Nvidia CEO Jensen Huang said running the company’s software for AI agents costs about $2,000 per GPU per year, a remark that has investors weighing how large a software business the chipmaker could build.
Huang made the comment at Nvidia’s GTC Berlin conference Oct. 22, Tom’s Hardware reported. The remark has circulated widely online since, as investors debate the economics of the artificial intelligence buildout.
“Software licensing will become increasingly a major part of our company’s business,” Huang said. “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.”
Nvidia has not said which software the figure covers, which GPUs would be eligible or when such pricing would take effect. It also has not announced a price change.
The company already sells software by the GPU. Its AI Enterprise suite is listed at $4,500 per GPU for a one-year subscription and $18,000 for five years, according to Nvidia’s licensing guide.
Huang’s figure is less than half that one-year price. TweakTown noted that $2,000 doesn’t match anything Nvidia currently sells and called it more of a ballpark than an actual price tag, cautioning against reading it as a price cut.
Why the Number Matters
The math explains the interest. Five million GPUs licensed at $2,000 a year would bring in $10 billion annually, an illustrative calculation that Notebookcheck stressed is not a company forecast.
Over five years, with customers renewing, that would total $50 billion. Nvidia has not disclosed how many GPUs might fall under such a license.
The remark followed Nvidia’s Oct. 21 announcement of its OpenShell Reference Architecture, which pairs Vera Rubin systems and Vera CPUs with BlueField-4 data processing units running a tool called Nvidia Sentry.
OpenShell is meant to give AI agents a secure runtime that limits what tools and data they can reach. Sentry monitors agent activity and can quarantine an agent that tries to cross those limits, according to Nvidia.
AI agents are programs that can use tools, pull data and carry out tasks with limited human direction. Businesses running them at scale need software to manage the workloads and keep agents away from resources they shouldn’t touch.
Huang has argued Nvidia is becoming more than a chip supplier, expanding into networking, systems and enterprise software. “One of these days, people might not categorize us as a chip company anymore,” he said on the BG2Pod podcast.
The software talk comes as investors question AI infrastructure spending. Shares of cloud computing providers CoreWeave, Nebius and Iren fell more than 4% in premarket trading Monday before paring losses, CNBC reported.
Huang told CNBC last week that demand for such providers is “off the charts” but that capital is scarce. “The demand of the capital is much greater than the availability of the capital. That’s going to get solved,” he said.
Content created by Conservative Daily News and some content syndicated through CDN is available for re-publication without charge under the Creative Commons license. Visit our syndication page for details and requirements.
