Nvidia floats $2,000 per GPU software fee for AI agents

CEO Jensen Huang says annual subscription could generate tens of billions in new revenue as company expands beyond chips

By LineZotpaper AI staff

Published25 October 2026

Read Time2 min

Nvidia CEO Jensen Huang said at GTC Berlin on October 22 that the company is considering charging around $2,000 per GPU per year for the software needed to run AI agents on its infrastructure, a pricing model that could generate $50 billion over five years under one scenario outlined by analysts. Huang stressed that software licensing will become an increasingly major part of Nvidia’s business but has not confirmed when the fee would take effect or which products it would cover.

Speaking at GTC Berlin, Huang described a future where enterprises running AI agents, systems that can use tools, access data and execute tasks, would pay an annual subscription for Nvidia’s software stack. “Software licensing will become increasingly a major part of our company’s business. The way that we license will be…for example, it’s about $2,000 per GPU per year to run all the software that Nvidia provides that runs agents on these AI factories,” he said.

At $2,000 per GPU per year, five million paid licenses would bring in $10 billion annually. Over five years that would reach $50 billion, assuming customers renew their subscriptions. However, Nvidia has not disclosed how many GPUs the fee would cover, whether that price is standard, or which software it includes. The company already charges $4,500 per GPU for a one-year subscription to its AI Enterprise platform, and it remains unclear how the new figure relates to that existing product.

The licensing discussion comes as Nvidia develops more of the infrastructure needed to deploy and manage AI agents. On October 21, the company announced the OpenShell Reference Architecture, combining its Vera Rubin computing infrastructure, Vera CPUs, and BlueField-4 data processing units running Nvidia Sentry. Sentry monitors agent activity and can quarantine agents that violate defined boundaries. Anthropic is integrating OpenShell and BlueField hardware with Claude Managed Agents, while Salesforce has developed a Slack Admin Bridge for OpenShell that lets administrators review and approve agent access requests.

Huang argued that Nvidia is becoming more than a chip supplier. “One of these days, people might not categorize us as a chip company anymore,” he said on a podcast, pointing to the company’s presence in healthcare, automotive, telecommunications and life sciences. Nvidia has not said when the $2,000-per-GPU fee would take effect or how it fits into its existing enterprise software business.

RSS

The Daily Edition

The front page in your inbox each weekday morning: the lead, the picks, and why each one earned its place.

Analysis

Why This Matters

The proposed fee signals Nvidia’s intention to build a recurring software revenue stream on top of its hardware sales, potentially shifting its business model and increasing costs for enterprises deploying AI agents.
If widely adopted, the pricing could standardise the cost of enterprise AI infrastructure and reinforce Nvidia’s position as the dominant platform provider, from chips to security software.
The outcome will affect enterprises’ bottom lines and may influence how competitors price their own AI infrastructure offerings.

Background

Nvidia has long been known primarily as a GPU maker, but in recent years it has expanded into networking, systems, enterprise software and full AI infrastructure stacks. Its existing AI Enterprise platform costs $4,500 per GPU for a one-year subscription. The new $2,000 figure is lower, but Nvidia has not confirmed it as a replacement or a separate product. The company’s push into agent security software, including the OpenShell reference architecture and the Sentry monitoring system, reflects a broader strategy to provide the entire software layer needed to run AI agents securely in enterprise environments.

What to Watch

When and how Nvidia clarifies the $2,000-per-GPU fee: which software it covers, how it relates to AI Enterprise, and what the rollout timeline looks like.
The adoption rate of Nvidia’s OpenShell and Sentry among enterprise customers and partners, as well as any competitive responses from other AI infrastructure providers.
Whether regulators or customers push back on the bundling of hardware and software subscriptions as Nvidia tightens its control over the AI stack.

Sources

Nvidia discusses $2,000-per-GPU annual software fee for running agents on AI infrastructure · Tom's Hardware

Checked

12 claims in this story checked against 1 source: 10 supported.

Show each claim

SupportedHuang said, “Software licensing will become increasingly a major part of our company’s business. The way that we license will be…for example, it’s about $2,000 per GPU per year to run all the software that Nvidia provides that runs agents on these AI factories.”Latest from Tom's Hardware: ““Software licensing will become increasingly a major part of our company’s business. The way that we license will be…for example, it’s about $2,000 per GPU per year to run all the software that Nvidia provides that runs agents on these AI factories,” said Huang.”
SupportedHuang said, “One of these days, people might not categorize us as a chip company anymore.”Latest from Tom's Hardware: “Speaking on the BG2Pod podcast recently, Huang said, “One of these days, people might not categorize us as a chip company anymore.””
Partly supportedHuang said an annual subscription could generate tens of billions in new revenue.Latest from Tom's Hardware: “Nvidia CEO Jensen Huang has put the annual cost of licensing the company’s AI software at around $2,000 per GPU, reports The Register, a pricing model that could generate $50 billion over five years under a proposed scenario involving five million paid licenses outlined by FourWeekMBA.”
SupportedNvidia is considering an annual software fee of around $2,000 per GPU for running AI agents.Latest from Tom's Hardware: “Nvidia CEO Jensen Huang has put the annual cost of licensing the company’s AI software at around $2,000 per GPU, reports The Register”
SupportedHuang discussed the pricing at GTC Berlin on October 22.Latest from Tom's Hardware: “Huang discussed the pricing at GTC Berlin on October 22”
SupportedFive million paid licenses at $2,000 per GPU per year would generate $10 billion annually and $50 billion over five years, assuming renewals.Latest from Tom's Hardware: “At $2,000 per GPU each year, Nvidia would bring in $10 billion annually if five million GPUs were covered by paid licenses. Over five years, that would amount to $50 billion, provided customers kept renewing their subscriptions.”
SupportedNvidia has not disclosed how many GPUs could be covered by the proposed fee, whether it is the standard price, or which software it covers.Latest from Tom's Hardware: “Nvidia has not disclosed how many GPUs could be covered by the proposed $2,000 fee, or whether that’s its standard price. It’s also unclear which software the license would cover.”
Partly supportedNvidia charges $4,500 per GPU for a one-year AI Enterprise subscription.Latest from Tom's Hardware: “Its AI Enterprise platform is listed at $4,500 per GPU for a one-year subscription.”
SupportedOn October 21, Nvidia announced an OpenShell Reference Architecture combining Vera Rubin infrastructure, Vera CPUs and BlueField-4 DPUs running Nvidia Sentry.Latest from Tom's Hardware: “On October 21, Nvidia announced its OpenShell Reference Architecture, a system that combines its Vera Rubin computing infrastructure, Vera CPUs, and BlueField-4 data processing units (DPUs) running Nvidia Sentry.”
SupportedNvidia Sentry monitors agent activity and can quarantine an agent that crosses defined boundaries.Latest from Tom's Hardware: “Sentry monitors agent activity independently and can quarantine an agent that attempts to cross those boundaries, according to Nvidia.”
SupportedAnthropic is integrating OpenShell and BlueField hardware with Claude Managed Agents.Latest from Tom's Hardware: “Anthropic is integrating OpenShell and BlueField hardware with Claude Managed Agents to give businesses additional control over what their AI agents can access inside execution sandboxes.”
SupportedSalesforce developed a Slack Admin Bridge for OpenShell that lets administrators review and approve agent access requests.Latest from Tom's Hardware: “Salesforce has developed a Slack Admin Bridge for OpenShell that lets administrators review agent access requests, approve or reject permissions, and monitor activity directly in Slack.”