Crypto's 2026 Shakeout: What's Surviving the Die-Off?
KPMG issued an unqualified opinion on Tether's 2025 financials, confirming $6.814 billion in excess reserves, in the same year more than 100 crypto projects folded. The open money read: the industry is being graded now, and the tests are revenue, reserves, and redemption.
Summary: On August 13, KPMG issued an unqualified opinion on Tether International's 2025 financial statements, confirming that reserves exceeded liabilities by $6.814 billion as of December 31, 2025, across a book of roughly $183.6 billion in token liabilities.
Auditors counted the gold bars by hand. This is the examination critics have demanded since 2017, and it arrived in a year when more than 100 crypto projects shut down, went dark, or filed for bankruptcy, including BitMEX and BitMart in a single week alongside Movement Labs and Storj.
BitMEX, which invented the perpetual swap and ran it for eleven years, has told customers to withdraw by August 26. The same week, Miden and Circle detailed USDCx, a privacy-preserving dollar backed one-to-one by USDC in an onchain reserve contract, due at mainnet in late August. And in Washington, the market structure bill that would settle who regulates what went into recess unresolved.
Thesis: 2026 is the first year the crypto industry sat a real exam, and the pass marks are specific: recurring revenue collected in dollars, reserves an auditor will sign, and the capacity to honor redemptions under stress.
Tether's KPMG opinion and the hundred closures are the same event seen from two ends. What makes the week worth a strategy team's attention is where those tests get written. Almost all of them sit offchain, in statute, in agency rulemaking, and in the professional judgment of four accounting firms.
An industry whose founding claim was that you would never again have to take an institution's word is being sorted by how well it satisfies institutions. The open question for the next two years is whether these systems can move verification back onchain, where anyone can check it continuously, instead of once a year, for a fee, through one firm.
What the die-off is actually selecting for
The closure list is long and it is not random. Moonbeam's parachain went offline on July 31. Tally, which built governance tooling for DAOs, closed. Step Finance and Everclear wound down. Stream Finance collapsed last November. Four names went in a single late-July week: BitMEX, BitMart, Movement Labs, and Storj Labs. Exploits took $1.1 billion in the first half of 2026 on top of that.
Set against the survivors, the sorting logic is legible. Hyperliquid crossed a billion dollars in cumulative fees by June 30. Aave holds north of $12 billion in deposits and generates more than $100 million in annualized fees. Ether.fi carries $7.8 billion. The shared trait across the list is dull and decisive: these businesses collect revenue in dollars and stablecoins from people using the product, rather than paying for their own growth in a token they print.
Every wind-down this year was a solvency event dressed in different clothing. A perpetual futures venue with no volume is a fixed cost base and a compliance department. A storage network paying more in emissions than it earns in storage fees is a subsidy program with a whitepaper. The subsidy stopped in 2026, and the businesses underneath either had customers or did not. Nobody legislated these closures, which makes this the closest thing to open pressure the sector generates.
The item circulating hardest this week said the opposite. Analytics firm Santiment reported a spike in social posts pairing crypto with "dead" and "finished," and the coverage read it as a contrarian bottom signal.
The report publishes no figures: no magnitude, no baseline, no dated comparison to a prior cycle. A sentiment claim with no denominator is a mood, and it inverts everything else on this page. The rest of the week's news is about the arrival of measurement.
The audit
Tether's full-scope audit by a Big Four firm covering a balance sheet backing $183 billion of circulating dollars, with an unqualified opinion and physical verification of the gold, is a different artifact from the quarterly attestations the company published for years. Attestations confirm a snapshot someone hands you. An audit tests the process that produced the snapshot. Tether spent a decade insisting the distinction did not matter and then went and got the one that does.
Size the claim honestly. USDT sits at $183.0 billion of a $287.1 billion stablecoin market, which is 63.7% of all circulating stablecoins and roughly nine tenths of one percent of the $19.5 trillion in deposits held at US commercial banks in the week ending August 5. The entire stablecoin sector is about 1.5% of American bank deposits. Tether is a large private financial institution operating at roughly the scale of a mid-sized regional bank, and it should be described that way.
Here is where the week turns. The GENIUS Act's implementing rules, proposed by the OCC in March and the FDIC in April, set out what a permitted payment stablecoin issuer above $50 billion must do: monthly public reserve disclosures, annual audited financial statements, quarterly condition reports, reserve segregation, documented liquidity risk management, and a redemption plan that survives more than 10% of outstanding issuance being presented inside 24 hours.
The rules also prescribe which assets count, and the list runs to cash, Federal Reserve balances, Treasuries, and money market instruments. Gold is not on it.
So Tether cleared the examination its critics spent nine years demanding, and the passing grade moved while the paper was being marked. The annual audit has become table stakes, and the harder tests are composition and redemption capacity, neither of which an unqualified opinion on last year's balance sheet answers. The honest verdict is hybrid, with the gate named clearly. The token moves on open rails anyone can watch. The reserves behind it are a private balance sheet you learn about once a year, from a firm the issuer pays.
Proving solvency without publishing the book
The most architecturally interesting thing this week was small and almost entirely unnoticed. Miden, a zero-knowledge chain that spun out of Polygon in April 2025 with backing from a16z crypto, 1kx, and Hack VC, is launching USDCx alongside its mainnet at the end of this month. The design is worth understanding even if the thing never gets traction.
USDCx is backed one to one by USDC, held in an onchain reserve contract Circle operates. Transactions execute and prove on the user's own device, so balances, counterparties, and history stay private by default.
The user can then produce a proof of a specific fact to a specific party, an auditor, a regulator, a counterparty, without publishing everything else. Prove solvency without disclosing positions. Prove a payment cleared without disclosing the book it came from.
That is a direct answer to the problem the Tether audit exposes. The current arrangement asks a stablecoin issuer to open its books to one firm, once a year, and asks everyone else to accept the summary. Selective disclosure proposes that verification be a continuous property of the system, available on demand to whoever has standing to ask. The gap between those two models is the whole argument this newsletter has been making since spring, and it now has a working example with a launch date.
Two honest qualifications. USDCx inherits the reserve question whole, because the backing asset is USDC: Circle can freeze, and the dollars sit in the traditional system regardless. And nothing has shipped. A pre-mainnet privacy dollar with no volume is a design document with a marketing budget. Hybrid on the settlement asset, genuinely open on the verification mechanism, once that mechanism exists.
The size test
The third filter is permission to scale. The Bank of England's June policy statement on sterling-denominated systemic stablecoins dropped the individual holding caps it had floated, £20,000 for people and £10 million for businesses, which the industry had fought hard.
In their place sits a temporary issuance guardrail of £40 billion per systemic coin, with backing composition fixed at up to 70% short-dated gilts and the remaining 30% in non-interest-bearing central bank deposits. Feedback closes September 22, the code of practice lands at the end of this year, and the regime goes operational in 2027.
The comparison that sizes it: the largest dollar stablecoin already carries more than four times what the Bank of England will permit any single systemic sterling coin to issue. The Bank has said it expects to loosen and eventually remove the guardrail once it is satisfied that credit provision is not at risk, which is a candid way of saying the cap exists to protect bank funding rather than consumers.
Nothing about this rule is open, and it should be scored that way. It is the official system naming the gate, in public, with a number attached. That clarity has value. A builder can now model a UK sterling stablecoin business against a hard ceiling and a known reserve mix, which beats modeling against a regulator's mood.
The contrast with Washington is instructive. The CLARITY Act, which would hand the CFTC authority over spot crypto markets and pull exchanges under the Bank Secrecy Act, missed the Senate's last working day before the August recess.
The House passed its version thirteen months ago. Senate Banking approved one in May. The floor time never appeared. American grading therefore continues through agency rulemaking and enforcement, which leaves the tests being written by the people who administer them.
The Eurodollar comparison, and where it fails
The obvious historical analogue is the Eurodollar market. Dollars accumulated outside the United States after the war, Regulation Q capped the interest domestic banks could pay, and capital controls made onshore dollar banking more expensive still.
Banks and corporations used the offshore pool anyway, and a parallel market for dollar funding grew up beside the official one, eventually getting large enough that regulators had to write rules for it rather than close it.
The parallel holds on the mechanism. Constrained official systems produce parallel systems, and parallel systems get regulated into existence. The Bank of England's guardrail and the GENIUS Act rulebook are that second phase arriving on schedule.
The analogy breaks in the place that matters most this week. The Eurodollar market grew precisely because nobody could see it. Its scale was estimated for decades and never really measured, and that opacity worked as a feature for participants and a recurring problem for everyone else, most memorably in 2008, when the size of offshore dollar funding needs turned out to be knowable only after the fact.
An open financial system wins by being cheap to check. That is the distinction worth carrying out of the comparison, and it is why the Miden design deserves more attention than its market cap will justify for a long time.
The honest case against this read
The scale objection lands first and lands hard. All of this is small. The stablecoin sector is 1.5% of American bank deposits. USDCx has no users because it does not exist yet. A hundred failed projects is a rounding error against the failure rate of any early industry, and the survivors' fee revenue, impressive by crypto standards, would be a modest quarter for a regional bank. Anyone arguing that this week reveals something structural about global finance is arguing from a very small base. The defense is narrow and should stay narrow: the claim here is about which properties survive selection, and selection is legible at small scale. Direction is observable before magnitude is.
The costume objection cuts closer. Tether's reserves are a private balance sheet audited by a firm it pays. USDCx inherits Circle's freeze capability. The Bank of England's guardrail is the old system deciding how much of the new one it will tolerate. A skeptic can look at this week and conclude that crypto is being domesticated on entirely conventional terms, with the same auditors, the same regulators, and the same gatekeepers, and that the ledger underneath is decoration. That reading is available and it is not stupid.
The response is about where the gate sits and what remains behind it. A USDT holder can verify their own balance and move it without asking Tether, and the audit changed what is knowable about the reserves while leaving the rail exactly as it was.
A USDCx user will hold an asset whose transfer logic is inspectable code, subject to a published rule instead of an institutional policy. Neither is a clean win, and hybrids should be scored by naming the gate rather than by declaring victory. The gate this week is reserve quality and issuer solvency, sitting at the front door. What sits behind it, transferability, verifiability, and custody the issuer cannot silently revoke, is the part the last twenty years of fintech never gave up.
A third objection comes from consistency. This publication discounted announcement-driven news repeatedly through the spring, and a pre-mainnet privacy dollar is exactly that. Fair. It is here as a design worth understanding, not a development worth pricing.
What to watch
The first read is Tether's response to the reserve composition rules. The gold is the tell. If the company begins rotating out of ineligible assets ahead of the GENIUS deadlines, it has decided to compete for the American regulated market. If it keeps the gold and optimizes for jurisdictions that allow it, the largest dollar stablecoin has chosen to be an offshore instrument on purpose, and the sector splits into onshore and offshore tiers with different rules and different customers. That fork is the most consequential open question in stablecoins right now.
The second is whether USDCx ships in August and whether anything moves through it. The metric that matters is whether a business with a real compliance obligation uses selective disclosure to satisfy an auditor. Total value locked will say nothing useful here. One payroll run or one treasury operation settled that way carries more evidentiary weight than a billion dollars sitting idle.
The third is the closure list through the fourth quarter. If the pace holds and the failures keep being subsidy businesses rather than revenue businesses, the selection story is confirmed. If the failures start including protocols with real fee income, then the diagnosis is wrong and something harder is happening to the demand side.
The fourth is CLARITY after the recess. A market structure statute moves the grading from agency discretion to law, which changes the planning horizon for anyone building a regulated crypto business in the United States. Another six months without one leaves the CFTC and SEC drawing boundaries case by case.
Strategic implications
For builders. Build the thing that is checkable. The businesses that survived this year charge fees to customers in dollars, and the designs that will matter next let an outside party verify a claim without being handed a report.
Those are one instinct applied at two levels: a real business can skip persuading the market to believe a story, and a verifiable system can skip persuading a counterparty to accept one. If your product's honesty depends on an annual document, you have built a company that will be graded like a bank while carrying none of a bank's advantages.
Treat proof generation, selective disclosure, and continuous reserve verification as product features rather than compliance overhead, because in eighteen months the procurement questionnaires will ask for them by name.
For capital allocators. The screen this year is dull and it works: recurring fee revenue collected in dollars, from identifiable users, with costs that do not scale with token emissions.
Apply it and most of the 2026 casualty list would have been excluded in advance. On stablecoin exposure specifically, reserve composition has become a regulatory variable as much as a credit one, and an issuer holding assets the American rulebook excludes carries a policy risk no audit opinion addresses. Size positions against redemption capacity under stress, which is the test the rules impose. The excess reserve figure is the number the marketing leads with.
For policymakers. Two of this week's three filters were written by regulators, which is a considerable amount of influence over what an emerging financial infrastructure becomes. Use it precisely. The Bank of England's approach, a public number, a named rationale, and a stated intention to loosen it, is better practice than the American pattern of leaving structural questions to enforcement while the statute waits for floor time.
The larger opportunity is to recognize that continuous onchain verification could eventually do work that annual audits do badly, and to write rules that accept a proof where they currently require a report. A supervisory regime that can only read PDFs will keep pushing this industry toward the least verifiable version of itself.
Daniel McGlynn
Last week's issue, ICYMI.
Sources
[1] CryptoPotato. "Tether Clears First Full Audit as KPMG Issues Unqualified Opinion on 2025 Statements." August 14, 2026. https://cryptopotato.com/tether-clears-first-full-audit-as-kpmg-issues-unqualified-opinion-on-2025-statements/
[2] Tether. "Tether Completes the Largest Inaugural Financial Audit in History." August 13, 2026. https://tether.io/news/tether-completes-the-largest-inaugural-financial-audit-in-history/
[3] Coinpaprika. "Tether Clears First Full KPMG Audit, Gold Bars Counted by Hand." August 2026. https://coinpaprika.com/news/tether-clears-first-full-kpmg-audit-gold-bars/
[4] CryptoSlate. "Tether finally got the audit critics demanded, just as Washington changed the test." August 2026. https://cryptoslate.com/tether-finally-got-the-audit-critics-demanded-just-as-washington-changed-the-test/
[5] CoinDesk. "Crypto is going through a massive dot-com style shakeout as over 100 projects fold in 2026." August 9, 2026. https://www.coindesk.com/business/2026/08/09/crypto-is-going-through-a-massive-dot-com-style-shakeout-as-over-100-projects-fold-in-2026
[6] CoinDesk. "BitMEX and BitMart may be first casualties of crypto trading slump." July 28, 2026. https://www.coindesk.com/business/2026/07/28/bitmex-and-bitmart-may-be-first-casualties-of-crypto-trading-slump
[7] CoinDesk. "Miden bets on privacy stablecoins with introduction of USDCx." August 12, 2026. https://www.coindesk.com/tech/2026/08/12/miden-bets-on-privacy-stablecoins-with-introduction-of-usdcx
[8] FDIC / Federal Register. "GENIUS Act Requirements and Standards for FDIC-Supervised Permitted Payment Stablecoin Issuers and Insured Depository Institutions." April 10, 2026. https://www.federalregister.gov/documents/2026/04/10/2026-06974/genius-act-requirements-and-standards-for-fdic-supervised-permitted-payment-stablecoin-issuers-and
[9] OCC. "GENIUS Act Regulations: Notice of Proposed Rulemaking." Bulletin 2026-3, March 2026. https://www.occ.gov/news-issuances/bulletins/2026/bulletin-2026-3.html
[10] Bank of England. "Sterling-denominated systemic stablecoins: policy statement and draft Code of Practice." June 2026. https://www.bankofengland.co.uk/paper/2026/ps/sterling-denominated-systemic-stablecoin
[11] Global Government Finance. "Bank of England drops individual caps in stablecoin rules rethink for 2027." June 2026. https://www.globalgovernmentfinance.com/bank-of-england-stablecoins-policy-statement/
[12] Federal Reserve. "Assets and Liabilities of Commercial Banks in the United States, H.8." August 14, 2026. https://www.federalreserve.gov/releases/h8/current/default.htm
[13] StableCoin.com. "Stablecoin Market Cap." Accessed August 16, 2026. https://stablecoin.com/market-cap/
[14] Bitcoin Foundation. "CLARITY Act: What's Next, and Why August 10 Could Be Crypto's Biggest Deadline in 2026." August 2026. https://bitcoinfoundation.org/news/regulation/clarity-act-whats-next-why-aug-10-could-be-cryptos-biggest-deadline-in-2026/
[15] Startup Fortune. "Crypto Just Lost BitMEX, BitMart, Movement Labs and Storj in One Week." August 2026. https://startupfortune.com/crypto-just-lost-bitmex-bitmart-movement-labs-and-storj-in-one-week/
[16] Bitcoin World. "Santiment: 'Crypto Is Dead' Sentiment." August 14, 2026. https://bitcoinworld.co.in/santiment-crypto-is-dead-sentiment/