NVIDIA used its Washington D.C. showcase to sketch the next phase of the AI build-out across chips, models, partners, and networks. The company said the U.S. Department of Energy is standing up seven supercomputers powered by more than 100,000 Blackwell GPUs, with manufacturing occurring domestically. NVIDIA also released new open-source models and large datasets spanning reasoning, robotics/physical AI, and biomedical research, and unveiled a string of enterprise partnerships across healthcare, automotive, and platforms.

The headline financial and ecosystem news was a strategic, $1 billion investment in Nokia aimed at accelerating the telecom industry’s pivot to AI-native networking. Nokia plans to redesign carrier infrastructure around AI processing—bringing inference and data-plane intelligence closer to the edge while tapping NVIDIA’s accelerated compute and software stack. The collaboration signals how core internet plumbing is being re-architected for latency-sensitive, high-throughput AI workloads rather than traditional traffic patterns. NVIDIA’s stock touched new highs on the updates, underscoring investor conviction that demand for training and inference silicon will remain intense through the Blackwell and forthcoming Rubin cycles.

Beyond networking, NVIDIA emphasized an “all of the above” approach: domestic chip manufacturing for national labs, domain-specific open models to seed new applications, and deeper ties with enterprises such as Eli Lilly, Palantir, Hyundai, Samsung, and Uber. The cumulative message: the AI boom is broadening from data centers to industrial systems, healthcare, and telecom, with GPUs and CUDA-aligned software as the connective tissue. Read the company’s announcement on the Nokia partnership here: NVIDIA–Nokia AI telecommunications investment.