Designed, not live. Nothing trades until two audits are clean, the pilot blockers found in our own risk review are fixed and checked again, counsel has answered, the B300A grade is ratified and default protection is in place and attested. The pilot will be open to certified businesses only.
What this market would do to AKT supply
AKT comes into play in three places in this market. Surplus default protection buys AKT and burns it, service nodes lock their own AKT in bonds from the Akash chain's cutover, and ACT posted as margin keeps AKT locked on-chain. None of the three is switched on yet. The market keeps its accounts in US dollars and has no token of its own. Every figure on this page is an estimate that could change heavily, and none of it is a forecast of AKT's price.
Where AKT enters
The burn, the bond and ACT as margin.
The burn. Once default protection stands above its buffer, the US dollars above the buffer buy AKT on on-chain markets, and the program burns that AKT in the same step, so it leaves the supply for good. The money is what reaches the venue's own part of default protection: the fees left after validators are paid, the premium that certified accounts pay for protection, and fines for defaults. Claims are paid before anything is burned, the burn stops whenever protection falls below the buffer, and no person decides when it runs.
The buffer and the pace. The buffer is the highest protection target of recent months, raised by the target's own recent growth, plus the headroom that certified accounts have already paid toward it, and never less than the starting level. The target grows with the market's open positions, so a market that grows fast reaches its buffer later, and the burn is likely to be small in the early years. A daily budget, which spreads the surplus over about a month and never buys more AKT than the on-chain markets can supply without moving its price more than on a normal day, limits the purchases, and what it cannot spend stays in default protection.
The bond. From the Akash chain's cutover, each service node posts a bond of its own AKT. The bond stays locked while the node serves and for a short unbonding period after it leaves, and it is burned only for a provable fault, such as signing two conflicting reports. Its smallest size is set by formula from what a majority of signers could gain by leaning on the market's reference price for replacement compute, so it grows with the compute in delivery. No one can delegate AKT to a service node, and a node's pay stays a share of fees in US dollars.
ACT as margin. Once ACT, Akash's dollar-pegged compute token, exists on Solana with the four transfer terms the design requires, a certified account may post it for part of its margin, at a haircut set by formula. AKT itself is never margin. ACT is minted against AKT, and Akash's mechanism locks that AKT on-chain rather than burning it, so ACT posted here keeps AKT locked while it is posted. If the account falls short, the program redeems its ACT for AKT and sells that AKT for US dollars, which puts the AKT back into circulation.
Where AKT moves
The estimate
At three sizes of delivered compute.
Delivered compute a year $7M · $70M · $700M
These are examples of size, not a forecast of this market.
AKT bought and burned, and AKT locked in bonds, at three sizes of the market
Each panel has its own scale, from zero to its tallest bar, so the height of a bar in one panel cannot be compared with the height of a bar in the other. The figure under each bar is its value.
Bought and burned in a year, at most
US dollars (estimate)
Delivered compute a year
Most the burn could spend in a year
Locked in service-node bonds, at least
US dollars (estimate)
Delivered compute a year
Least AKT the bond formula locks, once bonds apply
For each size, the bar under "Bought and burned in a year, at most" is about the most the burn could spend in a year: the fees left after validators are paid, with the validators' share at half, a placeholder still to be set. The burn spends nothing until its conditions are met, and in the early years it is likely to spend about $0.1M a year or less. The bar under "Locked in service-node bonds, at least" is the smallest total bond the formula asks of all service nodes at that size. Bonds apply only from the Akash chain's cutover, whose date a later Akash proposal will set. Estimate. Could change heavily.
AKT figures use AKT's price of 0.6098 USD on 8 Nov 2026 (CoinGecko), held constant on purpose. This page makes no forecast of AKT's price.
If the market delivered $70M of compute a year, the burn could spend at most about $1.1M a year (about 1.8M AKT), and service-node bonds would lock at least about $2.6M of AKT (about 4.3M AKT) once bonds apply. Estimate. Could change heavily.
The most AKT bought and burned in a year, and the least AKT locked in service-node bonds, at three sizes of delivered compute a year. Every figure is an estimate that could change heavily.
| Delivered compute a year | Most the burn could spend in a year, in USD | The same in AKT | Least locked in service-node bonds, in USD of AKT | The same in AKT |
|---|---|---|---|---|
| $7M | about $0.11M Estimate. Could change heavily. | about 0.18M AKT Estimate. Could change heavily. | about $0.26M Estimate. Could change heavily. | about 0.43M AKT Estimate. Could change heavily. |
| $70M | about $1.1M Estimate. Could change heavily. | about 1.8M AKT Estimate. Could change heavily. | about $2.6M Estimate. Could change heavily. | about 4.3M AKT Estimate. Could change heavily. |
| $700M | about $11M Estimate. Could change heavily. | about 18M AKT Estimate. Could change heavily. | about $26M Estimate. Could change heavily. | about 43M AKT Estimate. Could change heavily. |
There is no line for total supply. Total supply also depends on Akash's own issuance after the cutover, which is not yet decided, and on AKT's price, which this page does not forecast.
The rules
The rules, in plain words.
-
The burn. Only the surplus above a buffer is spent, on AKT that is bought and burned. Claims are paid before anything is burned. The program, not a person, decides.
Source: white paper §10; specification §7.7, decision D110.
-
The buffer. The buffer is the highest protection target of recent months, raised by its recent growth, plus headroom certified accounts have paid, and never less than the starting level.
Source: white paper §10; specification §7.7, decision D110.
-
The budget. A daily budget, which spreads the surplus over about a month and is tied to how much AKT can be bought on-chain without a large move in its price, limits the purchases.
Source: white paper §10; specification §7.7, decision D110.
-
The bond. From the chain cutover, each service node posts a bond of its own AKT, and a bond is burned only for a provable fault.
Source: white paper §12; specification §11.5, decision D111.
-
No delegation. No one can delegate AKT to a service node.
Source: white paper §12; specification §11.5, decision D111.
-
ACT, not AKT, as margin. ACT may count toward part of a certified account's margin once it exists on Solana with the transfer terms the design requires. AKT is never margin.
Source: white paper §7; specification §7.11, decision D112.
Before the burn runs
What the burn waits for.
The burn waits for three things. A pending correction to how the protection target is shared among certified accounts must be written into the specification. The market must have a record of about 90 daily closes of its own. Default protection must also stand above both its target and its buffer, with no claim unpaid. The burn pauses while the AKT behind any posted ACT is being sold to cover a shortfall. Counsel is reviewing the burn.
What this leaves out
- New AKT that Akash issues.
- The end of the old chain's issuance at the cutover, and the AKT that holders now lock with Akash's validators, whose role in securing the chain ends then.
- Spending that moves from Akash's lease market to this market. Akash's mechanism for paying leases locks AKT on-chain for as long as the lease credit it issues is held. Delivery here is paid in US dollars and does not use that mechanism, so where spending moves here, less AKT may stay locked there. This page does not estimate how much.
- ACT posted as margin here. Each ACT posted keeps AKT locked on-chain, but how much will be posted cannot be estimated yet.
- Competition for service-node seats, which can push bonds above the minimum.
None of these is netted against the figures above.
Notes
- How the figures are built. Each figure applies the market's rules to one size of delivered compute a year. The design worked them out for H100A nodes, before B300A became the opening contract, and they could move with that change.
- Why the burn bar is a ceiling. The burn can spend only what the venue's own part of default protection holds above its buffer. The bar counts all the fees left after validators are paid, with the validators' share at half, a placeholder still to be set. The premium and fines that also reach that part are small next to those fees once protection stands above its target.
- Why the bond grows with the market. The formula sizes each bond to what a majority of signers could gain by leaning on the market's reference price for replacement compute, which grows with the compute in delivery. At the starting settings the total is about 3.7% of a year's delivered compute, in US dollars of AKT. Estimate. Could change heavily.
- Status. Every rule on this page is written into the market's specification, and none of it is switched on. Every figure is an estimate that could change heavily.
Three ways in
Do you run, need or price GPU compute?
You run B300s.
Tell us your nodes and monthsYou need B300s.
Tell us what you run and whenYou make markets.
Tell us what you'd need to quote