AI WORLD HQ
Have you ever wondered whether AI has to run in a distant data center, or whether it can run on the computer already on your desk? NVIDIA's latest announcements point toward the second option, with a focus on Europe.
What NVIDIA announced
NVIDIA is expanding what it calls its family of open models. An open model is an AI model whose trained files are published so that developers and companies can download, study, and adapt them, instead of only using them through a company's website. NVIDIA says the goal is to give developers, researchers, and businesses what they need to build apps that work in local languages and fit specific industries.
The family now covers more than language. NVIDIA says it spans robotics, biology, and quantum computing as well. Among the new language models are Nemotron 3.5 Super VL and Nemotron 3.5 Bolt. NVIDIA also introduced new Nemotron Retriever models, which handle embedding and reranking. An embedding model turns text into numbers that capture its meaning, so a computer can find related content. A reranking model then sorts those results so the most relevant ones come first. Think of it as a librarian who first pulls every possibly useful book off the shelf, then puts the best ones on top.
At GTC Berlin, NVIDIA's developer conference in Berlin, the company showed its lineup of local AI hardware. "Local" means the AI runs on the device itself, not only in the cloud. NVIDIA describes three products:
- RTX Spark, which launched last week. NVIDIA says it powers personal AI agents on Windows PCs. An agent is an AI that can carry out tasks for you, not just answer questions.
- DGX Spark, which NVIDIA says supports developers building and scaling applications on Linux.
- DGX Station, which NVIDIA says serves enterprise teams running demanding AI workloads on Linux. NVIDIA says a Windows version arrives in December.
Why this matters to you
Running AI locally can mean more control over where your files and questions go, because the work stays on your device. NVIDIA's announcement does not go into detail about privacy, so treat that as a general idea to check with the product's own documentation.
Local AI also helps with language. Models built for specific languages and industries aim to understand local wording better than a general tool might.
Wrap-up
NVIDIA's message is that AI should work closer to where people live and work, on their own PCs and in their own languages. The company says it is adding open models for many fields and new hardware for both developers and everyday users. Your next step is simple: if you use a Windows PC, look up RTX Spark on NVIDIA's official site and read what it says about personal agents.