South Korea's KOSPI closed down 0.44% at 6,008.81 on Wednesday, its sixth consecutive decline, after early gains tied to New York Fed comments were erased by ASML's weaker 2026 sales outlook, according to Yonhap.
MemSilicon says it has completed tape-out of MountTai 2026, a 1Gb RRAM chip using a 1T4R cell structure on a 28nm CMOS platform, with 30Mb/mm² density and sub-20ns access latency, and is now verifying the part.
SK hynix expects the performance scale of AI infrastructure to grow about 20-fold by 2030 as AI inference demand rises, an executive said at SEDEX 2026 in Seoul.
Seoul shares opened higher Wednesday as tech stocks rebounded, despite Wall Street losses and lingering concerns over AI spending. The KOSPI rose 1.83% by 9:15 a.m.
TechRadar published a headline describing Minisforum's newest AI workstation as having 16 cores, 5.2 GHz, 55 TOPS and a 192GB memory upgrade. The supplied article text included no further product details, only an author biography.
AI Chips & Compute · Products & Applications · Companies
At the Open Compute Project Global Summit, Intel argued that inference and open models will define enterprise AI economics, not the largest training clusters. The company detailed its GPU D-series and high-density Xeon rack designs for agentic workloads.
Companies · AI Chips & Compute · Products & Applications
Tinygrad has introduced a redesigned Tinybox workstation for deep learning, with a G4 configuration that supports four AMD Radeon RX 9070 XT GPUs and starts at $11,000, according to TechRadar.
Nvidia is to make billion-dollar investments in European technology, with CEO Jensen Huang saying the chipmaker has committed up to $10 billion to the region's tech sector over the past year.
Engadget explains how NVIDIA DLSS renders games at lower internal resolutions and uses machine learning to upscale them, improving frame rates while trading some image fidelity. The article details DLSS modes, support history and DLSS 4 changes.
Ubitium raises $14M in additional seed funding led by Ibex Investors, bringing total seed investment to $18M to commercialize a universal RISC-V processor for physical AI machines.
Satlyt, founded by former Google and SpaceX product manager Rama Afullo, has raised $8 million to build software that runs AI models on satellites. Its software is set to launch on a SpaceX rocket alongside Google's Project Suncatcher prototype.
Microsoft introduced the Azure Maia 300 at the OCP Global Summit 2026, claiming 40% more tokens per megawatt than leading GPU-based systems as it seeks to cut its reliance on Nvidia and AMD.
AI Chips & Compute · Products & Applications · Companies
ChangXin Memory Technologies said its new architecture can pack more data into a smaller space, triggering a selloff in global memory-chip stocks and raising questions about competition with Micron, SK Hynix and Samsung.