Hugging Face CEO Clément Delangue on what Nvidia's open-source power grab really means — 3/19/26

Watch on YouTube ↗  |  March 19, 2026 at 18:35  |  29:40  |  CNBC
Speakers

Summary

  • Nvidia's GTC announcements mark a strategic pivot from a chip company to an open-source AI platform (e.g., NemoClaw) to build a new "moat" at the platform/operating system level for AI agents.
  • This open-source push is a defensive strategy: it commoditizes the model layer above to prevent players like OpenAI/Anthropic from gaining pricing power, and ensures no single customer becomes powerful enough to walk away from Nvidia's infrastructure (GPUs, data centers).
  • Hugging Face CEO states Nvidia has become the "American king of open source AI," dethroning Meta and competing with dominant Chinese providers.
  • Critical data point: In 2025, for the first time, the volume of open-source model downloads in the US from Chinese providers surpassed those from American providers.
  • Adoption is accelerating: ~30% of Fortune 500 companies now use Hugging Face/open-source models, often in a hybrid approach with proprietary APIs.
  • The "inference era" and the rise of AI agents are structural tailwinds for open-source adoption, as agents prioritize cost, speed, and control, and lower the skill barrier for building AI.
  • Chinese companies are currently leading in releasing high-quality, frontier-level open-source models, posing a strategic and values-based risk.
  • Meta has "fallen off" in the open-source race; contributions from US players (Google, OpenAI's releases, startups) are present but not at the same frontier level as China's.
  • The nature of open-source technology (forkable, modifiable) makes it harder to establish the same level of platform control and lock-in as seen in closed ecosystems (e.g., Apple's App Store).
Trade Ideas
Clément Delangue CEO, Hugging Face 16:00
Speaker says AMD's software is "so much better than it was maybe two three years ago," and they are emerging as a "good alternative for a lot of these workflows." As Nvidia pushes an open, multi-hardware software platform, the door opens for competitive GPU architectures to gain traction if their software stack improves. AMD is making tangible progress and is positioned as a viable alternative in the growing open-source AI ecosystem. Nvidia's first-mover advantage and ecosystem depth may be insurmountable.
Clément Delangue CEO, Hugging Face 17:00
Speaker states Nvidia needs as many people as possible to build AI themselves to avoid a world where only a few big labs (OpenAI, Anthropic) dominate. Their open-source contributions (e.g., >1 model/dataset daily on Hugging Face) align their incentives with the field's growth. Nvidia's core business is selling AI infrastructure (GPUs, data centers). By giving away open-source software (like NemoClaw), they expand the total addressable market of AI builders, all of whom will need compute. This strategic, infrastructure-centric open-source push solidifies Nvidia's central role and drives demand for its core products. The open-source model could be forked and community control could diverge if Nvidia's incentives shift.
Up Next

This CNBC video, published March 19, 2026, features Clément Delangue discussing AMD, NVDA. 2 trade ideas extracted by AI with direction and confidence scoring.

Speakers: Clément Delangue  · Tickers: AMD, NVDA