Ever since taking place this Tuesday, Nvidia CEO Jensen Huang's keynote at the Nvidia GTC 2025 has become an inescapable reminder of Nvidia's pivotal role in the AI ecosystem, with Huang declaring that we've reached an "inflection point" in AI technology as the computational demands of the more complex AI workloads required by "reasoning" models and AI "agents" continue to trigger investments in data center infrastructure poised to reach a staggering $1 trillion.

Major announcements: Blackwell, Vera Rubin, and strategic collaborations

Some of the most exciting developments announced at GTC include:

  • The NVIDIA Blackwell Ultra, coming in the second half of this year, is a next-generation AI factory platform that dramatically improves training and inference capabilities.
  • Open-source tools like CUDA Dynamo, designed to accelerate AI reasoning models, and the openly available Llama Nemotron models, which deliver up to 20% improved accuracy and 5x faster inference than competitors, and are designed to provide the "reasoning" capabilities required to build advanced AI agents.
  • A collaboration with General Motors to develop next-generation AI-powered vehicles and robots, and one with Alphabet and Google focused on physical and agentic AI.
  • Per Nvidia's yearly roadmaps, the next-generation NVIDIA Rubin Ultra GPU and NVIDIA Vera CPU architectures have a planned release date during the second half of 2026. Vera Rubin systems promise significant innovations and dramatic performance gains: capable of up to 50 petaflops while doing inference, they more than double the 20 petaflops delivered by the current Blackwell chips. Rubin Ultra systems, collections of four GPUs in a single package delivering up to 100 petaflops, are expected to be available during H2 2027.

Hardware is not all there is to the AI ecosystem

Intriguingly, Huang suggested that every company will soon have two factories: one for physical products and another for AI. Naturally, the company expects this will be made possible by the dominance and pervasiveness of the Nvidia CUDA ecosystem. Nvidia proudly reports that "CUDA-X GPU accelerated libraries and microservices are now serving every industry", and acknowledges its role in achieving what the company calls the "tipping point of accelerated computing".

The conference also explored exciting frontiers in agentic AI, robotics, and physical AI. Nvidia's Isaac GR00T N1, the world's first open foundation model for humanoid reasoning, and the Cosmos world models represent significant steps towards more sophisticated, adaptable AI systems.

Flexible, customizable, end-to-end solutions

With the announcement of technologies like photonics networking and personal AI supercomputers like DGX Spark and DGX Station, Nvidia is not just creating hardware—it's building the entire ecosystem for the AI revolution.

While not as splashy as previous iterations of the conference, this year's GTC was a clear demonstration of Nvidia's dominance in the industry and a confirmation that it shows no signs of slowing down. Huang's closing remarks conveyed Nvidia's confidence in its future as it prepares to support inference, which "is going to be one of the most important workloads in the next decade as we scale out AI" and to build a dizzying amount of infrastructures: "one for the cloud, a second for the enterprise and a third for robots."

Other noteworthy headlines this week

OpenAI to invest billions in CoreWeave as tensions with Microsoft continue growing: CoreWeave and OpenAI announced a strategic partnership to provide the latter with even more AI infrastructure. OpenAI has committed up to $11.9 billion to be disbursed across five years in exchange for $350 million in shares.

Gleamer adds AI-assisted MRI diagnosis to its portfolio after acquiring Pixyl and Caerus Medical: Gleamer recently acquired Pixyl and Caerus Medical to expand its portfolio to include neural and lumbar MRI scan analysis, furthering its goal of developing specialized AI tools for various radiology diagnostics tasks.

OpenAI Launches New Tools for Building AI Agents: OpenAI has launched a comprehensive suite of new tools including the Responses API, built-in capabilities for web search, file search, and computer use, and an open-source Agents SDK—all designed to make it significantly easier for developers to build AI agents.