Economics
Validator costs, owner operating costs, revenue planning scenarios, and why emissions are not revenue.
The figures on this page are planning estimates from the August 2026 baseline. They are not guarantees, quotes, or forecasts.
Validator economics
The V1 parameter and runtime caps keep validation inside one-GPU economics.
| Item | Estimate |
|---|---|
| GPU-hours per daily round, at maximum planned scoring load | 10–16 |
| GPU cost per round, at the reference cloud rate | $18–29 |
| Monthly GPU cost | $550–875 |
| Total monthly operating cost, including artifact storage and bandwidth | $600–925 |
Owner economics
| Item | Amount |
|---|---|
| Initial private corpus (one-time) | ~$15,000 |
| Data rotation | ~$3,000 / month |
| Subnet and reference-model engineering | ~$12,000 / month |
| Commercial API infrastructure (after launch) | ~$2,000 / month |
| Total owner opex after launch | ~$17,000 / month |
Revenue planning scenarios
| Scenario | ARR | Operating assumption |
|---|---|---|
| Bear | $1.5M | Pre-recorded educational and corporate media only |
| Base | $5.5M | Production quality in two fixed or asynchronous verticals |
| Bull | $30M | Conversational quality expands addressable live use cases |
These are planning scenarios, not guaranteed outcomes. The controlling variable is measured accuracy and generalization in a commercially selected vertical.
Emissions are not revenue
This distinction is deliberate and load-bearing.
Emissions finance research competition. They compensate miners and validators for producing and measuring model improvements. They are R&D funding, not sales.
Commercial revenue comes from customers. It depends on a deployable champion, product infrastructure, sales, contracts, and measurable accessibility value.
The two systems connect because emissions fund the R&D engine that creates the commercial artifact. They are not interchangeable accounting lines.
Investment thesis in one paragraph
Signet is not a bet that one team has already solved sign-language translation. It is a bet that the problem is measurable enough to fund continuous distributed search, and that the winning artifact becomes economically useful before a conventional small research team could discover and maintain the same result alone.
Four reinforcing assets:
- The subnet mechanism continuously buys model improvement.
- The private rotating corpus measures real generalization and is difficult to replicate quickly.
- The champion artifact history creates a transparent progression of deployable models.
- The commercial API converts research output into external revenue.
The next capital-intensive step is not inventing the protocol. It is acquiring the private evaluation asset, hardening the implementation, registering the subnet, and turning an open model competition into a commercial accessibility product.