Last week, NVIDIA CEO Jensen Huang joined a coalition of technology leaders in signing an open letter titled Open Weights and American AI Leadership. At first glance, it may appear to be another policy paper aimed at governments. It is much more significant than that.

The letter argues that open-weight artificial intelligence models are becoming a critical part of national competitiveness and should be supported, not unnecessarily restricted. While the paper is written from a United States perspective, the ideas it raises have important implications for regional communities like Ballarat and for organisations seeking to build long-term capability in

Why Jensen Huang matters

Jensen Huang is the founder and CEO of NVIDIA, the company whose technology underpins much of today’s AI revolution. Its graphics processing units (GPUs) power the training and operation of many of the world’s most advanced AI models and have become essential infrastructure for research institutions, governments and businesses alike.

When Jensen Huang speaks about the future of artificial intelligence, policymakers, investors and technology leaders pay attention. His signature alone would have made this letter noteworthy. However, it was far from the only significant name attached.

A remarkable coalition

The letter was also signed by organisations including Microsoft, Meta, IBM, Hugging Face, Mozilla, Mistral AI, Palantir, ServiceNow, Perplexity, Y Combinator and the Linux Foundation, among many others.

One signature, in particular, stands out: Microsoft.

At first, Microsoft’s support appears surprising. After all, it has invested billions of dollars in OpenAI, whose flagship models are largely closed. Yet Microsoft’s position reflects a much broader strategy.

The company is not simply betting on one AI model. It is betting on becoming the platform on which AI runs. Whether organisations choose OpenAI’s GPT models, Meta’s Llama models, Microsoft’s own Phi models or other open-weight alternatives, Microsoft benefits when those workloads are deployed on Azure.

In other words, Microsoft is hedging its position. It recognises that the future of AI is unlikely to belong exclusively to either open or closed models. Supporting a healthy ecosystem of both protects its long-term strategic interests while giving customers greater choice.

What the letter argues

The central argument is straightforward.

The authors believe AI leadership will not be determined by a single frontier model, but by the strength of the ecosystem built around artificial intelligence.

They argue that open-weight models expand access to advanced AI, encourage competition, reduce costs and allow organisations to retain greater control over their technology. They also contend that openness can improve security by enabling more researchers and developers to identify vulnerabilities, test systems and develop safeguards.

Importantly, the letter does not argue that closed models are bad. Instead, it argues that governments should avoid restricting open-weight AI simply because it is open. Rather than limiting legitimate innovation, policymakers should invest in the infrastructure, datasets and research capability needed to support a competitive AI ecosystem.

Open weights versus closed models

Understanding this debate begins with a simple distinction.

A closed model is one you can use but cannot view how the model works directly. Models such as OpenAI’s GPT series or Anthropic’s Claude are accessed through online services. The organisations behind them retain the model itself, while users access its capabilities through an interface or API. To understand these models, we rely on the organisation itself to release information about it.

An open-weight model, by contrast, makes its trained model weights available for organisations to download, run and adapt on their own infrastructure. Examples include Meta’s Llama family, Microsoft’s Phi models, Mistral AI, Qwen and DeepSeek. You can download these model weights, and run them yourself on your own hardware (although the hardware requirements for many models can be extremely expensive).

For many organisations, this distinction is significant. Open-weight models can be customised with internal knowledge, operated within private environments and improved over time without becoming permanently dependent on a single technology provider. Closed models often offer exceptional performance and convenience but involve a greater level of reliance on external platforms. You must rely on a closed model’s organisation to get access to the model, which normally means ongoing subscription pricing.

Neither approach is inherently better. Each has strengths, limitations and appropriate use cases.

BRAIN’s position

At BRAIN, we believe that organisations need to make informed decisions on whether to use open or closed models, and how they interact with AI. Our focus is on helping organisations understand the trade-offs and make these informed decisions that align with their objectives, governance requirements and long-term resilience.

The best models in the world are currently closed models, so organisations might prefer to use these for challenging tasks with a high complexity. However, open models are not far behind, so for many tasks you’ll find a closed model sufficient. Organisations can choose open models for self-hosting (running on your own hardware), control over long-term usage, and for the added transparency an open model provides. If you have digital sovereignty concerns, open-weight models may be important to consider.

We also believe that Australia’s AI future should not be measured solely by the models we develop or use. It should be measured by the capability we build across our businesses, institutions and communities. Regions that understand these technologies, develop governance capability and retain greater control over their digital infrastructure will be better positioned to create economic opportunity and adapt to the changes ahead.

Join the conversation

Artificial intelligence is evolving rapidly, but the decisions organisations make today will shape their capability for years to come.

BRAIN exists to help regional businesses, governments, educators and community organisations navigate these changes with confidence. Through practical guidance, research and collaboration, we are building the capability our region needs to prosper in the age of AI.

If you would like to be part of that journey, we invite you to become a BRAIN member, participate in our events and contribute to the growing conversation about the future of AI in regional Australia. The future will not simply happen to our region. Together, we can help shape it.

The link has been copied!