The Future of AI: Open and Proprietary Models from NVIDIA

Artificial intelligence is becoming the defining technology of our time, rapidly transforming into a core business infrastructure. It feeds on a diverse ecosystem of models: large and small, open and proprietary, general-purpose and specialized. This diversity is essential for a future where every application will use AI, every country will develop it, and every company will apply it.

NVIDIA's founder and CEO Jensen Huang noted during a special session at GTC that "the choice between proprietary and open is not a choice. It is both proprietary and open." The future of innovation in AI is not tied to a single massive model. Each industry, whether healthcare, finance, or manufacturing, faces unique challenges and requires AI capable of processing their data and workflows.

NVIDIA is actively involved in the development of open AI and is the largest organization on the Hugging Face platform with nearly 4,000 participants. At GTC, the company announced the creation of the NVIDIA Nemotron coalition, a unique global collaboration of model developers and AI labs aimed at promoting open models through shared expertise, data, and computing resources.

The first project of the coalition will be a foundational model developed in collaboration with Mistral AI, supported by the data and expertise of coalition members for post-training and further development of the model. It will be available to the open ecosystem and will serve as the basis for the next generation of NVIDIA Nemotron models, which have already been downloaded over 45 million times.

A panel discussion at GTC featured leaders in open models, discussing key points such as the growing capability of AI agents to perform complex tasks, the need for a multi-layered and multimodal system of models, and the importance of openness for innovation in the model ecosystem.