Rafay launches Token Factory for token-based AI billing

Rafay Systems has launched Token Factory, a product that lets GPU providers sell token-metered access to AI models and services.

The launch adds metering, pricing, quota management and access controls to the Rafay platform for operators such as neoclouds and sovereign AI providers. The goal is to help them offer AI services through application programming interface endpoints rather than rely solely on GPU rental.

Rafay is targeting a market in which AI use is shifting toward token-based consumption. That model is already common among large model providers, where customers pay by usage rather than reserving raw infrastructure.

The shift becomes more significant as AI agents take on longer, more complex tasks. Users consume larger volumes of tokens over time, creating a commercial opportunity for operators that own GPU capacity but have not built their own charging and access systems.

Market shift

Token Factory is designed to help infrastructure operators compete on service delivery rather than hourly hardware pricing. It lets them expose AI models through API endpoints and apply usage rules and pricing for enterprise and retail customers.

Rafay said the system has been validated to work with OpenClaw and NVIDIA NemoClaw, which it described as frameworks associated with heavy token consumption. Under this model, users can connect existing agent workflows to the operator's API endpoint and consume AI services on a token basis.

The complexity of GPU hardware, scaling and connectivity remains hidden from end users. For operators, the focus shifts to controlling access, assigning quotas and tracking consumption across users, applications and agentic workflows.

That approach reflects a broader shift in the economics of AI infrastructure. Much of today's token spending goes to hyperscale cloud operators and foundation model developers, while regional infrastructure providers often remain tied to the lower-margin hardware rental market.

Rafay is positioning Token Factory as a way for those regional operators to capture more of that spending. It is also aimed at sovereign AI projects, where governments and local cloud providers want AI services hosted within their own jurisdictions.

Industry context

Interest in token-based charging has been rising across the AI sector. NVIDIA Chief Executive Officer Jensen Huang recently highlighted "tokenomics" as a central theme, describing tokens as a new commodity in AI consumption.

Market forecasts suggest room for growth if the model becomes more widely adopted. Research and Markets projects the GPU-as-a-Service market will reach USD $7.36 billion in 2026 and expand to USD $26.43 billion by 2031. IDC also projects that by 2028, 60% of multinational firms will split their AI stacks across sovereign zones.

Those trends matter for operators building national or regional AI platforms. If customers want to keep workloads in-country while buying AI through a usage-based model, local providers may need the kind of billing and governance layer that hyperscalers already run internally.

Rafay says it has worked with AI factory operators across six continents. It cited customers including Cassava Technologies in Africa, Firmus Technologies in Australia and Telus in Canada, alongside deployments in the Middle East, Latin America and Southeast Asia.

The launch builds on Rafay's broader business in infrastructure orchestration for AI and cloud-native workloads. Its platform is used to add self-service automation, governance and multi-tenancy to compute infrastructure, including Kubernetes environments and AI deployments.

For customers considering how to commercialise GPU estates, the key challenge is moving from provisioning to monetisation. Token Factory addresses that part of the stack by adding a ready-made commercial layer on top of model access.

"Token Factories are the new cellphone companies," said Haseeb Budhani, Chief Executive Officer and Co-Founder of Rafay Systems. "Similar to how cellphone companies used to sell pre- and post-paid minute plans, AI factories are beginning to sell pre- and post-paid token plans. Team Rafay is looking forward to supporting the success of a thousand AI factories across the world with our Token Factory offering."