Perplexity, best known for its AI-powered search engine, is pushing its technology off the internet and onto your own hardware.
According to Gizmodo, the company has launched a local AI model designed to run on your GPU instead of the cloud. Techzine Global reports that the move is tied to Nvidia, framing it as Perplexity opting for local AI in partnership with the chipmaker.
The distinction matters more than it might sound. Most AI assistants today work like a phone call: you type a question, it travels to a data center packed with expensive chips, the answer gets computed there, and it comes back to your screen. A local model flips that arrangement. The computation happens on the graphics card already sitting inside your computer, so the request never leaves your machine.
That shift carries a few practical consequences. Data that stays on your device is data that isn't sitting on someone else's servers. Responses don't depend on a network connection or on how busy a provider's data centers are that day. And for the company running the model, every query handled locally is a query it doesn't have to pay cloud costs to serve.
The Nvidia connection is the other half of the story. Nvidia's GPUs dominate AI computing in data centers, but the company also sells the graphics cards inside consumer gaming PCs and workstations. If AI models increasingly run on personal hardware, those consumer chips become a lot more interesting as an AI platform, not just a gaming one.
The available reporting is thin on specifics — neither headline details which model, what hardware it requires, or how it compares to Perplexity's cloud offerings.
It matters because the AI industry has spent years assuming intelligence must live in giant remote data centers, and a serious push toward running capable models on ordinary personal hardware would change who controls your data, what AI costs, and who profits from it.