A new AI networking startup emerged from stealth this week with an unusually large war chest for a company that has yet to ship a product. Upscale AI said it has raised more than $100 million in seed funding to build open-standard networking infrastructure for AI data centers, a direct challenge to the proprietary interconnect technology that has helped cement Nvidia's dominance of the AI compute stack.
The round was co-led by Mayfield and Maverick Silicon, according to the company's launch announcement, with additional backing from StepStone Group, Celesta Capital, Xora, Qualcomm Ventures, Cota Capital, MVP Ventures and Stanford University. The Palo Alto, California-based company was incubated inside Auradine, the blockchain- and AI-infrastructure company led by Upscale's executive chairman, Rajiv Khemani.
Betting on open standards over proprietary silicon
Upscale is led by chief executive Barun Kar, who previously co-founded networking chipmakers Cavium and Innovium and helped build Palo Alto Networks, alongside a team the company says numbers more than 100 engineers drawn from Marvell, Broadcom, Intel, Cisco, AWS, Microsoft, Google and Juniper Networks. Rather than building around proprietary interconnects, the company says its network fabrics, operating system and rack-scale hardware will be built on open standards including SONiC, Ultra Ethernet and the Ultra Accelerator Link, positioning it as an alternative to the closed networking stack Nvidia has used to lock AI compute customers into its ecosystem.
Now, our world-class team is applying its depth and breadth of expertise to the next frontier: open AI-native infrastructure. Our deep expertise enables us to architect cohesive AI infrastructure solutions, partnering with leading xPU vendors and AI innovators to meet today's demands and power what's next.
Barun Kar, CEO, Upscale AI
Khemani, who is also chief executive of Auradine, framed the spinout as a deliberate bet on open infrastructure gaining ground as AI buildouts scale. "We incubated Upscale AI with a bold vision to reimagine networking for the AI era anchored in openness, performance, and scale," he said in the announcement, adding that "next-generation infrastructure must keep pace" as AI adoption spreads across industries.
The raise lands amid a broader wave of venture money flowing into AI networking and interconnect startups this year, as hyperscalers and neoclouds look for alternatives to Nvidia's InfiniBand and NVLink technology that could lower costs and reduce single-vendor dependence in increasingly expensive data-center buildouts. Investors including Qualcomm Ventures and Stanford University's endowment joining a $100 million-plus seed round — an unusually large check at that stage — signals the scale of capital chasing the networking layer of the AI buildout specifically.
Upscale has not disclosed a product-shipping timeline or named early customers. The company said it plans to move quickly to bring its networking suite to market across silicon, systems and software, a process that will test whether open standards can match the performance hyperscalers have come to expect from Nvidia's tightly integrated hardware.
