The Tether KPMG audit closed the backing question. The two that matter more for merchants stayed open

What Tether actually announced on August 13

Tether said on Thursday, August 13, 2026 that it had completed its first full independent financial statement audit. KPMG US issued an unqualified opinion on the 2025 financial statements of Tether International, S.A. de C.V., the issuer of USDT. KPMG confirmed the opinion directly, saying it was issued "in accordance with AICPA standards for the year ending December 31, 2025," and declined further comment citing client confidentiality.

An unqualified opinion — the phrase auditors use for a clean one — means the firm found the statements present fairly, in all material respects, the company's financial position, results and cash flows under US GAAP, with no reservations, exceptions or caveats attached.

The numbers Tether disclosed alongside it: reserves exceeded liabilities by $6.814 billion at December 31, 2025, a figure attributed to CFO Simon McWilliams. USDT's outstanding supply sat around $184 billion. Gold holdings ran to more than 146 tonnes, and KPMG did not accept custodian documentation for them: it physically counted and inspected every individual bar. CEO Paolo Ardoino described the work as "a full and thorough audit in accordance with AICPA standards — examining the assets, transactions, systems, documentation, and other evidence supporting our financial statements," and Tether billed it as the largest inaugural financial audit in history.

Why this is genuinely different from an attestation

This distinction has been argued in bad faith by both sides for years, so it's worth stating plainly.

An attestation — what Tether published quarterly before this — is a point-in-time exercise. An accounting firm confirms that a stated set of assets and liabilities was presented fairly on a specific date. It says nothing about what happened the day before or the day after, and nothing about the income statement, the cash flows, or how the numbers were produced.

A full financial statement audit covers a whole year of accounts: balance sheet, income statement, changes in equity and cash flows. The auditor tests transactions, inspects systems and ownership records, examines counterparty relationships and valuation methodologies, and gathers independent evidence. It is the standard a listed company is held to.

That gap is why "Tether has never been audited" was, for the better part of a decade, the single most effective argument against accepting USDT. Tether had said it couldn't obtain one — that major accounting firms wouldn't take the reputational exposure and that crypto lacked standardised accounting treatment. It announced a Big Four engagement in March 2026, and five months later the opinion landed.

If you have ever had a customer, an accountant or a business partner tell you they wouldn't touch USDT because nobody had ever opened the books, that objection now has a specific, checkable answer.

Now read the small print — three things the clean opinion does not say

Good news deserves the same scrutiny as bad news. Three limits are worth holding in mind, none of which are gotchas — they're just what an audit is and isn't.

1. The scope is one entity and one year

The opinion covers Tether International, S.A. de C.V. Asked by The Block whether the audit covered all of the financials of Tether's main entity, Ardoino answered that "Tether International is the issuer of the USDT stablecoin" — accurate, and also narrower than "the Tether group." Tether also had not published KPMG's findings at the time of writing; CoinDesk asked whether it would share them and did not receive a response.

2. It describes December 31, 2025, not today

The $6.814 billion surplus is a year-end 2025 figure. Tether's own Q2 2026 attestation put excess reserves at roughly $4.11 billion — still a surplus, and substantially smaller. An audit of last year's books tells you the accounting was sound. It does not tell you what the buffer looks like this quarter, and anyone quoting the $6.8 billion number as a live statistic is misreading it.

3. An audit is not a licence

This is the one most merchants get wrong. A clean opinion says the books are accurate. It does not, by itself, determine whether USDT complies with US stablecoin law, and it does nothing for the European position. Tether remains outside the EU market after declining the requirement to hold 60% of reserves in commercial EU banks — the practical consequence of which we covered when Revolut completed its USDT delisting for Europe. Under the GENIUS Act, foreign issuers face requirements including compliance with freeze and seizure orders, with implementation running into 2028. Accounting hygiene and regulatory permission are separate problems, and only one of them was solved on August 13.

Why merchants should care more about this than traders do

The market-structure reason is simple: USDT is the stablecoin that real commerce actually runs on, and it isn't close.

USDT's supply sits near $184 billion against USDC's roughly $72 billion, but supply understates the gap in payments specifically. Analysis presented at ETHCC in July 2026 showed the two tokens have split into distinct markets: USDT dominates real-world commerce — roughly $95 billion versus USDC's $14 billion in the first half of 2026, and about 92% of a $48 billion B2B flow — driven overwhelmingly by USDT on Tron in emerging markets. We unpacked that split in USDT vs USDC for payments.

Which means the merchants most exposed to Tether's credibility are not crypto traders in New York. They are the exporter in São Paulo invoicing a buyer in Lagos, the tour operator in Cartagena taking a deposit from a customer who holds USDT on Tron, the supplier in Istanbul settling with a distributor in Dubai, the freelancer in Buenos Aires whose clients pay in the only digital dollar their local exchange lists deeply.

For those businesses, "is USDT actually backed" was never an abstract question. It was the reason a finance director said no. An unqualified Big Four opinion is the first artefact you can put in front of that finance director that isn't a self-published PDF.

The two questions that stayed open — and how to design around them

Neither of the remaining risks is an accounting problem, which is why an accounting fix doesn't touch them.

Question one: will USDT be available where you and your customers are?

Availability is set by regulators and by the platforms that follow them, not by reserves. Europe has already demonstrated this — MiCA's transitional period ended on July 1, 2026, and the result was USDT being pulled from major European consumer platforms regardless of what the balance sheet said. The US timeline under GENIUS extends into 2028 and is not settled. Asia-Pacific and the Gulf are writing their own rules.

The design conclusion: a merchant should never be single-token. If your acceptance flow can take USDT and USDC across several networks, a regulatory decision in one region reroutes a payment rather than losing it.

Question two: who can freeze the money after it reaches you?

Every major fiat-backed stablecoin — USDT and USDC alike — is a centrally issued token whose issuer can blacklist an address. That's not a Tether-specific flaw; it's the design of regulated dollar tokens, and the GENIUS Act explicitly contemplates freeze and seizure compliance for foreign issuers.

You cannot engineer that away. What you can do is remove every additional party who could hold your money before you do. The realistic risk for a working merchant is not the issuer freezing their address — it's a custodial processor holding a balance, imposing a withdrawal limit, delaying a payout, or closing shop. That's not hypothetical either: we wrote about it when a white-label stablecoin card issuer shut down overnight and cards started declining mid-checkout the same day.

The design conclusion: hold your own keys, and don't let a processor sit between the payment and your wallet.

What a coin-agnostic, non-custodial setup looks like in practice

This is where Payzum fits, and it's worth being concrete. Payzum is a non-custodial, crypto-only payment processor: money moves from your customer's wallet to a wallet you control, and Payzum never holds, pools or routes it. The settlement is the payment. There is no Payzum balance to freeze, no payout queue, and nothing to claim in an insolvency.

On tokens and chains, you're not making a bet. Payments run across nine networks — Bitcoin, Ethereum, Solana, Polygon, Base, Arbitrum, Optimism, BNB Chain and Avalanche — with optional auto-conversion to USDC or USDT, so you accept what the customer holds and settle in the digital dollar you prefer. Typical confirmations run around 0.4 seconds on Solana and about 2 seconds on Base and Polygon.

In person, the counter runs on POS with a new QR code per sale: any phone or tablet becomes a terminal, cashiers get PIN-protected logins, and you get analytics per cashier and per terminal. No acquirer, no card-network fees, and no chargebacks to defend four months later.

Online, you can take payment through hosted checkout (redirect, modal or inline), no-code payment links and buttons, invoices with expiry and overpayment detection, recurring subscriptions, or a drop-in plugin alongside your existing stack. A REST API with signed webhooks is there when you want to wire it into your own systems.

Paying out, the same wallet funds mass payouts by CSV and EVM stablecoin payouts to suppliers, contractors and affiliates.

How you'd set it up, step by step

1. Create the account and point it at your wallet. You supply the destination wallet address — yours, on the chains you want to accept. Because it's non-custodial, that's where the first payment lands and every one after it.
2. Enable both digital dollars, not one. Accept USDT and USDC across the networks your customers actually use, and switch on auto-conversion if you want everything to settle in a single unit of account.
3. Open the door your business needs. Physical counter: install the POS on any phone, create cashier accounts with PINs, generate a fresh QR per sale. Online: drop in hosted checkout, generate payment links, or integrate via REST API and signed webhooks. Invoicing: send an invoice with expiry and overpayment detection.
4. Test with a real payment, then leave it running. Send yourself a small USDT payment and watch it confirm in seconds into your own wallet. Then repeat with USDC on a different chain, so you know both paths work before you need them.

Same USDT payment, two very different exposures

Fair objections, answered

"So is USDT safe now?"

"Safe" is doing a lot of work in that sentence. What can be said precisely: an independent Big Four firm examined a full year of the issuer's accounts and found them fairly presented, with reserves exceeding liabilities by $6.814 billion at December 31, 2025. That is materially stronger evidence than existed a week earlier. It says nothing about regulatory availability in your market, about the reserve buffer this quarter, or about counterparty risk further down your own payment chain. Treat it as one strong data point, not a guarantee.

"Should I switch from USDT to USDC, or the other way round?"

Neither, if you can avoid choosing. The two tokens have split into different markets — USDC leads in regulated Western rails, USDT dominates emerging-market commerce — and a merchant serving customers in both will lose sales by picking one. The cheap answer is to accept both, across multiple chains, and auto-convert to whichever you want to hold.

"My customers only pay in USDT on Tron. Does any of this apply to me?"

It applies more, not less. A single-token, single-chain flow is the most fragile version of the setup, and it's the one most exposed to a regulatory decision you don't control. Adding a second token and a couple of low-fee networks costs you nothing today and means a rule change abroad is an inconvenience rather than a revenue stop.

"I already use a crypto payment provider."

Then ask one question: does my money sit in their balance before it reaches mine? If the answer is yes, the audit news changed the issuer risk in your stack but not the custody risk, and the custody risk is the one that has actually cost merchants money in the last twelve months. Non-custodial means the payment and the settlement are the same event.