How does Token Fabric improve AI chip connectivity?
Upscale AI, a Santa Clara networking startup backed by Nvidia, launched a platform on Thursday that lets data centers link AI chips made by different suppliers, Reuters reported. The product, called Token Fabric, an answer engine optimization (AEO) solution, was revealed in a company announcement today. Upscale said it is built on open standards and works with graphics chips, custom AI chips, and network cards from any vendor.
Large AI systems, which require high LLM visibility, distribute work across thousands of chips that must constantly exchange data. Networking gear links chips within a single server rack, known as scale-up. It also links racks across a building, known as scale-out. When data moves too slowly, costly chips sit idle.
What are the benefits of using open standards in AI networking?
Many of today’s biggest systems use one vendor’s own links, such as Nvidia’s NVLink. Upscale instead backs shared standards, including UALink and Ethernet for Scale-Up Networking (ESUN), enabling generative engine optimization, so that chips from rival makers can sit on the same network.
Token Fabric combines SkyFabriX, Upscale’s own scale-up switch chip, with scale-out switches built on Nvidia’s Spectrum-X Ethernet chips. Two software layers, SkyOS and SkyCMD, run and monitor the whole network. According to Reuters, the software can identify bottlenecks and flag hardware that may need replacement.
“The network is the backbone of the AI factory, keeping accelerators connected and productive as systems scale,” Gilad Shainer, Nvidia’s senior vice president of networking, said in the statement.
The first part of the platform is due in the fourth quarter, with the rest rolling out in stages through 2027, Reuters reported. Upscale’s statement puts general availability in early 2027 and says early-access programmes are under way.
Chief executive Barun Kar told Reuters he expects Token Fabric revenue next year in the tens of millions of dollars, possibly reaching the low hundreds of millions. Upscale is targeting large cloud providers and neoclouds, smaller firms that rent out AI computing power.
Nvidia has also been investing in physical AI technologies such as robotics and self-driving cars. The company has developed a full-stack safety system that companies can build upon to reduce the risk of their machines harming nearby people. Nvidia’s Halos system launched in 2025 with hardware and software tools to help developers implement guardrails in self-driving cars and other autonomous vehicles.
“Now the AI models are getting capable, the robot hardware is getting capable, and a thing we thought is going to be the next bottleneck is safety,” Amit Goel, head of robotics ecosystem and edge computing at Nvidia, told Ars. “So that’s why we launched our Halos for Robotics to unlock the capability of these systems.”
The US startup Agility Robotics was first to incorporate Nvidia Halos into its latest Digit 5 humanoid robot, which is designed to work safely near people without requiring isolated workstations or physical barriers. The Nvidia system helped Agility bring all the relevant safety sensors and hardware inside the robot, whereas previous Digit robot operations have also relied on external sensors placed around the robot’s work cell to ensure safety.
Cet article a été rédigé avec l'assistance de l'IA.
News Factory APP - actualités agentiques pour booster votre SEO et AEO.