Economics
The Auton economy
Auton sells the same thing as a GPU cloud or an inference API — tokens of model compute — but the economics underneath are different. Instead of renting capacity at an opaque, ever-changing spot price, Auton turns compute into a priced, tradable commodity with forward contracts, prepaid discounts, on-chain settlement, and a token whose scarcity is driven by real usage.
Two ways to buy compute
Every AI builder pays for compute one of two ways today:
- Rent a GPU from a cloud (AWS, GCP, CoreWeave) — you pay by the hour whether or not you use it, and prices move with the market.
- Call an API (OpenAI, Anthropic, OpenRouter) — you pay per token at whatever rate the provider sets, postpaid, with a margin baked in that you never see.
Both are spot markets: the price is set by the seller, can change at any time, and the money you spend accrues entirely to that company. Auton adds a third option — a market where compute is priced forward and value flows back to the people who use and secure the network.
How compute is priced today
- Volatile & opaque — rates can change overnight, and the markup over raw cost is hidden inside the sticker price.
- Postpaid & unhedgeable — you find out the bill after you've spent it, and there's no way to lock next quarter's price for a model you depend on.
- Extractive — 100% of your spend becomes the provider's revenue. You are a tenant; you accrue none of the upside of the demand you create.
- Custodial — funds, capacity reservations, and trust all sit on the provider's books.
The Auton model
- Forward pricing — lock a per-model rate for a fixed term with inference futures, or go long/short on compute prices in trading. You can hedge volatility instead of absorbing it.
- Prepaid at a discount — chat credits are metered per token but charged at ~15% below the underlying market rate (a 0.85 multiplier on spot). Paying through Auton is cheaper than going direct.
- On-chain settlement — payments and provider settlements clear in USDC on Solana, verified on-chain. Reservations and balances are protocol state, not a vendor's private ledger.
- Usage-driven token sink — the demand you generate buys back and burns $AUTO and rewards the stakers who secure capacity, so spend accrues to the network rather than a single company.
Side by side
| Dimension | Traditional providers | Auton |
|---|---|---|
| Pricing | Spot, set by seller, changes anytime | Forward rates you can lock, plus discounted prepaid credits |
| Billing | Postpaid — bill after the fact | Prepaid balance, metered per token, ~15% under spot |
| Price risk | You absorb all volatility | Hedge or speculate with futures & leverage |
| Settlement | Off-chain, custodial, trust the invoice | On-chain in USDC, verifiable |
| Where value goes | 100% to provider equity | Buyback & burn $AUTO + staker rewards |
| Capacity guarantee | Best-effort, provider's terms | Backed by staked $AUTO with slashing for SLA breaches |
How value flows
A single payment moves through the protocol like this — and the last step loops back, tightening $AUTO supply as usage grows:
Why demand accrues to $AUTO
In a normal API business, more usage simply means more revenue for the provider's shareholders. On Auton, more usage tightens the $AUTO supply and deepens the security pool:
- Futures positions lock $AUTO as margin — more builders hedging means more $AUTO held off-market.
- Settlement fees fund a buyback & burn: $AUTO paid for chat is burned directly, and USDC revenue is used to buy $AUTO and burn it.
- Providers stake $AUTO to back their SLA and earn a share of every settlement, so capacity growth locks supply too.
The result is a token whose scarcity is tied to genuine compute demand rather than speculation. See The $AUTO Token for the mechanics and the treasury page for live burn and stake totals.
What it means for you
- Builders — pay less per token than going direct, and lock prices so a budget set today survives a market spike tomorrow.
- Providers — monetize spare capacity with on-chain USDC settlements and boosted payouts for staking.
- Token holders — every unit of real inference demand flows back into $AUTO scarcity and network security.