern for China's Domestic AI Markets The AI market in China is booming at an incredible pace, led by Huawei and several other chipmakers. Despite facing strict bans on advanced chip production, the AI market has endured, & domestic DRAM makers are now sampling their very first HBM3 solutions, which will be coupled with the latest AI chips. At […]

Read full article at https://wccftech.com/china-ymtc-hits-hbm3-roadblock-domestic-makers-rush-hbm-manufacturing/

The CEO of Qualcomm has landed in Korea to discuss crucial plans with the executives at Samsung and SK Hynix. Qualcomm CEO's Visit To Korea Can Be A Push To Secure Chip & Memory Capacity at Samsung & SK Hynix As supply constraints continue to grip the AI and PC markets, chipmakers are rushing to secure supply across the globe. Reports are emerging from Korean outlets that Qualcomm's CEO, Cristiano Amon, has visited the country today (April 21st), and is likely to hold meetings with executives from Samsung Foundry and SK Hynix. The main agenda that is likely to be […]

Read full article at https://wccftech.com/qualcomm-ceo-flies-to-korea-hunting-2nm-wafers-at-samsung-lpddr-supply-at-sk-hynix/

AMD will be leveraging GlobalFoundries for the development of its MRM Co-packaged Optic solution for the next-gen Instinct MI500 AI accelerators. GlobalFoundries & AMD Working Together on Co-Packaged Optics Hardware For Instinct MI500 Accelerators CPO or Co-Packaged Optics (Silicon Photonics) is the next-generation solution that reduces reliance on copper and harnesses light to transfer signals. These CPOs are packaged alongside hardware accelerators such as GPUs and will be a key solution for next-gen AI factories, offering improved interconnect latency and creating high-bandwidth connections between CPU and GPU. Both AMD and NVIDIA will be leveraging these technologies for their next-gen AI […]

Read full article at https://wccftech.com/amd-taps-globalfoundries-for-mi500-co-packaged-optics/

SK Hynix is now mass-producing SOCAMM2 memory with up to 192 GB capacities for NVIDIA & next-gen AI Data Centers. Prepped For NVIDIA's Vera Rubin & AI Data Centers, SK Hynix Begins Mass Production of SOCAMM2 Memory With 192 GB Capacities At CES 2026, SK Hynix announced that it had delivered its next-gen memory solutions, such as SOCAMM2, to NVIDIA for its upcoming AI data center solutions. Now, four months later, SK Hynix is announcing that it has begun mass production of the memory, and it will be utilized by NVIDIA's next-gen platform for AI. NVIDIA will be utilizing SOCAMM2 […]

Read full article at https://wccftech.com/sk-hynix-mass-produces-192-gb-socamm2-memory-for-nvidia-vera-rubi-ai-datacenters/

Samsung will be rolling out a new HBM memory design every new year instead of two as it tries to catch up with rising AI demands. AI Super Cycle Pushes Samsung To Move From 2-Year HBM Memory Development Cycle To 1-Year HBM is an integral part of various accelerators that are powering the AI ecosystem. Samsung is one of the vital players in the HBM and DRAM segment, & aims to shorten its development cycles drastically to align itself with the growth trajectory in the AI segment, reports Korean outlet Busan. For years, Samsung has been developing new HBM standards […]

Read full article at https://wccftech.com/samsung-slashes-hbm-development-cycle-from-2-years-to-1-betting-future-on-ai/

Recent chatter suggests that AMD has secured a major AI customer deal with its upcoming Instinct MI450 GPU accelerators. Supply Constraints Push Anthropic To Sign A Major Deal With AMD, Will Be Leveraging Next-Gen Instinct MI450 GPUs Given the tightness surrounding the industry's supply chain, major AI players are looking to fulfill their infrastructure requirements from various firms. AMD is aware of this opportunity and has already hinted at multiple "Open-AI" scale customers that are lined up for its existing and upcoming Instinct AI accelerators. The company has OpenAI & META already onboard, with the latter signing a 6 Gigawatt […]

Read full article at https://wccftech.com/amd-instinct-mi450-secures-major-ai-customer-win/

AMD confirms that its 6th Gen EPYC Verano CPUs will be the first to utilize SOCAMM2 memory, further expanding its perf/W AI solutions. AMD Confirms SOCAMM2 Memory Support With 6th Gen EPYC Verano CPUs, Tightening The DRAM Supply Chain Even More Last year, AMD confirmed its next-gen EPYC Verano CPUs for AI at rack scale. These chips will be coupled with the company's next-gen MI500 accelerators and will come together using Vulcano interconnect. At the time, AMD didn't disclose much, and the first impression was that Verano would likely use Zen 7 cores. But that isn't the case, as AMD […]

Read full article at https://wccftech.com/amd-epyc-verano-cpus-lpddr5x-socamm2-memory-perf-watt-optimized-ai-rack/

NVIDIA has officially announced its GTC Taipei 2026 keynote, where CEO Jensen Huang will take the stage to unveil next-gen AI breakthroughs. NVIDIA To Host Its Own Special GTC Taipei Keynote Ahead of Computex, Will Focus on AI, But Consumer Updates Also On The Table Well, NVIDIA has finally confirmed that it's coming to Taipei in June, and will be hosting a special GTC Taipei keynote just like last year, hosted by CEO Jensen Huang. NVIDIA's appearance in Taipei during Computex 2026 was never in doubt, but we were waiting for an official confirmation, & now that has finally happened. […]

Read full article at https://wccftech.com/nvidia-ceo-jensen-huang-to-host-gtc-taipei-keynote-ahead-of-computex-2026/