Originally published at https://mrnasdog.com/research/fet/inflation

FET Inflation Analysis · September 2026 · Supply growing, projected to keep growing

FET, the token of the Artificial Superintelligence Alliance, is described almost everywhere as a capped asset, and on Ethereum that is true: the ERC-20 contract held 2,714,384,546.672 FET at both ends of the last 90 days, unchanged to the wei. The Pressure Framework still reads FET at +15.42% over the trailing 90 days and +15.48% over the next 90, because the Fetch.ai native chain is a second, uncapped ledger — and between Jun 19 2026 and Aug 28 2026 the administrator of its token bridge minted 342.0M FET in six batches and withdrew every coin. Sell pressure is 356.5M FET, buy pressure is 0, and no burn or buyback fired on either chain.

The verdict, in one paragraph

Against a circulating base of 2,311.6M FET, the framework books 356.5M FET of sell pressure and 0 of buy pressure over the trailing 90 days — a net of +15.42% — and projects +15.48% for the next 90 on the same observed cadence. The inflation monitor reads +2.41% for the same window, a gap of 13.02 percentage points, far outside the framework's 0.5pp tolerance, so the overview page ships with a monitor-gap warning. That gap decomposes exactly and leaves no residual: 13.40pp is numerator, because the monitor's circulating series is a classification of the Ethereum leg alone and cannot see coins created on the Fetch.ai chain, and the other 0.38pp runs the other way as base convention, since the monitor divides its supply change by the 90-day-old supply while the framework divides by today's. The label for FET is a token with a hard cap on one chain and a live printer on the other.

Sell pressure: where new FET comes from

It does not come from Ethereum. The FET ERC-20 contract at 0xaea46A60368A7bD060eec7DF8CBa43b7EF41Ad85 reported 2,714,384,546.671999999999656122 FET at block 25,280,789 on Jun 9 2026 and the identical figure at block 25,926,394 on Sep 7 2026, confirmed on five independent archive nodes. That flat reading was tested rather than trusted: the total lives in mutable storage, its value is absent from the contract's compiled code, and a call to the contract's mint function from an unprivileged address reverts with the contract's own message about lacking the minter role. FET therefore has a live, role-gated mint function on Ethereum, held by a single address — the supply is fixed by policy, not by code — and the flat quarter is a real measurement rather than a compiler artefact.

The new FET comes from the Fetch.ai native chain, a Cosmos chain whose mint module runs a fixed 3.00% annual staking inflation with no cap of any kind. Measured rather than assumed — the chain ran about 12.7% slower than its own block reference over the window, and Cosmos issuance is paid per block — that produced 9.8M FET of staking rewards, which is Sell #1, protocol inflation. Sell #2, vesting unlocks, is 4.7M FET, and it is not a cliff: no FET vesting event falls anywhere between Jun 9 2026 and the end of the year on any unlock tracker. What moved is the merger conversion escrow, whose balance fell from 13.7M to 9.0M FET as legacy holders converted. Those coins already existed, so the framework books the realised escrow outflow rather than a schedule, and never counts it as issuance.

The quarter's real story is Sell #5, native-chain issuance outside the bridge, at 342.0M FET. Six times — 50M on Jun 19 2026, 60M on Jul 3 2026, 60M on Jul 25 2026, 62M on Aug 5 2026, 70M on Aug 14 2026 and 40M on Aug 28 2026 — the administrator of the Fetch.ai token bridge contract executed a mint on it, and within minutes withdrew the coins. Seven withdrawals moved 352,745,419 FET out to two holding wallets, which forwarded 240,162,709 and 112,582,710 onward to a single high-throughput wallet that has sent over a thousand transactions this quarter and holds under a thousand FET. The two legs sum to the withdrawals exactly, with no residual. Sell #3 and Sell #4 are both 0: no foundation wallet released anything, and FET has no bankruptcy estate, no trustee and no court distribution.

Buy pressure: where new FET goes

Nowhere. All four buy rows read 0, and each is a measured zero rather than an assumed one. Buy #1, programmatic buyback, is 0: the Alliance's repurchase programme belongs to 2025 and nothing ran in this window, while the Fetch.ai Foundation multisig at 0x8400ac235ed4f139a3e05670a9a3c724e448129b held 26,691,468.09 FET at both ends, identical to eight decimal places, so there is no accumulation to book. Buy #2, protocol fee burn, is 0, and this is the row most write-ups get wrong: the announced buy-and-burn programme exists on paper but did not fire once in 90 days. Both surfaces a burn can appear on were read at both ends. The ERC-20 total was identical to the wei — and this token burns by calling burn and burnFrom, which cut that total directly, so a real burn would have shown there. The unspendable address held 0.47 FET on both dates and the zero address held nothing. The native chain's supply never fell at any sampled height. Neither surface moved, so nothing was destroyed; the last genuine burn, roughly 5.1M FET, sits in early 2026, before this window opens.

Buy #3, foundation buy, is 0 for the same reason the Foundation multisig closes flat. Buy #4, new long-term lock, is 0: nothing new was locked, and staking on the Fetch.ai chain would not remove supply from this reading in any case, because staking is what earns the newly minted coins already counted on the sell side. A token whose sell side is 356.5M FET and whose buy side is a hard zero has nothing structural pulling the other way.

Foundation and overhang

Four team-controlled overhangs are tracked, and each is watched rather than projected. The Fetch.ai Foundation multisig holds 26.7M FET and did not move a unit across the window; it is read from the chain on every rebuild. The Ethereum side of the official bridge holds 214.2M FET in escrow against native coins, under a contract cap of 350M; it is on-chain and read directly. The merger conversion escrow still holds 9.9M FET that legacy holders can draw at will, with no published calendar. Largest and least discussed is the 110M FET minted in 2024 for a partner migration that never completed after that partner left the Alliance in October 2025; a governance proposal to destroy those coins was rejected in November 2025, so they remain live, discretionary and under Alliance control. None of these carries a value in the ledger today, because none of them fired in the window. If any of these balances falls between refreshes, the outflow enters Sell #3 at the next refresh.

How FET compares to other multi-chain AI tokens

The right comparison for FET is not another AI token by market cap; it is any token that exists as a capped ERC-20 on Ethereum alongside a sovereign chain that can issue. A single-ledger capped token — an ERC-20 with a fixed total and a renounced or absent mint path — cannot do what FET did this quarter, because there is no second surface to issue from. A single-ledger inflating chain, the ordinary Cosmos or delegated-proof-of-stake design, issues openly at a published rate that every aggregator reads and prices; its inflation is boring precisely because it is visible. FET sits in the gap between the two, and the gap is the risk: the number the market quotes comes from the capped leg, while the issuance happens on the sovereign leg.

The mechanism that decides whether that gap is real is the bridge design, and FET's is a dual-escrow lock-for-lock bridge rather than a burn-and-mint one. Coins crossing from Ethereum are locked in the Ethereum escrow and released from a matching escrow on the Fetch.ai chain, so genuine crossings move supply without creating it. That accounting closes here to the unit: the Ethereum escrow's own balance fell by 52,755,219.045 FET across the window, and the Fetch.ai side's crossings — 28,415,981.87 paid out against 81,171,200.92 taken in, over 154 and 123 transfers — produce the same figure to six decimal places, with 154 crossings recorded identically on both chains. The bridge balances to zero without the six mints, which is exactly what places those 342.0M FET outside it. Nothing was locked or destroyed on Ethereum to match them, and no Ethereum address gained anything close — the largest 90-day gain among the top holders is 49.6M. Under the Alliance's own bridge source the ERC-20 mint and burn functions exist to reflect changes in native FET supply; over these 90 days that reflection was not applied.

What to watch in the next 90 days

First, the mint cadence on the Fetch.ai bridge contract. Six firings in 90 days at a rising size, from 50M in June to 70M in August, is an observed pattern rather than a published schedule; the framework carries the trailing total forward and does not extrapolate the trend, so the next firing is the number that matters. Second, whether the Ethereum ERC-20 total ever moves to reflect that native issuance — a burn there would convert the finding from additive supply into an accounting mirror, and it would be visible immediately in the contract's total. Third, the 110M FET partner-migration overhang, which is unretired after the burn vote failed and is the single largest discretionary block outside the escrows. Fourth, the announced buy-and-burn programme: it has not fired since early 2026, and a resumption is the only mechanism on the buy side capable of moving this ledger. Fifth, any governance proposal touching the Fetch.ai mint module, which has not been amended since January 2024.

Summary

The Pressure Framework reads FET at +15.42% over the trailing 90 days and +15.48% projected, against a monitor reading of +2.41% — a 13.02pp gap that decomposes entirely into a numerator the monitor cannot see. The structural mechanism is a two-ledger token: the Ethereum ERC-20 is capped, was flat to the wei, and has a live but unused role-gated mint, while the Fetch.ai native chain is uncapped and issued 351.8M FET — 342.0M of it from six administrator mints on the bridge contract that fall outside the bridge's own lock-for-lock accounting, plus 9.8M of staking rewards. The key risk is that the coin's headline supply figure is a classification of the capped leg only, so a reader checking the cap will not see the issuance. The ceiling that matters is not the 2,714,384,546.672 FET printed on the Ethereum contract; it is whatever the Alliance chooses to mint next on a chain that has no ceiling at all.

---

MrNasdog Pressure Framework analysis of FET, Metric 1 — Inflation. Data + explanation only. Not financial advice. Updated Sep 7 2026.