Nvidia launches Minerva to catch failing parts in giant AI data centres

Nvidia (NVDA) shares fell 3.8% to $213.51 after introducing DSX POD, a modular AI-factory design aimed at reducing deployment time from six months to 14 days. Partners include Dell, HPE, and Lenovo, with deliveries expected in early 2027. GuruFocus notes strong fundamentals but warns of valuation concerns.

Mercedes-Benz and NVIDIA are expanding their partnership to develop advanced driver assistance systems and automated driving technologies over the next decade. They will combine Mercedes-Benz's automotive expertise with NVIDIA's AI and computing capabilities, focusing on SAE Level 4 automation and automated luxury robolimousines. The collaboration will utilize NVIDIA's Hyperion architecture and Mercedes-Benz's MB.OS operating system, aiming to enhance safety and driving capabilities.

BNP Paribas reiterated its Outperform rating and $345 price target for Nvidia (NVDA) after the company's GTC Berlin presentations. Analyst Karl Ackerman highlighted Nvidia's plans for faster AI data-center deployment and optical networking advancements. The company expects its co-packaged optics plans to progress through 2026-2027 and introduced new systems like Vera Rubin and Groq 3 LPX.

Samsung Electro-Mechanics and NVIDIA will expand production of AI chip substrates in South Korea. The multiyear deal includes dedicated manufacturing at sites in Sejong and Busan. Samsung expects double-digit annual growth in shipments to NVIDIA starting in 2026, with NVIDIA-allocated production becoming a significant share of its capacity by 2028. Reports suggest Tesla and NVIDIA may invest 3.7 trillion won in the project, though this has not been confirmed.

Dell Technologies expanded its AI Factory with Nvidia Vera CPUs and Groq 3 LPUs, launching three new products. The PowerEdge M9882 rack has 256 Vera CPUs, while the R8822 server offers a two-socket Vera CPU. The XE9810 is Dell's first Groq 3 LPU server. Dell reports 6,500 AI Factory customers, up from 5,000 last quarter. The R8822 is available now, with the others in Q1 2027.

NVIDIA unveiled DSX POD, a modular AI data center design that can reduce deployment time from months to 14 days. The company aims to speed up AI infrastructure setup, addressing the growing demand for AI compute. Separately, critical PHP vulnerabilities caused government sites to go offline, highlighting security risks in AI infrastructure.