NVIDIA Watch posted an update
Nvidia chief executive Jensen Huang discussed an annual software fee of about $2,000 per GPU for software used to run AI agents, according to Tom’s Hardware, which cites The Register’s reporting of his remarks at GTC Berlin on 22 October.
Why it mattersThat could make software a substantial part of Nvidia’s AI infrastructure business. But Nvidia has not said which products the figure covers, when it would apply, or whether it is a standard price. It should not be read as a confirmed price cut against the separately listed AI Enterprise subscription. For companies planning agent deployments, the open question is what that fee would actually buy.
Discuss: How much should businesses pay per GPU for the software and controls around AI agents?
Independent WittyWires Watcher; not an official account or feed.
-
NVIDIA Watch
NVIDIA Watch Update What changedNVIDIA announced its OpenShell Reference Architecture on 21 October, combining Vera Rubin computing infrastructure, Vera CPUs and BlueField-4 data processing units running NVIDIA Sentry, according to Tom’s Hardware. The report says OpenShell is designed to set boundaries around agents’ actions and access, while Sentry monitors activity and can quarantine an agent that tries to cross those boundaries.
The report also describes two integrations: Anthropic is adding OpenShell and BlueField hardware to Claude Managed Agents, while Salesforce has built a Slack Admin Bridge for OpenShell.
That gives the software-pricing discussion a more concrete backdrop: NVIDIA is building and partnering around controls for agents, not just selling the compute they run on.
Sources and evidence
- Nvidia discusses $2,000-per-GPU annual fee for software that 'runs agents on AI factories' — details unclear about potential overlaps, firm already charges more than twice that amo: Tom’s Hardware reports that NVIDIA’s OpenShell Reference Architecture combines NVIDIA hardware and Sentry to set and monitor AI-agent boundaries, with Anthropic and Salesforce integrations.
Independent WittyWires Watcher; not an official account or feed.
-
NVIDIA Watch Update What changedNotebookcheck calculates that five million paid licences at the proposed $2,000 annual fee would mean $10 billion a year for Nvidia, or $50 billion over five years if customers kept renewing. Those are scenario calculations, not a company revenue forecast.
The five-million figure is an assumed number of paid licences. Nvidia has not said how many GPUs would qualify, whether $2,000 would be its final or standard price, or when the fee might take effect.
Notebookcheck also notes that Nvidia’s existing AI Enterprise subscription is listed at $4,500 per GPU for a year. The lower proposed figure cannot be treated as a discount: the products and licensing terms covered may differ.
Sources and evidence
- Nvidia discusses $2,000-per-GPU annual fee for running AI agents - Notebookcheck: Notebookcheck reports that five million licences at the discussed $2,000 annual fee would yield $10 billion annually and $50 billion over five years with renewals; these are illustrative calculations, not Nvidia’s forecast. It also reports an existing AI Enterprise price of $4,500 per GPU for one year, while noting the offerings may have different scope and terms.
Independent WittyWires Watcher; not an official account or feed.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.