CUDA and AI software stack
Open Agent Safety Platform (OpenShell runtime and BlueField-based Sentry) launched September 28, 2026 with over 100 launch partners. NVIDIA does not disclose software revenue separately.
What it is
CUDA, Nemotron open models, Omniverse, and the new OpenShell and Sentry agent runtime and safety layer. It is sold mostly as part of the hardware business and as enterprise subscriptions.
Why it matters to the stock. Software is the reason customers keep buying NVIDIA hardware, and it is a source of recurring revenue that the market values more highly than chips. NVIDIA does not report software revenue, so the value is hard to size.
News
NVIDIA shows Omniverse, Isaac Sim and Alpamayo simulation workflows for developers
NVIDIA's October 8 developer post walks through simulation workflows built on Omniverse, Isaac Sim, Newton and Alpamayo, with work involving PTC Onshape, Unitree and NASA.
SoftBank, AT&T and Indosat cite NVIDIA open models in telecom AI plans
NVIDIA's October 6 telecom post says SoftBank Corp., AT&T and Indosat Ooredoo Hutchison are building AI on open models, with SoftBank using Nemotron for its telecom model.
NVIDIA features four Inception breast-cancer AI startups built on MONAI
NVIDIA's October 5 post profiles four startups from its Inception program: iSono Health, Whiterabbit.ai, Ataraxis AI and SimBioSys.
NVIDIA adds 64GB DGX Spark at $4,999, with two-unit clustering for 200B models
NVIDIA's October 2 post adds a 64GB unified-memory DGX Spark alongside the existing 128GB model, starting at $4,999.
NVIDIA launches Open Agent Safety Platform with OpenShell and Sentry
NVIDIA released an open platform and reference design for governing AI agents, built from OpenShell, an open-source runtime, and Sentry, a watchdog running on BlueField-4 DPUs.
NVIDIA adds CUDA-Q Logical orchestration layer for fault-tolerant quantum work
On September 14, 2026 NVIDIA added CUDA-Q Logical, an orchestration layer in its open-source CUDA-Q platform, for designing and testing fault-tolerant quantum applications.
NVIDIA and Palantir launch sovereign AI supply-chain stack, starting with NVIDIA's own operations
On September 10, 2026 NVIDIA and Palantir announced a collaboration that places NVIDIA Nemotron models inside Palantir Foundry and AIP on the Palantir Ontology.
NVIDIA launches Ising open quantum AI models for calibration and error correction
NVIDIA released Ising, which it calls a family of open quantum AI models.
NVIDIA launches Nemotron Coalition of open frontier AI labs
NVIDIA formed the Nemotron Coalition, an open-model group whose inaugural members include Mistral AI, Perplexity, LangChain, Cursor and Thinking Machines Lab.
NVIDIA launches NemoClaw stack for OpenClaw agents on local RTX and DGX systems
NVIDIA announced NemoClaw on March 16, 2026, a stack for the OpenClaw agent platform that installs Nemotron models and the OpenShell runtime in one command.
NVIDIA and Eli Lilly launch up to $1 billion AI co-innovation lab
NVIDIA and Eli Lilly announced a co-innovation AI lab in the San Francisco Bay Area, where Lilly scientists will work alongside NVIDIA AI engineers on drug discovery.
NVIDIA expands BioNeMo drug discovery platform with Thermo Fisher, Chai and Boltz
NVIDIA said life sciences companies are adopting its open BioNeMo platform, naming Eli Lilly, Thermo Fisher Scientific and a long list of model builders, robotics firms and lab-automation companies.