GPU providers
Enterprise customers for your GPUs, at premium prices.
Our agent integrates with your cloud stack and runs the AI ops, so you can offer enterprises far more than GPU hours, without extra effort from your team. The exchange brings you the volume your GPUs need to pay back.
Enterprise workloads
Matched to you on the Covextra Exchange
Covextra agent
Runs the AI ops
Your cloud stack
Such as Canonical OpenStack
Your GPUs
On your sites, under your operation
The service ladder
Our agent takes you up the service ladder, without the effort.
Hourly rental is becoming a commodity, and its price keeps falling. With our agent you sell a complete service, which is worth far more to a regulated enterprise, and your team doesn’t have to build it.
GPU rental by the hour
You sell hours, and hours are becoming a commodity.
Certified, sovereign capacity
Enterprises can filter for your jurisdiction and your certifications.
Attested enterprise AI compute
You sell a complete service: the workload deployed, protected, proven and settled.
Volume
Enterprise demand is what will pay back this generation of GPUs.
The return on a GPU depends on how many hours it sells. The exchange matches you with enterprise workloads you don’t reach today, sized to the capacity you have available, so more of your fleet is earning more of the time.
Neocloud GPUs, 2026
69%
allocated under contract
47%
actively in use
AI Industry Reviews, GPU Utilization Rate Statistics: 2026 Report, 23 September 2026.
Flexible AI ops
Our agent runs the AI ops, around your availability.
It offers the capacity you have free, and adjusts as that changes. Connect your power price feed, and it moves your short-term prices when electricity prices change.
Your team keeps running the data centre.
- 1
Tracks
Follows your available capacity and, with your power price feed, your energy costs.
- 2
Quotes
Answers enterprise requests with a price that reflects both.
- 3
Deploys
Places each workload across your machines.
- 4
Protects
Sets up confidential computing on each node.
- 5
Attests
Produces the hardware-signed evidence the enterprise asked for.
- 6
Settles
Meters usage against the quote and closes the contract, so you get paid.
The economics
The same GPUs earn more.
Enterprises pay hyperscalers $15 to $18 per GPU-hour on demand for the B300s you rent out at about $5 on reserved terms. They’re paying for compliance and data protection, and Covextra lets your GPUs offer both.
Neocloud rental
About $5 per GPU-hour, reserved
Your GPUs on Covextra
Priced by the market, for everything above the GPU
Hyperscaler
$15 to $18 per GPU-hour, on demand
List and reserved prices for B300 GPUs, October 2026. GPU-hour prices are moving quickly, so treat these as indicative. The Covextra bar shows what we add and is not a price.
Confidential computing is where the premium is.
Questions from GPU providers
- Which cloud stacks do you support?
- Our agent integrates with your cloud stack. Most neoclouds run Canonical OpenStack.
- Which contracts can I sell?
- Spot and forward contracts, matched to the capacity you have available.
We’re signing up GPU providers now, ahead of launch in March 2027.