Put your hardware to work.
Three kinds of operator, one network.
Home Labs
Hobbyists running AI at home, on their personal computers.
A single machine — a gaming GPU, a workstation, an Apple Silicon box. Serve a small model when you are not using it and earn for the work it does. No commitment, no minimum, and nothing to sign.
- —Join with one command on hardware you already own
- —Serve when idle, stop whenever you like
- —Earn per request, paid out in TNZO
Micro Data Centers
Small-scale operators.
A rack, a closet, a colocation cage, or a small room of machines. Enough capacity to serve real workloads and take paying demand, without the overhead of a full facility.
- —Routed demand sized to what you can actually serve
- —Billing, metering, and settlement handled for you
- —Deployment and runtime support from the team that builds the stack
Data Centers
Full-scale operators.
Existing GPU fleets and facility-scale capacity. Keep your stack — if you already run vLLM or SGLang, register the endpoint and keep serving the way you do now.
- —Sustained demand routed at facility scale
- —Your existing serving stack, unchanged
- —Capacity planning, scheduling, and priority support
You run the hardware. We do the rest.
Routing
We connect demand to your capacity. Requests are matched by price, latency, region, and track record — so idle hardware gets work rather than waiting to be found.
Payments
Buyers pay however they want — card, bank transfer, stablecoin, or TNZO. You earn in TNZO. Conversion, escrow, metering, and payout are handled end to end, so you never have to build billing or take card details.
Ticketing
Rentals are issued as tickets: a buyer books capacity for a term, the ticket is redeemed against your hardware, and it closes out with a receipt. Bookings, extensions, cancellations, and disputes run through us.
Credits
Buyers top up credits and spend them as they go. You see committed demand rather than card declines, and settlement is already funded before the work starts.
Distribution
Your capacity is listed where buyers are already looking, and reachable through the same API and RPC endpoints every developer on the network uses.
Marketing
Positioning, listing copy, and go-to-market for operators who would rather run hardware than write a landing page.
Operational support
Monitoring, upgrades, incident response, and a route to the people who build the software you are running.
Measured, not negotiated.
Price
You set what you charge per million tokens or per GPU-hour. Cheaper capacity wins more of the work it can actually serve well.
Latency and throughput
Time to first token and tokens per second are measured per endpoint, not claimed. Faster hardware earns more traffic on latency-sensitive work.
Uptime
Availability is tracked continuously. Consistent endpoints keep normal routing priority; an endpoint that keeps failing gets less traffic until it recovers.
Published, not private
These measurements are visible to the buyers choosing between operators. You compete on what your hardware actually does, and a new operator with good numbers is not disadvantaged for being new.
Or we run it for you.
Managed node
We stand up and run a Tenzro node for you — provisioning, keys, monitoring, upgrades, and on-call. You get the capability and the earnings; we carry the operations.
Managed cluster
A multi-node cluster operated on your behalf, sized to your workload and your power envelope, on your hardware or capacity we source.
Managed network services
Private deployments, dedicated endpoints, and network services run as a service — for teams that need the infrastructure without becoming an infrastructure team.