Jensen Huang tells investors open and closed AI models can both lift Nvidia’s chip demand

Nvidia Chief Executive Jensen Huang pushed back on a bearish line of thinking tied to the rise of open models, arguing that the market does not hinge on one approach winning outright. Speaking in remarks reported by Yahoo Finance on August 28, Huang said that both closed and open models are likely to succeed, and that the resulting demand for compute is the practical outcome Nvidia is positioned to supply.

The CEO’s comments come as the AI industry has increasingly split along two development philosophies. “Closed” models are typically built and deployed by a single vendor using proprietary weights, infrastructure, and distribution. “Open” models, by contrast, are released in a way that allows broader access to model weights and modification by other builders, including enterprises and software developers. Huang’s point, as characterized in the report, was that either pathway can still drive the training and inference workloads Nvidia hardware is designed to run.

Huang’s message to skeptical investors was framed around how AI adoption translates into chip and systems demand. Even if open models reduce the market share of a particular proprietary provider, they can still expand the total number of developers running large-scale models, scaling demand for accelerators used in data centers. In that sense, the CEO’s stance is less about forecasting which model category will dominate and more about arguing that Nvidia benefits from sustained growth in overall model usage.

The report also reflects a wider debate in financial markets about whether open-source distribution could erode margins or constrain spending on the most expensive, full-stack commercial offerings. Huang’s position suggested that Nvidia’s business is not fully dependent on any single ecosystem, because customers running open models still need substantial compute and specialized hardware for performance and efficiency.

For Nvidia, the commercial core of that compute demand is its data center platform, which is built around accelerating AI workloads such as training (learning model parameters) and inference (using a trained model to generate outputs). Nvidia’s role is typically upstream in the “compute supply chain,” providing GPUs and related systems that other companies integrate into their platforms. That upstream posture is one reason the company is often viewed as a barometer for AI infrastructure spending, regardless of which model providers are getting the attention.

Still, there are clear limits to what can be concluded from the reported remarks alone. Yahoo Finance’s coverage, as summarized in the alert that reached this desk, does not provide detailed financial guidance, specific customer commitments, or quantified impacts from any open model trend. It also does not spell out whether Nvidia expects any change in pricing or mix between customers using open versus closed model stacks.

As the market digests Huang’s framing, investors are likely to focus on follow-through: whether Nvidia later ties the “both will succeed” thesis to concrete indicators like data center order patterns, platform adoption rates, or commentary on how open model deployments map to Nvidia’s software and hardware ecosystem. Those specifics are what would turn a high-level industry argument into a clearer read-through for near-term revenue drivers.

Until Nvidia provides additional disclosure, the practical takeaway is directional. Huang is effectively arguing that open models do not eliminate the compute cycle, and that the company can benefit as long as the overall AI workload demand continues to expand, whether the models themselves are proprietary or broadly distributed. That may reassure some investors, even if questions remain about how competitive pressures and customer procurement choices could affect Nvidia’s margins.