Nvidia

Nvidia released Nemotron-Personas-Finance-Japan on Hugging Face, a synthetic persona dataset modeling Japan's demographics and financial context. Positioned as a privacy-conscious foundation, it targets teams building AI applications in Japan and developers pursuing regional deployment.

At CoRL, NVIDIA introduced an open robotics stack spanning world action models, simulation tools, benchmarks and agentic workflows, shifting robot learning from locomotion toward delicate manipulation such as gear alignment and cable connection. Two NVIDIA papers, GRAIL and RoboTTT, were named among six Best Paper finalists from thousands of submissions, while Ambi Robotics, Bosch Research and Siemens are already exploring its agentic robotics workflows in industrial settings.

Google has added two variants to its eighth-generation TPU lineup: TPU 8tx, which keeps confidential data encrypted during processing, and TPU 8i Max, which doubles memory and compute per rack by pairing two chips in a single package. Alongside software such as TorchTPU, Google is lowering the barrier for developers and marketing its in-house silicon to external customers for everyday inference.

Google is expanding its custom TPU lineup with two new chips: TPU 8tx for confidential computing and TPU 8i Max for ultra-long-context AI workloads, bringing the family to four products. Both are reconfigured rather than fully redesigned silicon, with 8tx designed to work alongside AMD chips. Google's chip chief said iteration has accelerated well beyond the previous two-year cadence as the company seeks to capitalize on Nvidia's supply constraints.

NVIDIA and South Korea's Ministry of Science and ICT hosted an AI startup partnership forum at COEX in Seoul, where NVIDIA announced an investment in AI interconnect firm Point2 and plans for additional investments in Motif Technologies, RLWRLD and Upstage. The four targets span AI interconnect, agent foundation models, robotics foundation models and enterprise LLMs, signaling NVIDIA's capital and compute ties across Korea's AI ecosystem.

Nvidia held its AI Day in Seoul, unveiling partnerships with South Korea's MSIT and companies including Samsung, LG Electronics, Hyundai Motor, Kakao Mobility and SK Telecom. The deals span sovereign AI infrastructure, accelerated semiconductor R&D, robotics, autonomous driving and inference optimization for Korean foundation models, but most remain exploratory or demo-stage with no concrete metrics disclosed.

Nvidia's board has launched an independent investigation into CEO Jensen Huang after he told staff he was unfaithful to his wife; the other party, a Megaspeed executive, is suspected by U.S. officials of being a Chinese spy. The stakes extend well beyond a personal scandal, since Huang says he is "responsible for an entire industry" and the outcome will shape how employees, customers, partners and ordinary shareholders view the AI economy.

NVIDIA is leveraging Korea's manufacturing and engineering strengths to advance the country's physical AI ecosystem. Samsung Electronics is expanding its collaboration with NVIDIA to accelerate semiconductor materials, process and device engineering via accelerated computing and AI physics, while robotics and autonomous vehicle efforts extend the same stack toward production. The piece is largely a vendor-driven ecosystem narrative without verifiable parameters or quantified results.

NVIDIA is expanding its agentic AI footprint in Korea: Hyundai Motor Group is exploring NemoClaw blueprints and the OpenShell runtime, while Korean AI security firm AIM Intelligence is integrating its Starfort agent-control platform with OpenShell to combine sandboxed execution and policy-based controls with detection of malicious instructions and data leakage. On the model side, LG AI Research is opening API access to K-EXAONE-2.5, trained with Nemotron datasets. The announcement remains largely vendor ecosystem partnership and platform promotion, without verifiable technical benchmarks.

NVIDIA announced an expanded sovereign AI collaboration with Korea, extending from GPU compute infrastructure to local AI talent and skills development. The announcement is a vendor-led industry partnership without specific model parameters, technical methods, or verifiable deployment metrics.