Nvidia built its name selling the expensive chips that power AI. Now it is giving away the software that runs on them.
The company released an open-source AI model called Nemotron 3.5 Lightning on Aug 11, 2026. Nvidia calls it lightweight.
The model runs on a single GPU inside a PC or laptop. A GPU is the chip that does the heavy lifting for AI.
Open-source, in plain terms, means the design is public. Anyone can copy it, change it, and run their own version without paying the creator.
Businesses can download Nemotron 3.5 Lightning, use it, and modify it at no cost, and they do not need Nvidia's permission. It will be available on HuggingFace and Nvidia's website.
Nvidia built it for AI agents, which are programs that can carry out tasks on their own in the background.
This is Nvidia's first open-source model since CEO Jensen Huang started backing the approach in late July, when he made his first-ever post on X.
His argument is simple: free models still need Nvidia's GPUs to run, so open-source AI ends up driving chip sales.
Get the free Always Be Buying eBook and learn the simple system for building wealth on any income
For a company that built its business on selling AI chips, that is a boon for chip sales. Huang put it this way: "Free AI should be great for hardware. Free AI should be great for chips."
The debate got louder in Washington after a Chinese startup called Moonshot AI released a model named Kimi K3. That model reportedly narrowed the performance gap with top American models.
Some policymakers raised concerns about national security and intellectual property theft. The suspected method is called distillation, where a lighter model is trained using responses from a more advanced AI service.
Nvidia representatives said Monday that distillation was also used to give Nemotron 3.5 Lightning abilities on par with its larger Nemotron models. That puts Nvidia's release right in the middle of the same debate.
Huang answered with an open letter arguing that open-weight models, where the AI's design stays open for others to build on, give companies more control, foster competition, and lower prices.
"Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty," he wrote. Sovereignty, in this context, means countries get AI they can control instead of renting it from a few giant companies.
Nvidia also formed an AI safety consortium with Microsoft, focused on applying open models and open-source software to cybersecurity.
Several companies have already tested and customized Nvidia's new model, including CrowdStrike, CodeRabbit, and Harvey.
Nvidia also released software called NeMo Switchyard, which helps identify the lowest-cost and most suitable AI model for a given task.
Why does it matter? Because Nvidia, which trades under the ticker NVDA, is betting that free software sells more chips.
Lighter models that run on a single PC chip could put AI into smaller businesses that could never afford a data center.
For your portfolio, the real question is whether free models grow the whole AI market or just move business between the big players.
Cheaper technology tends to reach more people, and every new user still needs hardware to run it.
If you own AI stocks beyond Nvidia, this debate decides whether the biggest players keep their edge or smaller ones catch up. Either way, the open-source experiment is now part of how AI companies compete.
Download the free Always Be Buying eBook and start putting your money to work today