Google DeepMind Chief Backs Nvidia's Call for Open AI Ecosystem

Google DeepMind chief Demis Hassabis joined the ranks of open artificial intelligence model supporters on Saturday, publicly backing the stance of Nvidia chief executive Jensen Huang. It's another voice in the growing dispute over whether the future of AI should belong to the closed systems of major companies or to models with openly released weights.

The spark for the whole discussion came from Jensen Huang, who created an X account on Friday and immediately used it to share an open letter titled "Open Weights and American AI Leadership." The document was signed by more than twenty five companies and organizations from across the industry: hardware makers like Nvidia and Dell, model developers like Meta, Microsoft, Mistral and Hugging Face, application companies like Perplexity, Palantir and CrowdStrike, investors like Andreessen Horowitz and Y Combinator, and organizations such as the Linux Foundation, Mozilla and IBM.

What Huang's Letter Argues

The letter's authors argue that open model weights expand economic access for startups and universities that don't have to train their own systems from scratch. A second argument is competition - open models are meant to prevent innovation from concentrating in the hands of a few of the largest companies. A third point concerns control - organizations using open models retain greater sovereignty over their data and how the system is deployed.

Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty - Jensen Huang, CEO of Nvidia

The letter also addresses directly the security concerns that opponents of open models have raised for months. Signatories argue that open code allows for broader community auditing, offering better protection than relying on the assumption that closed code alone guards against attacks. They also stress that cybersecurity defenders need the same tools that attackers have at their disposal.

Hassabis Weighs In With DeepMind's Record

Demis Hassabis joined the discussion on Saturday, sharing Huang's post on X and adding his own comment. The Google DeepMind chief stressed that a strong and secure open ecosystem matters for the whole world to benefit from artificial intelligence.

A strong and secure open ecosystem is important for the world to benefit from AI - Demis Hassabis, CEO of Google DeepMind

Hassabis used the occasion to recall DeepMind's own track record in open science, citing the JAX, Transformers and AlphaFold projects as examples of tools shared with the research community. He also pointed to the open Gemma model family, which by his count has been downloaded more than 300 million times. At the same time, he noted that both open and proprietary systems need responsible safety standards.

Chinese Models Dispute in the Background

The debate over open weights is playing out against the backdrop of a tense dispute between Washington and Beijing over Chinese AI models. In an interview with Axios, Huang explicitly named the Chinese models DeepSeek and Moonshot AI's Kimi K3 as examples of high-quality open systems that, in his view, should be widely used rather than banned. White House science adviser Michael Kratsios had previously accused Moonshot AI of using large-scale distillation to copy an American model. The letter's signatories defend ordinary distillation as a legitimate technique for improving models, rejecting only the unlawful extraction of intellectual property.

It's telling who did not sign the letter. Anthropic and OpenAI were absent, companies that have previously warned publicly that powerful open Chinese models could pose security risks and infringe on intellectual property. Their absence underscores how deep the industry's divide is between the camp favoring maximum openness and companies betting on closed, controlled systems.

What It Means for the Market

For Polish companies and developers, this dispute has practical implications. Open models such as Gemma or competing open source solutions are cheaper to deploy locally and help avoid dependence on a single cloud provider, which can matter under data sovereignty requirements. The involvement of such a large group of technology companies, from chipmakers to venture capital investors, suggests that lobbying for open models in Washington will intensify in the coming months, especially given planned regulations on the import and use of Chinese AI systems.