Reserve with confidence
Commit the capacity your planning depends on, with the reservation guarantees your roadmap requires.
AI Commons
AI Commons is a marketplace for reserved capacity. When your cluster is not running your workloads, it runs for others — and the value flows back to you.
What it is
Reserve the capacity your roadmap needs. Keep it for your workloads. And when it sits idle, let it work for you.
Commit the capacity your planning depends on, with the reservation guarantees your roadmap requires.
Windows your team is not using are offered to the marketplace — automatically, or on your own schedule.
Reserved capacity generates value during idle periods, offsetting the cost of capacity you hold for peak demand.
How it works
01
Commit GPU capacity sized to your roadmap — with the guarantee that it is yours when you need it.
02
Use your reservation for training, fine-tuning, and inference exactly as you would any dedicated environment.
03
When your cluster is not running your work, it becomes available to the marketplace — under your control.
04
Compensation for shared capacity returns to your account, turning reserved cost into revenue.
Capacity sharing is strictly opt-in. Your capacity is never co-tenanted with other tenants while your workloads run — isolation is logical. Between workloads, the environment is logically re-provisioned and data is sanitized. Commons capacity is available only to vetted enterprise customers.
Impact on the ecosystem
What is good for one customer's utilization is good for every customer's economics.
Capacity that once sat idle now serves real workloads — raising utilization across the entire platform.
Teams that need short, bursty compute get it from the commons, without long-term commitment.
Higher utilization lowers the cost base for everyone — including the price of your own reserved capacity.
Tell us about your reserved-capacity needs and usage patterns. A senior member of our team will respond within one business day.