# The battle for open AI: Google, OpenAI and Anthropic propose restricting it

> Google, OpenAI and Anthropic advocate limiting open AI, while NVIDIA, Meta and Microsoft lead the commitment to open source
> By José Maldonado · Artificial Intelligence · 2026-07-25

The debate around artificial intelligence has ceased to be a purely technical discussion and has become the most bitter political and economic battle of the decade. At the center of the conflict is open-source artificial intelligence (or open-weight models), a technology that allows developers, researchers and companies around the world to download, modify and run AI models on their own infrastructure.

On one side, entrenched giants of the closed model such as Google and Anthropic are pressing governments —especially the United States— to impose severe restrictions, prior licenses or explicit bans on the open deployment of frontier models. On the other, a coalition as diverse as it is powerful, led by Jensen Huang (NVIDIA), Mark Zuckerberg (Meta), Satya Nadella (Microsoft) and Elon Musk ([**SpaceXAI**](https://x.ai/)), has stepped forward to defend free software as the indispensable engine of global innovation.

This dispute is not only about security or lines of code: it represents a strategic crossroads where it is decided who will retain control over the infrastructure of the future and how global economic power will be distributed.

**[NVIDIA, Meta and Microsoft back open AI without OpenAI, Google and xAI](https://observatorioblockchain.com/ia/nvidia-meta-y-microsoft-apoyan-la-ia-abierta-sin-openai-y-google/)**

## The position of Google, Anthropic and OpenAI

Google and Anthropic have championed the thesis that the most advanced artificial intelligence models pose a catastrophic and unsustainable risk if their weights are released publicly without filters or direct safeguards from the developer.

At the center of their arguments [**they maintain**](https://www.theregister.com/ai-and-ml/2026/06/26/google-wants-ai-regulation-but-on-its-own-terms/5263276) that releasing models capable of discovering complex vulnerabilities in operating systems or designing biological pathogens represents a direct threat to national security if it falls [**into the hands of malicious actors**](https://www.axios.com/2026/07/22/openai-anthropic-open-models-trump-china). They also argue that once a model's weights are downloaded locally, it is impossible to apply guardrails or reverse modifications that remove the ethical limits originally imposed.

However, behind this discourse focused on human safety there is a clear strategy of regulatory capture to protect their business model. Training frontier models requires multi-million-dollar investments in compute and infrastructure, so if an open model reaches performance similar to a closed alternative accessible by subscription or API, the pricing power of both companies collapses.

Since Google, OpenAI and Anthropic sell access to AI as a cloud service (API-as-a-Service), the demand for state licenses and regulatory barriers ends up restricting the market to a few corporations with the financial and bureaucratic capacity needed to comply with regulators, thus guaranteeing a commercial monopoly.

## The pro-free AI front: Why do Meta, NVIDIA, Microsoft and xAI defend open source?

Faced with the pressure exerted to restrict the distribution of open models, various influential figures in the technology industry have formed a heterogeneous alliance in defense of open-weight AI. Although they present themselves as a united front, the motivations of this group respond to radically opposed commercial strategies and business models.

## NVIDIA's position

NVIDIA, under the leadership of [**Jensen Huang**](https://x.com/JensenHuang/status/2080643682408321103), acts as the key supplier in this technological race. Since it does not sell artificial intelligence models as a final product or service, its business centers on selling the computer hardware needed to process these technologies.

The expansion of open AI directly benefits the company, since thousands of developers, companies and institutions require local infrastructure to fine-tune and run such models, which drives up global demand for its graphics processing units.

![Jensen Huang's position is clear: open source AI must be protected](https://content.observatorioblockchain.com/wp-content/uploads/2026/07/Captura-de-pantalla_20260725_021645.png)
*Jensen Huang's position is clear: open source AI must be protected*

By contrast, if AI is confined exclusively to the data centers of a handful of closed firms, the hardware market would become drastically concentrated, limiting its long-term growth opportunities.

**[Why the future of the global crypto industry is being decided these days in Washington](https://observatorioblockchain.com/en/crypto/el-futuro-de-la-industria-cripto-mundial-se-decide-en-washington/)**

## Meta and Microsoft also take a stance

On Meta's side, Mark Zuckerberg has promoted the Llama family of models to position the company as the benchmark for open source. Since the company does not sell access to language model APIs, its strategy seeks to turn its developments into the industry standard.

By allowing the global community to optimize and integrate its models, Meta absorbs collective innovation without bearing all the research and development cost alone, while neutralizing its competitors' pricing power. A government restriction on open models would nullify this strategic move, leaving the market in the hands of the closed APIs of rivals such as Google or OpenAI.

Microsoft's position, led by Satya Nadella, balances its commercial alliance with OpenAI and the adoption of open architectures. The core of the firm's business lies in its cloud platform, Azure, which bills regardless of whether customers run proprietary models or open alternatives such as Llama or Mistral. Diversifying its offering allows it to capture enterprise compute consumption without depending on a single software provider, while strict regulation of open source would reduce the activity of independent developers and, therefore, the widespread use of its cloud infrastructure.

## Elon Musk supports the idea on X

Finally, xAI, backed by Elon Musk, approaches the conflict from direct competition against the centralization of the sector. Promoting open-access architectures allows the company to challenge the hegemony of the bloc formed by OpenAI and Google, avoiding submitting to their operating standards or ideological filters.

![Elon Musk's response to the NVIDIA CEO's post](https://content.observatorioblockchain.com/wp-content/uploads/2026/07/Captura-de-pantalla_20260725_021136.png)
*Elon Musk's response to the NVIDIA CEO's post*

For new competitors and startups, the imposition of severe regulatory barriers to open source would mean the impossibility of competing on equal terms, permanently consolidating the dominance of established players.

## The key questions about the future of the ecosystem

Although the cybersecurity and disinformation risks presented by defenders of the closed model are real, the history of software development consistently shows that security based on secrecy tends to be inferior to the strength of open auditing. Allowing the global community to review the code, audit the weights and identify vulnerabilities strengthens the overall resilience of digital defenses worldwide, whereas limiting access to a small group of corporations does not eliminate threats, but simply centralizes control of the most strategic technology of the 21st century.

**[Valve launches Proton 11 and gains ground on Windows in gaming](https://observatorioblockchain.com/en/technology/valve-lanza-proton-11-y-gana-terreno-a-windows-en-el-gaming/)**

Likewise, this discussion has a direct impact on global technological sovereignty, since if the United States imposes severe restrictions on its own open-source developers, it risks pushing regions such as Europe, Asia and Latin America toward alternative ecosystems or models created in jurisdictions without such brakes. Open AI ensures that different countries can adapt these models to their languages, legal frameworks and cultural realities without depending on the bias or infrastructure of a few US multinationals.

## Protecting an open source future

Ultimately, the tension between closed models and free artificial intelligence transcends the technical to become the struggle for the architecture of future knowledge. If the restrictive stance of Google and Anthropic prevails, AI will be confined behind paywalls and state licenses, consolidating itself as a privatized service under the control of a corporate oligarchy.

By contrast, if the vision of the alliance led by NVIDIA, Meta, Microsoft and the global free software community prevails, the technology will maintain the decentralized dynamic that made the Internet prosper, guaranteeing an ecosystem where any developer or institution can freely audit, adapt and build on the most advanced tools of our time.
