Protopia and Rafay deliver multi-tenancy for shared GPU AI factories

Coverage from SiliconANGLE's livestreaming video studiohelp_outline UPDATED 10:25 EDT / JULY 30 2026 INFRA Protopia and Rafay deliver multi-tenancy for shared GPU AI factories by Jonathan Anthony Enterprise AI infrastructure providers are turning to multi-tenancy paired with upstream data protection to convert idle GPU capacity into secure, token-metered services that enterprises will actually adopt at scale. That tension is pushing infrastructure providers toward a model that treats utilization and security as two sides of the same equation, as the market shifts from raw GPU rental toward token-metered access packaged and abstracted across models via a consistent application programming interface. A partnership between Protopia AI Inc. and Rafay Systems Inc. aims to expand the addressable market for enterprise AI infrastructure by pairing multi-tenant token delivery with upstream data protection, according to Eiman Ebrahimi (pictured, left), chief executive officer of Protopia AI. “One of the biggest realizations for us … is that often [operators are] thinking about how do they increase the amount of supply that they have,” Ebrahimi said. “But when you dig into how you are operating the underlying factory … you see that there’s a lot of over-allocated capacity. Just because something is booked, if it’s not busy, that’s revenue that they’re leaving on the table.” Ebrahimi and Haseeb Budhani (right), co-founder and chief executive officer of Rafay Systems, spoke with theCUBE’s John Furrier at theCUBE + NYSE Wired: Robotics & AI Infra Leaders event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how a virtual private token factory can expand enterprise AI infrastructure by uniting multi-tenancy, privacy and cost efficiency. (* Disclosure below.) Multi-tenancy emerges as table stakes for enterprise AI infrastructure At the core of the partnership is Rafay’s Token Factory, which lets operators deliver metered, serverless access to open-source models without dedicating GPUs to each customer. Rafay made Token Factory generally available in April, delivering token-based access to AI models and services, arriving as the GPU-as-a-service market is projected to reach $7.36 billion in 2026. Protopia’s technology adds a data-isolation layer that keeps each enterprise’s context private on shared hardware, Budhani noted. “Now we have a slice of essentially a confidential GPU,” Budhani said. “As a provider, I get to monetize every second of my infrastructure, but for the enterprise, it’s actually a better deal. They get a better price point, because the infrastructure is now shared holistically, versus on a time-sliced basis.” That isolation hinges on Protopia’s Stained Glass technology, which transforms prompts and context out of raw plain text before they reach a model. The approach is meant to complement zero-data-retention policies so that even inadvertent leaks don’t expose readable information, Ebrahimi noted. “What Stained Glass is meant to do is to create an inference privacy layer that’s way upstream,” Ebrahimi said. “It transforms the data from that raw representation into … a stochastic re-representation … that can be consumed by the target model without ever needing to go back to the original form.” For operators, the economics only hold at high degrees of multi-tenancy — a point echoed across the industry as agentic AI inference surges and tokenomics becomes the enterprise’s defining budget constraint. By packaging security alongside consumption, providers can turn governance into a revenue driver rather than