Article · AI translation
Thanks to IT Home readers 华南吴彦祖 and 不一样的体验 for the tips!
IT Home reported on October 26 that Nvidia CEO Jensen Huang said the annual license fee for accompanying AI software is about $2,000 per GPU (IT Home note: approximately 13,355 yuan at current exchange rates), to “run the full suite of software Nvidia provides to support AI agents on these AI factories.” A video of the remarks recently surfaced. Under one scenario calculated by FourWeekMBA, if 5 million paid licenses are sold, this pricing model could generate $50 billion (approximately 333.872 billion yuan at current exchange rates) over five years.
Huang discussed the pricing at the Berlin GTC conference on October 22. Nvidia also announced a broader plan at the event: to build an AI infrastructure business spanning chips, software, and AI agent security capabilities. “Software licensing will gradually become a very important part of our business. Our licensing model is roughly like this… For example, to run Nvidia’s full suite of software for deploying agents on AI factories, the cost is about $2,000 per GPU per year (approximately 13,355 yuan at current exchange rates),” Huang said.
The potential revenue is easy to calculate. If 5 million GPUs are covered by paid licenses, at $2,000 per GPU per year (approximately 13,355 yuan at current exchange rates), Nvidia could bring in $10 billion annually (approximately 66.774 billion yuan at current exchange rates); if customers keep renewing, the total could reach $50 billion over five years (approximately 333.872 billion yuan at current exchange rates). There are uncertainties, however: Nvidia has not said which GPUs the $2,000 quote (approximately 13,355 yuan at current exchange rates) applies to, or confirmed whether it is a uniform standard price. It is also unclear exactly which software the license includes and whether entitlements overlap across products.
The distinction matters because Nvidia already sells paid enterprise software. The listed annual subscription price for its AI Enterprise platform is $4,500 per GPU (approximately 30,049 yuan at current exchange rates). Although the $2,000 figure Huang mentioned (approximately 13,355 yuan at current exchange rates) is lower, Nvidia has not confirmed that the two prices cover exactly the same products and license terms, so it cannot be treated as an official price cut.
The discussion of software licensing comes as Nvidia develops more supporting infrastructure to help companies deploy and manage AI agents. These systems can call tools, read data, and perform tasks automatically, creating new security and management challenges when companies deploy them at scale.
On October 21, Nvidia released the OpenShell reference architecture, which integrates Vera Rubin computing infrastructure, Vera CPUs, and BlueField-4 data processing units (DPUs) equipped with Nvidia’s Sentry security component. OpenShell provides a secure runtime environment for AI agents, defines the boundaries of their actions, and limits their permissions to call tools and access resources. According to Nvidia, Sentry can monitor agent behavior independently and isolate an agent if it tries to cross those boundaries. The architecture places security controls in the underlying infrastructure instead of relying solely on safeguards within AI models or applications.
Nvidia launched an open agent security platform back in September. The new reference architecture builds on that platform, extending its approach to controls across enterprise infrastructure and edge devices.
Other companies are adding similar controls to their products. Anthropic is combining OpenShell with BlueField hardware and Claude managed agents so companies can control what AI agents can access within execution sandboxes. Salesforce has developed a Slack administration bridge for OpenShell that lets administrators review agents’ access requests, grant or deny permissions, and monitor their activity directly in Slack.
The new software pricing also reflects Huang’s longstanding view that Nvidia is no longer just a GPU supplier. Its business now extends to networking equipment, complete systems, enterprise software, and infrastructure for AI workloads. Huang recently said on the BG2Pod podcast, “One day, people may no longer classify us as a chip company.” As examples, he said Nvidia has entered industries including healthcare, automotive, communications, and life sciences, providing a complete technology stack rather than just chips.
Even so, Nvidia has not announced when the $2,000-per-GPU annual fee (approximately 13,355 yuan at current exchange rates) will take effect or which software it includes. It is also unclear how this pricing will fit with its existing enterprise software business.