Opinions expressed by Entrepreneur contributors are their own.
Listen to this post
- Unlike closed models, open-weight models let organizations control model versions, fine-tuning and deployment while keeping sensitive data within a chosen legal jurisdiction.
- Open-weight models reduce dependence on a single provider’s API, infrastructure and pricing, making models and accumulated knowledge easier to transfer between providers and projects.
- Open ecosystems allow businesses to benefit from community contributions and independently test security, while licences such as Apache 2.0 define how models may be modified and used commercially.
On July 24, Nvidia joined Microsoft, Meta, IBM, Hugging Face, Mistral and other technology organizations in signing Open Weights and American AI Leadership. The letter argues that AI leadership will depend not only on developing powerful frontier models, but also on creating an open ecosystem that distributes those capabilities throughout the economy.
The most important idea in the letter is sovereignty.
For governments, sovereignty means maintaining control of critical technology. For businesses, it operates at several levels. It includes control of model behavior, company data, hosting infrastructure, costs and the knowledge accumulated through years of work.
According to NIST’s definition of a model weight, a weight is “a numerical parameter within an AI model that helps determine the model’s outputs.” In practical terms, weights contain much of what a model learned during training.
An open-weight model makes those parameters available for download. Subject to its licence, an organisation can run the model, evaluate it, adapt it and deploy it on infrastructure of its choice.
The licence defines how published weights may be used. Under the Apache License 2.0, they can generally be used commercially, modified and redistributed, provided the required notices are preserved. Businesses should still check each model’s specific licence before deployment.
A closed weight model keeps those parameters private. Customers access the model through an application or API, while the provider controls the underlying model, its hosting and usually its update cycle.
When a company deploys an open-weight checkpoint, the underlying weights remain fixed. The model does not unexpectedly change because its provider introduced a new version. The business decides whether to preserve it, replace it or fine tune it.
This stability matters when AI becomes part of an operational workflow. A model used to classify documents, extract information or assist employees may have been tested against hundreds of internal cases. A provider-controlled update can change its output structure, tone or performance and force the business to repeat that evaluation.
Open weights give the company control over that schedule. The model can remain unchanged while engineers improve the prompts, retrieval layer and application around it.
This does not mean every output will always be identical. Quantization, inference software, prompts and generation settings can still affect results. The important point is that the organization controls those changes rather than receiving them automatically.
When a company uses a closed model through an external API, its data is processed in an environment controlled by another provider. Contractual protections may limit retention, but the organization still depends on that provider’s policies and legal jurisdiction.
The U.S. CLOUD Act shows why server location alone does not guarantee sovereignty. A provider subject to U.S. law may be required to disclose data under valid legal process, even when that data is stored in Europe.
This does not give U.S. authorities unrestricted access. However, it can create uncertainty when US disclosure obligations conflict with European data protection rules. The CNIL therefore warns that sensitive data may remain exposed when handled by companies subject to non-European laws.
Open-weight models reduce this dependency by allowing businesses to process data in an environment they control. Data sovereignty means knowing who controls the infrastructure, which laws apply and whether prompts or outputs are retained.
Open weights also give organizations greater control over the infrastructure on which their AI operates.
A business can deploy a model on its own servers or choose a cloud provider according to its requirements for geography, regulation, cost and performance. European companies, for example, can use providers such as Scaleway or OVHcloud to keep workloads on infrastructure located in France or elsewhere in Europe.
This creates several deployment options. A company can self-host for maximum control, use a European managed provider to reduce operational complexity or move the same model between providers as its needs evolve.
With a closed-weight model, the model and its infrastructure are usually inseparable. Changing the hosting provider often means changing the model as well. Open-weight models separate those two decisions, allowing the organization to choose both the intelligence it uses and the environment in which it runs.
“Open weights expand access to the AI economy,” the letter states. Organizations can match the right model to each task instead of paying frontier prices for every operation.
With a closed model, the provider controls both access and pricing. API prices, usage limits or service tiers may change, leaving customers to absorb the cost or migrate.
Open models still require infrastructure and engineering, but businesses retain more options. They can change hosting providers, self host or optimise inference without necessarily replacing the model.
Economic sovereignty does not eliminate costs. It gives businesses greater control over how those costs evolve as usage grows.
An AI project accumulates evaluation data, fine-tuning examples, retrieval systems, prompts, adapters and specialized business knowledge. With an open-weight architecture, those assets remain under the company’s control and can be transferred between applications.
A model adapted for one internal assistant can support another product. A retrieval layer created for customer support can later serve an analytics platform or an AI agent. The knowledge does not remain trapped inside the project or provider that first produced it.
Open-weight models are therefore not simply cheaper alternatives to closed systems. They allow organizations to decide how their models evolve, where their data travels, which infrastructure they use and how their accumulated knowledge moves from one project to the next.
The letter states, “Relying solely on closed models is not inherently safe.” Closed systems can still be breached, misused or fail in ways that external researchers cannot detect.
Open weights allow a broader community to examine model behavior, identify vulnerabilities, conduct red teaming and develop safeguards. They do not guarantee security, but they enable independent evaluation and let businesses benefit from protections created by researchers and developers beyond the original provider.
That is what AI sovereignty looks like at the level of a business.
