Crypto daily market recap - Aug 27, 2026

BTC----NVDA----

Nvidia delivered everything the AI trade wanted and bitcoin still could not take 80,000

August 27 is the cleanest natural experiment this market has offered in months. Nvidia reported after Wednesday's close and beat on every line that matters. Revenue of 96.22 billion dollars against a 92.17 billion consensus, up 106% year over year and 18% sequentially. Data center revenue of 89.0 billion, up 117%, which is now functionally the entire company. Adjusted earnings of 2.22 dollars against 2.10 expected. Net income of 59.69 billion. Then the guide: 108 billion dollars for the current quarter plus or minus 2%, against a street looking for about 104.2 billion. On the call, CFO Colette Kress said the company expects fiscal 2028 revenue growth of about 70% while analysts were modeling 44%, and said customer forecasts point to growth doubling next year with guidance held back by supply constraints, not demand. She also flagged an additional 2 million GPUs going to Amazon's cloud arm across 2027 and 2028.

The equity market did what you would expect. Nvidia had closed the regular session on August 26 down about 1.6% at 209.66, its eighth decline in nine days. On August 27 it opened around 222.94 and traded up to 226.40, closing near 225.72, a gain of about 7.66%. Semiconductor ETFs rose about 2.5% before the bell. Memory names ran harder. Nasdaq 100 and S&P 500 futures jumped. The AI trade got its answer.

Bitcoin got nothing. It sat at roughly 79,960 through the session, a few hundred dollars under the 80,000 level it took on Tuesday for the first time since May 15. Total crypto market cap was about 2.70 trillion, down about 0.47% over 24 hours. Dominance held near 59.2%. Fear and Greed cooled to 71 from 74 earlier in the week. The daily relative strength index read 81.71, still deeply overbought. Daily pivot sat at 79,675, R1 at 80,805, S1 at 78,831, and price spent the day pinned between them.

That divergence is the story of the day, and it is more informative than another 2% candle would have been. When the single biggest risk-appetite catalyst on the calendar lands in your favor and your asset does not move, the asset is not trading risk appetite. It is trading something else. This week that something else is the funding of US government debt, and the reason it stalled is that Wednesday's inflation print made the funding story harder, not easier.

The two prints that actually set the tape

Work backwards through the 48 hours, because both drivers are still live.

July PCE landed Wednesday morning. Core came in exactly as forecast at 3.3% year over year and 0.2% month over month. Headline came in at 3.7% against 3.6% expected, with a 0.2% monthly increase against a 0.1% forecast. That 0.1 percentage point headline miss is what mattered. Bitcoin had printed an intraday high near 81,235 to 81,265 on Tuesday, its best level since May. It lost 80,000 within hours of the release and traded near 78,000 by Wednesday's Wall Street open. Gold, which had run to roughly 4,713 an ounce on Tuesday, fell 1.38% to about 4,594.71 on Wednesday, then traded near 4,645 on Thursday.

The rest of Wednesday's 8:30 bundle argued against easing too. Q2 GDP's second estimate held at 1.5% annualized, unrevised, versus 2.1% in Q1. But personal income rose 0.4% against 0.2% expected. Durable goods orders jumped 1.1% against 0.4%. Corporate profits surged 9.1% quarter over quarter on Haver's compilation, the fastest since Q2 2021, or plus 8.2% and 400.9 billion dollars on another cut, against a 0.5% forecast. Consumer spending rose 0.2% while real personal consumption was flat at 0.0%. Personal consumption expenditures in the GDP accounts were revised up to 3.4% annualized from 3.2%. Inventories subtracted 0.7 percentage points for a fifth straight quarterly decline, an unusual thing to see outside a recession, and net exports subtracted 1.1 points on a 12.5% import surge.

Read that mix honestly. Underlying private demand is stronger than the headline growth number, corporate earnings are accelerating, and inflation is stuck near double the target. That is not a combination that gets you rate relief, and rate relief is what the August rally was built on.

Then Nvidia confirmed on Thursday that the AI capex cycle is still spending. Strong nominal growth plus sticky inflation is the argument for higher-for-longer, not lower. The 10-year yield sat at about 4.67% on Thursday, up nearly 2 basis points from Wednesday's close. The dollar index rose 0.1% to about 99.22. Gold slipped 0.2%. Bitcoin flatlined. The AI trade and the debasement trade decoupled on the day the AI trade won.

Where the rally came from, in numbers

The full picture matters because the composition determines what can repeat.

Bitcoin rose about 23% to 24% in the week through August 23, its largest weekly gain since the November 2024 post-election move and the largest nominal weekly dollar gain on record, taking it from roughly 63,000 to 64,000 up to Tuesday's 81,265. From the cycle low near 57,000 the move is closer to 40%.

The squeeze half of that was extraordinary. K33's head of research Vetle Lunde put August 19 short liquidations at 1.37 billion dollars, a record and nearly double the prior peak from July 2021. August 21 added another 739 million in shorts. Two-day totals across Thursday and Friday exceeded 4 billion, with roughly 1.2 billion of shorts inside a 1.4 billion total across 156,211 traders in one 24-hour window, and a single 25.13 million dollar bitcoin position closed on a decentralized perpetuals venue. Perpetual open interest fell to 284,000 BTC, the lowest since May, and funding returned to neutral. Ether recorded its largest short liquidations in history. Solana short liquidations ran about 75% of its liquidation total. Historically, squeezes of this scale cluster near cycle bottoms, which is the constructive read; the mechanical read is that forced buyers cannot be forced twice.

The flow half can repeat, and it has. US spot bitcoin ETFs are now on an eight-session inflow streak through August 26:

• August 17: +297.6 million

• August 18: +189.3 million

• August 19: +517.2 million

• August 20: +606.3 million, with 503.0 million into BlackRock's IBIT

• August 21: +307.5 million

• August 24: +337.6 million

• August 25: +314.4 million, with 284.4 million into IBIT, about 90.5% of the category

• August 26: +232.1 million, with 200.8 million into IBIT against a 50.4 million redemption from Grayscale's GBTC

August month-to-date bitcoin ETF inflows reached about 3.03 billion dollars through August 25, roughly 390 million short of October 2025 with four sessions left in the month. Year-to-date net outflows have been more than halved to about 2.26 billion. Category net assets stood at 98.63 billion on August 26 with cumulative net inflows since the January 2024 launch at 54.59 billion. Daily traded value fell to 2.26 billion on Wednesday from 5.36 billion on Monday, which is worth noting: flows held, but turnover cooled.

Ether funds are on the same streak. They took 179.8 million on August 25 with BlackRock's ETHA supplying 146.44 million, Fidelity's FETH 25.75 million and BlackRock's ETHB 7.61 million, with no fund reporting an outflow. Ether ETF net assets closed near 14.88 billion. August ether inflows reached about 1.06 billion. The week through August 21 brought 697.2 million, the largest weekly figure of 2026.

The breadth extended past the top two. Solana funds drew 33.49 million on August 24 and 32.25 million on August 25, led by Bitwise's BSOL, taking cumulative Solana ETF inflows to a record near 1.22 billion with category assets around 1.27 billion. Weekly Solana ETF inflows of about 65.74 million were the strongest of 2026. XRP funds added 23.87 million on August 25 after 13.82 million the prior session, with August totals near 80.74 million against 27.29 million in July. HYPE funds gained 7.51 million. Grayscale launched a Zcash product with about 314 million in assets. Across bitcoin, ether, Solana and XRP, August 25 alone drew more than 550 million, with BlackRock's two flagship funds accounting for roughly 430.8 million of it.

Aggregate ETP flows for the week hit 31,740 BTC equivalent, the highest since October 2025. Spot and perpetual volumes across the market rose 188%. CME bitcoin futures volume rose 152% and annualized basis climbed to 11.1%, the highest since January 2025. CryptoQuant's Bull Score jumped from 30 to 80 in one week, its highest since October 2025, with eight of ten indicators bullish and spot demand growing at its fastest monthly pace since late December. Spot and futures demand rose together for the first time since October 2025.

What is fragile underneath

Two things.

First, concentration. One issuer supplied about 90.5% of bitcoin ETF flows on August 25 and about 87% on August 26. On the ether side a single issuer took 78% of category flows on one late-August session. A bid this concentrated is one allocation decision away from stalling. It is real money, but it is not broad money.

Second, positioning has flipped and rebuilt. Ether open interest fell roughly 500,000 ETH over the week as shorts were cleared, and the estimated leverage ratio dropped to 0.74, its lowest since early March. That sounds healthy, and on the derivatives side it is. But bitcoin open interest rebuilt into the mid-50 billions, funding turned positive again, and liquidation tapes flipped: one 24-hour window this week showed bitcoin longs at 56.85% of a 51.42 million total while ether and Solana were still short-dominated at 63% and 75%. Earlier in the week the whole-market split was 222 million longs against 172 million shorts out of 394 million. The crowded side is now long in bitcoin specifically, which is the asset carrying the index.

On-chain, unrealized profits are elevated and exchange inflows have increased, which CryptoQuant flags as latent selling pressure. Realized profit hit about 1.72 billion on August 21, the highest daily figure since late November 2024. Against that, ether exchange reserves fell about 101,000 ETH on the week despite those unrealized gains, which is genuine net accumulation, and ether net taker volume flipped positive. Bitmine disclosed 5,847,611 ETH as of August 24 after adding 32,447 in a week, a position worth roughly 14.3 to 14.9 billion at an average cost near 2,440.

So the honest summary is that spot demand is real and improving, derivative leverage was cleaned out and then partially rebuilt, and the marginal buyer is dangerously concentrated in one asset manager.

Warsh is the last gate this week

Jackson Hole opened August 27 and runs through August 29 under the theme "Financial Innovation: Implications for Payments and Policy," with roughly 120 central bankers from more than 70 countries. Fed Chair Kevin Warsh delivers his first keynote as chair on Friday morning. Most schedules list 10:00 a.m. Eastern; at least one previewer published 8:00 a.m., so treat the exact minute as unsettled and the morning as the window.

The substance is unusually tangled. US federal debt passed 40 trillion dollars. The 30-year yield reached its highest since 2007 earlier in August, printing near 5.337% on August 18, before easing to about 5.17% by August 25. The 10-year was 4.736% on August 21 and about 4.64% to 4.67% this week. The two-year fell from 4.24% to 4.17% over the same stretch, which is a front end backing away from a hike even as the July minutes read hawkish and four regional Fed boards favored one. The July decision itself was a 9-3 hold with three members voting to hike.

Meanwhile Treasury is doing the work monetary policy has not. Bessent doubled long-end buybacks from 2 billion to at least 4 billion per operation effective September 9, then let officials brief CNBC that the roughly 950 billion to 1 trillion dollar General Account could fund them. Gold is up about 15% in August, on track for its strongest month since 1999. The dollar index is heading for a third consecutive monthly loss even while it rose against all six other majors on August 25, which tells you the marginal dollar buyer is trading fiscal confidence rather than the policy path.

The specific thing to listen for is not September. It is how Warsh characterizes the long end. If he frames higher term premium as the market usefully doing part of the Fed's tightening, that is a hawkish outcome dressed as humility, and it argues for yields staying high and the scarce-asset bid facing a stiffer discount rate. If he treats long-end stress as a financial stability problem the Fed should address through portfolio management, that validates the trade that carried bitcoin from 63,000 to 81,000. TD Securities framed dollar risk as skewed modestly lower, on the logic that failure to address inflation credibility would hurt the dollar more than a hawkish clarification would help it. The ACM 10-year term premium near 82 basis points sits well under the roughly 150 average of the pre-quantitative-easing decades, so there is room for that channel to keep running.

Policy and industry

The September 15 cloture vote on H.R. 3633 is unchanged. The motion to proceed ripens at 2:15 p.m. Eastern, the Senate returns September 14, and cloture needs 60 votes in a chamber with 53 Republicans, 45 Democrats and 2 independents. Ethics language, stablecoin rewards and Agriculture Committee jurisdiction remain unresolved, and any Senate amendment sends the bill back to the House.

One CLARITY-adjacent story is worth flagging because it is being reported loosely. Cryptex Finance filed a pre-effective amendment on August 24 for a proposed Digital Market Cap ETF that assigns XRP a 4.88% fund weighting, and the prospectus states that Ripple "has indicated" it may release additional XRP from escrow to support on-ledger liquidity if regulatory clarity arrives via CLARITY. That sentence carries no citation, no date and no source. Ripple has made no equivalent public statement and it does not appear in the company's quarterly XRP markets reports. Mechanically, XRP Ledger escrows release on immutable time conditions; an EscrowFinish transaction fails if the programmed time has not passed. Ripple can choose to distribute scheduled monthly releases instead of re-escrowing them, which is a real lever, but no legislation can accelerate the ledger. Treat this as a sponsor disclosure, not an SEC finding and not a Ripple announcement.

Elsewhere: BitMart halted all spot, futures and copy trading at 01:00 UTC on August 26 with a 05:00 UTC withdrawal request cutoff, users reported small withdrawals shrinking hours beforehand, and the restructuring engagement announced August 21 did not move the wind-down calendar. Full platform termination remains January 31, 2027, with a roadmap promised by September 9. BNB Chain's Pasteur hard fork activated on schedule. Pakistan's licensing registration deadline is September 5. Australia tightened know-your-customer rules and Thailand opened an ETF consultation. Solana's decentralized exchange spot volume has beaten several major centralized venues for nine consecutive weeks and the network touched about 5 million daily active addresses. The SEC's Regulation Crypto Assets comment window and Treasury's stablecoin implementation proposal both remain open.

Security and custody: CoinGecko published its 2026 security report today

CoinGecko Research released its State of Crypto Security Report on August 27, which makes it the only crypto-native document of the day worth reading next to the price tape. One disclosure first: Toobit is the report's featured exchange and the Toobit Shield Fund appears in its protection fund panel. Read the numbers on their own terms.

The headline is 3.63 billion dollars lost across 245 documented incidents between January 2025 and July 2026, with the ten largest attacks accounting for more than 72.5% of all value stolen. That concentration is the finding, not the total. Losses are not spread evenly across a long tail of small failures; they cluster in a handful of catastrophic ones.

By vector:

• Infrastructure and supply chain: more than 1.8 billion dollars, the single most damaging category for both centralized and decentralized venues. Bybit and KelpDAO are the named cases

• Application layer: 546 million dollars drained from decentralized applications through smart contract exploits

• Centralized exchanges: private key compromise remains the most prevalent point of failure

• Oracle and market manipulation, plus internal mechanism errors, hit both venue types. Bitget, Binance and Hyperliquid are named

The audit finding is the uncomfortable one. 147 of the 245 incidents involved protocols that had been audited before they were compromised, and those audited entities accounted for 88.44% of total capital drained. Only about 11.0% of incidents traced back to in-scope smart contract flaws, though those still cost 396.0 million. An audit is a snapshot of code. It says nothing about the humans holding the keys, and the report's threat-actor section is explicit that the adversary has changed: away from opportunistic individuals and toward organized cartels and state-linked groups including North Korean operators, laundering through mixers, bridges and staggered withdrawals.

On-chain insurance is moving the wrong way against that. Active coverage across the major insurance protocols fell 20.2%, from 163.2 million to 130.2 million, while cumulative payouts stayed roughly flat near 33.0 million. As of August 2026, five of the nine on-chain insurance protocols have gone inactive or pivoted. The decentralized answer to custody risk is shrinking, not scaling.

That leaves exchange-funded reserves as the working coverage layer. The report's protection fund panel, priced as of August 24, 2026:

Two mechanical points are worth more than any of the marketing around these funds.

First, every reserve denominated in bitcoin is marked to market, which means the coverage figure moves with the asset it is meant to insure, and it moves down hardest exactly when a panic would make you want to check it. Toobit's pool carried a 50 million valuation at its October 2025 launch and prices at 40.2 million on the same basket on August 24. Bitget's peaked at 789 million in October 2025 and prices at 423.6 million now. Neither shrank because anything was spent. Read these as coin quantities with published addresses, not as dollar promises, and the comparison becomes useful rather than promotional.

Second, and this is the boundary that matters this week, none of these funds cover the failure mode that took BitMart's users offline on August 26. They are built for platform-side breaches and technical faults. They do not cover your own compromised credentials, and no exchange-administered reserve can cover the case where the exchange itself is the problem, because the administrator of the fund cannot indemnify losses the administrator created. The report's line on attestation is appropriately dry: compliance frameworks and proof of reserves bolster user confidence but offer little defense against social engineering or catastrophic failures in private key security. Published wallet addresses you can check yourself are worth more than a headline number you cannot.

Levels

The band is narrow and the confirmation threshold is above it.

• Immediate resistance: R1 at 80,805, then the 50-week moving average around 81,033 to 81,085, then the 81,235 to 81,265 high from Tuesday, then 82,814, the May high before June's break under 60,000. The daily upper Bollinger sits at 83,446

• The number that matters: 83,000, the 365-day moving average. Multiple desks converge on a weekly close above 82,000 to 83,000 as the line that converts this from relief rally to trend change. Below it, the October 2025 series of lower highs is technically intact

• Immediate support: S1 at 78,831, then 78,000, then the 77,251 fifty-week exponential average that Rekt Capital flagged as the monthly-close line, last held in October 2025

• Below that: 75,500 to 75,968, last weekend's base and the true market mean. Losing it opens 74,000 and validates the long-liquidation warnings

• Structural: the 200-day exponential near 72,019 to 72,114, then 71,800 to 70,000. The 50-day exponential at 68,242 is where the trend argument would actually break

Daily average true range is about 2,636 dollars, so a single session can cover the whole 78,800 to 80,800 range. Ether needs 2,400 to 2,431 to hold for its January reclaim to mean anything; its on-chain cost basis near 2,300 is the deeper floor. Solana cleared 100 to about 102.59 and traded near 97.50 to 105 with a daily relative strength near 79 to 84, and the question is whether 100 to 103 flips to support. XRP near 1.37 to 1.40 is the weakest large cap and has given back most of a 70% four-day move.

Three scenarios into the weekend:

• Warsh treats long-end stress as a Fed problem: 80,805 and 81,085 clear, 82,814 and then 83,446 come into play, and the monthly close makes the trend-change argument

• Warsh endorses higher term premium as useful tightening: 78,831 goes, 77,251 becomes the fight, and August closes as another macro lower high

• Warsh says nothing decisive: bitcoin ranges 78,000 to 81,000 into the monthly close and the August employment report inherits the repricing

Alpha watch

The Nvidia non-reaction is the highest-signal event of the day

A 108 billion dollar guide, a 70% forward-year growth comment against a 44% consensus, and a 7.66% move in the largest company in the AI complex produced no bitcoin follow-through. That kills the lazy version of the correlation argument. It also means the September setup for crypto runs through rates and the dollar, not through tech earnings. Anyone still sizing bitcoin as high-beta Nasdaq got a free lesson today.

Eight straight inflow days is the only genuinely new information

The squeeze is finished and cannot repeat. Flows have now printed positive for eight consecutive sessions and 3.03 billion in August, cutting year-to-date outflows by more than half. But the daily number is decaying: 606 million, then 307, 338, 314, 232. Turnover fell from 5.36 billion to 2.26 billion. If Thursday and Friday come in under 200 million while price holds, that is absorption. If they come in under 200 million while price slips, the baton was never passed.

83,000 is the whole argument, and nothing below it settles anything

Every constructive framework this week routes through the 365-day moving average near 83,000. CryptoQuant's Bull Score at 80, the first simultaneous spot and futures demand increase since October 2025, and the record short liquidation all describe a possible bottom. None of them override a monthly close under a descending resistance line. August closes Monday. That candle is worth more than any intraday print this week.

Concentration risk is now a market structure issue, not a footnote

One asset manager supplied roughly nine of every ten bitcoin ETF dollars on August 25 and about seven of eight on August 26, plus 78% of ether flows on one session. Grayscale's GBTC was still redeeming 50.4 million on Wednesday. The institutional bid is real and it is narrow. Watch whether Fidelity, Bitwise and ARK re-engage; breadth in the flow table is what would turn a streak into a regime.

The venue tape deserves attention while sentiment sits at 71

BitMart went dark on schedule with unresolved withdrawal complaints and no published creditor data, in the same week that bitcoin traded 25% above its August low and greed readings sat in the seventies. Counterparty risk does not correlate with your macro thesis. CoinGecko's report landing the same week is a useful accident of timing: 3.63 billion lost over 19 months, private key compromise still the top centralized failure mode, on-chain insurance coverage down 20.2%, and five of nine insurance protocols gone. This particular wind-down is not systemic, but exchange concentration and verifiable reserve addresses matter more in greed than in fear, and greed is where we are.

Bottom line

The most useful thing that happened on August 27 is what did not happen. Nvidia handed the market a blowout print and an acceleration guide, semiconductors ripped, and bitcoin sat at 79,960 with a relative strength reading of 81.71 and a pivot it could not clear. Crypto is not trading the AI cycle right now. It is trading whether the United States can fund 40 trillion dollars of debt without letting the long end run, and Wednesday's 3.7% headline PCE made that question harder while core behaved.

What is genuinely improved: eight consecutive days of ETF inflows, 3.03 billion in August, year-to-date outflows more than halved, ether and Solana funds on their own streaks, record cumulative Solana inflows, CME basis at a 19-month high, spot and futures demand rising together for the first time since October 2025, and a record short liquidation that historically clusters near bottoms. What is not: a daily relative strength above 81 with longs crowded again, flows decaying session over session, one issuer supplying the overwhelming majority of the bid, elevated unrealized profits with rising exchange inflows, and a 365-day moving average near 83,000 that has not been reclaimed.

Warsh speaks Friday morning and August closes Monday. Those two events decide whether this was a bull market reset or the best relief rally of a bear market. Reclaim 80,805 and then close the month above 82,000 to 83,000 and the first reading wins. Lose 78,831 and then 77,251 and the second one does. Either way, the variable is the same one that started this move four sessions ago: the price of long-dated US government debt.

 

Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.