Verifying Falcon USDf: What a Holder Can Actually Check, and Where the Trail Goes Private

After a year of synthetic-dollar depegs, the question holders actually ask before depositing is no longer "what is the APY." It is narrower and more useful: "if I wanted to check the backing myself, how far could I get before I had to trust someone." Most coverage never answers it, because answering it means picking one real token and walking the trail until it goes dark.

So we did. This is a live read of Falcon USDf, one of the larger synthetic dollars in the market, with roughly 1.26 billion USDf in circulation as of June 30, 2026. We picked Falcon because it has a genuinely public reserve surface to walk, not because it has a problem. We are not claiming USDf is under-backed. Falcon's own disclosures and an independent accountant say the opposite, and we have no evidence otherwise. The only question here is epistemic: which parts of the backing can a holder verify without anyone's permission, and which parts require trusting a third party. In plain terms, what does Falcon USDf's proof of reserves actually let you prove yourself?

The rules we held ourselves to: name Falcon's real method accurately, allege no wrongdoing, link every claim to a primary source, and end on the same boundary in our own product. If you hold a synthetic dollar, the skill this post is really teaching is how to run this walk yourself, on any of them.

What you can read on-chain in two minutes

Start with the part that needs no one. USDf is an ERC-20 on Ethereum at 0xFa2B947eEc368f42195f24F36d2aF29f7c24CeC2. The contract is verified on Etherscan as a transparent upgradeable proxy (implementation 0x3aDf34C09DAC24E4BAeFB1b1df4C2992edC2b789, Solidity 0.8.28), so you can read its source and its live state directly. As of June 30, 2026 it reports a total supply of 1,261,794,575.93 USDf across 6,436 holders, trading at about $0.99.

The staked token, sUSDf, sits at 0xc8CF6D7991f15525488b2A83Df53468D682Ba4B0. It is a verified ERC-4626 vault, so its share-to-asset ratio is a live on-chain value: about 1.123 USDf per sUSDf at the time of writing, which matches the sUSDf-to-USDf value Falcon shows on its dashboard. Three commands against an Ethereum RPC get you all of it.

Aggregators agree with the chain: DefiLlama shows roughly 1.26 billion USDf in circulation on June 30, 2026, about 1.147 billion on Ethereum and 114 million bridged to BNB Chain, at a market cap near $1.255 billion. This half of the picture is fully verifiable. The amount of USDf that exists, the claim on the reserves, is a number you can pull yourself and reconcile against three independent sources in a couple of minutes. Plenty of tokens cannot say even that.

Where the dashboard takes over

Now turn the question around. You have verified how much USDf exists. What backs it? Falcon publishes the answer at app.falcon.finance/transparency, and to its credit the page is detailed. As of the June 30, 2026 read it shows total reserves of about $1.3 billion, a protocol backing ratio of 103.32 percent that drifts daily as crypto prices move, and a full reserve breakdown by asset and by custody location.

The composition is Bitcoin-heavy: BTC at 64 percent, plus two wrapped-Bitcoin assets (MBTC at 14.5 percent and ENZOBTC at 14.2 percent) for roughly 93 percent in Bitcoin and Bitcoin derivatives, then stablecoins at about 4 percent and a long tail of smaller tokens. The custody columns are labeled Ceffu at 2.63 percent, Fireblocks at 2.25 percent, and Multisig at 95.1 percent. That is a more granular reserve disclosure than most synthetic dollars give you, and it is updated daily.

Here is the pivot. Every entry in that reserve table is a dollar figure, not an address. We checked which links on the page actually resolve to a block explorer, and there are exactly three: the USDf token, the sUSDf token, and a single insurance-fund address (0x432CDcc4516B21302985b639Ef9a7853727A4e49). The roughly $1.23 billion sitting in the Multisig column (about 95.1 percent of the roughly $1.3 billion reported on the June 30, 2026 read), the Bitcoin that makes up the bulk of the backing, the custodial balances at Ceffu and Fireblocks: none of those are exposed as addresses you can open and read for yourself. You can see what Falcon reports it holds. You cannot do to the assets what you just did to the supply.

This is not concealment. Off-exchange and multi-party-computation custody is a standard, deliberate institutional design, and the bulk of the backing is Bitcoin, which does not live on the Ethereum block explorer you used for the supply anyway. The point is narrower, and it is the whole point of this post: at the reserve table, you stop reading the chain and start reading Falcon's measurement of the chain.

Who does the verifying, and when

If a holder cannot re-derive the assets, who does? Falcon's answer is a named professional firm, and this is the strongest part of its setup. The reserves are attested by HT Digital Ltd, Chartered Accountants and Statutory Auditors, and the digital-asset division of Harris and Trotter LLP, a BKR International member firm. The dashboard hosts the actual signed reports as downloadable PDFs, alongside weekly reserve reports going back to mid-2025.

We read the latest one in full. The report dated as of 30 April 2026, signed by a Senior Statutory Auditor on 19 June 2026, is a reasonable assurance engagement under ISAE 3000 (Revised). That distinction matters and it is to Falcon's credit: reasonable assurance is the higher of the two levels the standard defines, above the limited assurance that many crypto attestations stop at. The procedures listed are not cosmetic. The firm states it reviewed the controls over creation of and access to the reserve wallets, verified the ownership and control of the custodial and non-custodial addresses holding the reserve assets, recalculated the aggregate reserve value, and confirmed it exceeded the USDf in issuance at the reporting timestamp ($1,676,378,223 of reserves against 1,625,018,685 USDf issued, as of 30 April 2026 23:59 UTC). We want to be precise here, because it is easy to be unfair: this is a real audit by a real firm that did real work, and nothing in it suggests the reserves are anything other than what Falcon says.

What it is, though, is a point in time. The report says so itself, in its emphasis-of-matter paragraph: the reporting date is limited to a single timestamp, and the firm "did not perform procedures or provide any assurance at any other date or time." And Falcon's own reserves report draws the line for holders explicitly. It tells you that you can view the USDf in circulation "through the public Ethereum network block explorer anytime," while the USD value shown on the dashboard "is aggregated from various independent third party market data sources." That is an honest description, and it is exactly the boundary this post is about. The supply is yours to check, continuously. The reserves are checked for you, by a firm, at points in time, with a daily dashboard in between that reads from market-data feeds rather than from a proof you can reconstruct.

The exact point you must trust someone

Put the walk together and the handoff is precise. Up to the supply, you trust no one. At the reserve table, you take on three things at once, none of which is a flaw and all of which are real:

1. The custodians and exchanges hold what is reported. Falcon names BitGo, Fireblocks, Ceffu, and ChainUp as custodians, with trading on venues including Binance and Bybit. You are trusting those balances exist and are controlled by Falcon, as the attestor confirmed at its last reporting date.
2. The off-chain strategy holds the positions it reports. The dashboard's own strategy allocation shows the yield engine is not passive collateral: roughly 61 percent options-based strategies, plus funding-rate farming, staking, and several arbitrage books. Those are open positions on centralized venues. They are not on the chain at all, and no holder can read them.
3. The point-in-time snapshot still holds today. Between attestations, you are trusting that nothing material has changed since the last timestamp a professional signed off on.

This is also the fair reading of Falcon's one notable wobble. On July 7 and 8, 2025, USDf slipped its peg, with a representative low around 80 basis points on aggregated pricing and deeper prints on thin single-venue DEX pools, before recovering to about $0.998 within days. It coincided with a liquidity withdrawal and a public debate about reserve composition, in which the risk firm LlamaRisk, in a May 2025 assessment that predated the depeg, had argued that the team held unilateral operational control of the reserves and that too much sat off-chain to verify independently. Falcon's response was to expand exactly the disclosure this post just walked: the granular dashboard, the weekly attestations, and the move to reasonable-assurance reporting. The episode was a confidence and liquidity event, not a reserve shortfall, and it is the clearest illustration there is of why holders care where the trail goes private. When you cannot check the assets yourself, your only tools in a scare are someone else's report and the depth of a DEX pool.

Kerne's own trust boundary, disclosed here

It would be dishonest to run this walk on someone else and not on ourselves. Kerne issues a synthetic dollar too, kUSD, and it has the same class of boundary. The difference is where we put it and how loudly we label it.

On the reserve leg, a kUSD holder is closer to the trustless end than the USDf walk above. Kerne's collateral lives on Base and is readable with raw on-chain calls, and we publish an hourly, re-derivable proof of reserves signed with an EIP-191 key, so you can reconstruct the on-chain reserve figure yourself rather than read our measurement of it. There is no third-party attestor standing between you and that figure. You can run it at /verify and against /api/por.

Then comes our private point, and it is the direct analogue of Falcon's. Kerne's delta-neutral hedge runs on Hyperliquid, a single venue. Those hedge positions are self-reported and signature-bound. They are not something you can independently re-derive on-chain the way you can our Base reserves. That is our attestor-equivalent: the exact line where a kUSD holder stops verifying and starts trusting us. We are also pre-audit, at Genesis scale, and exposed to funding-rate risk. We are not telling you we removed the trust boundary that every off-chain-hedged synthetic dollar has. We are telling you we moved it to the smallest surface we could and named it on the same page as the part you can check. The full field, including where Kerne ranks and where it does not, is laid out in the synthetic-dollar scorecard, and our own open gaps live at /legible.

What proof of reserves actually proves

None of this makes Falcon a bad synthetic dollar. By the standards of the field its transparency is strong: a granular daily dashboard, weekly reserve reports, a signed reasonable-assurance report from a named accountancy, and a supply you can read straight off the chain. If anything, USDf is a fairer target precisely because it tries harder than most.

The lesson is the one underneath the marketing word. "Proof of reserves" almost always means a professional firm proved it, at a point in time, and you are reading their number. That is genuinely valuable. It is also not the same thing as a holder re-deriving the number from the chain whenever they want, and the gap between those two is where most of the synthetic dollars that failed in the last two years actually failed. Knowing which one you are holding, for any token you hold, is what actually matters. We ran that check on Falcon above, and named our own boundary on the same page. Do it to us, do it to them, and deposit into whichever one survives the walk.

Sources for this piece are linked inline: the USDf and sUSDf contracts on Etherscan, the live Falcon transparency dashboard and its signed HT Digital assurance report, and DefiLlama for cross-chain supply. For Kerne's own verification surface, see /verify, /api/por, and the synthetic-dollar scorecard.