Understand this story
On October 22, 2026, NVIDIA CEO Jensen Huang proposed at the GTC conference in Berlin that the licensing fee for running NVIDIA's full suite of AI agent deployment software would be approximately $2,000 per GPU per year. Based on a calculation of 5 million GPUs paying the fee, this would bring in $10 billion per year, or $50 billion over five years. The report also noted that NVIDIA did not specify the range of GPUs to which this pricing applies, which software it includes, or whether it is consistent with the AI Enterprise subscription priced at $4,500 per GPU per year; subsequent reports stated that this figure is currently only a price reference and has not yet become a formal, unified pricing standard.
AI-generated from public reports
Latest development · Report
NVIDIA Plans to Charge Annual AI Agent Software Fee per GPU; Multiple Bandwidth Specs for Vera Rubin LoweredDevelopments
1 confirmed developments
Jensen Huang proposes charging software license fees based on the number of GPUs
Source:Phoenix Technology
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Oct 27
Semiconductor Industry Observer
NVIDIA Plans to Charge Annual AI Agent Software Fee per GPU; Multiple Bandwidth Specs for Vera Rubin Lowered
At GTC Berlin, NVIDIA CEO Jensen Huang said that the NVIDIA software fees required to run various AI Agents in an AI factory are approximately $2,000 per GPU per year, but this figure is currently only a price reference and has not yet become a formal unified pricing standard. By comparison, the existing public list price for NVIDIA AI Enterprise is $4,500 per GPU per year. In the same period, according to media comparisons of NVIDIA's product pages and historical archives, on the Vera Rubin platform the Rubin GPU HBM4 bandwidth was adjusted from 22 TB/s to 19.2 TB/s, sixth-generation NVLink bandwidth was adjusted from 3.6 TB/s to 3.0 TB/s, the total HBM4 bandwidth of the NVL72 system dropped from 1580 TB/s to 1400 TB/s, while system network bandwidth increased from 28.8 TB/s to 32.4 TB/s; NVIDIA has not yet issued a formal statement on the specification adjustments.
Oct 26
TechSpot
Jensen Huang says NVIDIA AI software is priced at about $2,000 per GPU per year
During the Q&A session at GTC Berlin, Jensen Huang said that running NVIDIA's full software stack for AI agents costs about $2,000 per GPU per year, and said that software licensing will increasingly become an important part of the company's business. That figure is lower than the list price of the existing AI Enterprise suite at $4,500 per GPU per year (including support), and NVIDIA has not yet explained the relationship between the two or when the new price will take effect. On that basis, every 1 million licensed GPUs could generate $2 billion in recurring revenue per year, while the company's revenue last quarter was $96.2 billion.
Phoenix Technology
Nvidia Targets AI Software Subscription Revenue, Huang Renxun Proposes About $2,000 per GPU per Year
At the GTC conference in Berlin, Huang Renxun said that the license fee for running Nvidia's full suite of AI agent deployment software is about $2,000 per GPU per year. Based on 5 million GPUs paying, this would bring in $10 billion per year, totaling $50 billion over five years; however, Nvidia did not specify the range of GPUs to which this quote applies, which software is included, or whether it is consistent with the AI Enterprise subscription at $4,500 per GPU per year. On October 21, Nvidia also released the OpenShell reference architecture, integrating Vera Rubin compute, Vera CPU, and BlueField-4 DPU equipped with Sentry security components, to define action boundaries for agents.
Jiwei.com
NVIDIA plans to charge a software subscription fee of about $2,000 per GPU per year
At the GTC event in Berlin, Jensen Huang said that the software provided for running AI agents can be licensed by the number of GPUs, with an example price of about $2,000 per GPU per year, but NVIDIA has not yet officially announced the full terms and implementation date. NVIDIA's existing NVIDIA AI Enterprise subscription is $4,500 per GPU per year, or $18,000 over five years, and the scope of software covered by the new price is not yet clear. A day earlier, NVIDIA announced the OpenShell reference architecture, which integrates Vera Rubin, Vera CPU and BlueField-4, and introduced NVIDIA Sentry to monitor unauthorized behavior by agents; Anthropic and Salesforce have already begun cooperation around this architecture.
Heat trend for this story
- Current
- 8
- Period peak
- 24 Oct 27 16:00
- 24-hour change
- -45%
Move the pointer or select the chart to inspect hourly heat. Use the left and right arrow keys with a keyboard.