Crypto News and Tradfi News - COINOTAG

• OpenAI told investors annualized revenue neared $50 billion by the end of September
• The Nasdaq 100 dropped more than 300 points inside roughly half an hour Thursday afternoon
• Reports on September 29 had put OpenAI's annualized run-rate near $70 billion
Nasdaq 100 Loses 300 Points After the Note
OpenAI has told investors that annualized revenue neared $50 billion by the end of September, roughly $20 billion below the figure widely reported a month earlier, and the Nasdaq 100 dropped more than 300 points inside roughly half an hour once the news crossed trading screens on Thursday afternoon. Reports on September 29 had put the Sam Altman-led firm's annualized run-rate near $70 billion. The lower number comes from a note the company circulated to investors, summarized as it spread across trading desks. The move ranked among the session's sharpest, and it began minutes after the figure started circulating.
Annualized revenue takes a company's current sales pace and stretches it across a full year, so the figure projects momentum rather than money already booked. Neither disclosure amounted to an official earnings release; the revised figure reached the market through the investor note itself. The gap still landed hard in a market already positioned nervously. Nasdaq 100 futures had been down 0.5% in early Thursday trading, the 30-year US Treasury yield traded at levels last seen more than two decades ago, and Brent crude pushed past $100 per barrel amid Middle East supply fears. AI borrowing formed a third weight on sentiment. Broadcom is lining up more than $50 billion of financing to fund a custom chip it is developing alongside OpenAI. Michael Burry has argued publicly that markets must fall far enough to shut out the coming wave of artificial intelligence listings. Skeptics of the trade have spent recent weeks arguing that AI valuations rest on revenue lines no auditor treats consistently. Revenue at OpenAI and Anthropic is read as the clearest gauge of demand for AI, because it drives how much computing infrastructure gets built. On Thursday, two ways of counting that same sales pace sat $20 billion apart.
The pre-IPO OPENAI instrument changes hands at $1,642.25, down 3.69% over the past 24 hours after a session range of $1,577.00 to $1,722.40, and COINOTAG's composite scoring frames the map around it. The nearest resistance sits at $1,732.39 and scores 86 out of 100. The nearest support, at $1,548.84, scores 66 out of 100 and is marked in part by a Fibonacci retracement level. Positioning stays light: perpetual open interest holds near $25.2 million, with funding at 0.0050% per interval. The daily trend reads sideways and the MACD signal is bearish, leaving the instrument capped below the $1,732.39 ceiling.
Two Accounting Methods, $20 Billion Apart
The discrepancy traces to how each lab books sales, not to any sudden shift in demand. Anthropic, which is preparing its own Nasdaq listing, books revenue earned through cloud partners including Amazon Web Services and Google Cloud inside its own line. OpenAI does not recognize those partner-channel sales and bills only customers it serves directly. Cloud partners such as Amazon and Google resell model access under their own brands, which is why the treatment of those sales varies between labs. Most large enterprise buyers procure model access through existing cloud agreements, so the partner channel carries substantial weight. Investors who restated OpenAI's figures on Anthropic's basis produced the higher $70 billion estimate that circulated last month.
Jim Cramer pressed the same point as the note spread, noting that the two companies report revenue differently and that the comparison is not apples to apples. The company itself told investors revenue expanded by more than 70% during the period covered by the note, a pace few businesses of any size match. Audited documents put OpenAI's revenue for all of 2025 at $13.07 billion, a base the current run-rate implies has multiplied several times over. The accounting choice matters beyond one afternoon of selling, because whether partner resale dollars appear in reported revenue changes how investors value forward demand for AI capacity. It also shapes how each lab's progress is judged between funding rounds, when annualized figures do much of the talking. The debate lands as Anthropic readies its listing, since its partner-inclusive figure will sit directly beside OpenAI's in investor models. No reconciliation of the two methods has been published; the $13.07 billion 2025 base and the 70%-plus growth rate are the hard numbers both camps share.

• Polkadot a lancé le stablecoin dotUSD le 8 octobre via le référendum de gouvernance 1944.
• La première phase émet dotUSD un pour un contre les réserves en USDT de Tether.
• La liquidité du trésor pour le pool DOT et dotUSD a été réduite de 5 millions de dollars à 3 millions de dollars.
dotUSD est lancé via un référendum 1944
Polkadot (DOT) a lancé dotUSD, un stablecoin adossé au dollar que le protocole décrit comme n’ayant aucune société émettrice, jeudi 8 octobre. Ses débuts arrivent enveloppés dans une contradiction que les propres supports du protocole rendent difficile à manquer : la version passée en ligne a été frappée un pour un contre l’USDT de Tether, le même type de jeton dollar contrôlé de façon centralisée que le discours de lancement met en garde. Dans le message d’annonce sur X, Polkadot a fait valoir qu’une poignée d’entreprises émet et contrôle la majeure partie des stablecoins du monde, que ce marché représente plus de 250 milliards de dollars, et que ces entreprises décident qui peut détenir leurs jetons, « à la manière des banques ». dotUSD, affirme le post, repose sur un autre principe. Pour l’instant, le produit en ligne ne s’y conforme pas. Le déploiement passe par le référendum de gouvernance 1944, la proposition on-chain à l’origine du lancement, dont la première phase est déjà construite et enregistrée sur la blockchain. Elle permet aux utilisateurs d’émettre dotUSD un pour un avec l’USDT, dans une limite d’offre, et le jeton de Tether se trouve derrière la réserve à ce stade. Pour lancer la négociation, la proposition s’appuie sur le trésor de Polkadot afin de mettre en place un pool de liquidité DOT et dotUSD sur l’Asset Hub de Polkadot. Le projet initial prévoyait 2,5 millions de dollars en USDT pour frapper dotUSD et 2,5 millions de dollars en DOT pour le pool, soit un démarrage de 5 millions. Une révision ultérieure a réduit chaque volet à 1,5 million de dollars, pour atteindre 3 millions. La Polkadot Community Foundation, qui a présenté la proposition, indique qu’elle joue un rôle administratif uniquement et qu’elle n’émet pas, n’exploite pas et ne détient pas dotUSD, DOT ou USDT. Pour les traders qui suivent le prix de Polkadot, l’effet du marché pour l’instant passe par les dépenses du trésor en liquidité plutôt que par un rachat direct quelconque ; ainsi, le lancement touche d’abord les rails du réseau avant de toucher la demande.

• Oracle (ORCL) stock fell over 5% intraday on October 8 and last traded down 4.51% at $136.72.
• A report placed OpenAI annualized revenue near $50 billion, $20 billion below the $70 billion earlier figure.
• OpenAI's revenue reportedly grew more than 70% between July and late September.
Update 21:12 UTC: The selloff landed alongside fresh macro pressure in the same session.
Oracle Leads a Broad AI Stock Selloff
Oracle's stock led a broad selloff across AI-linked equities on Thursday, October 8, dropping more than 5% during the session after a Financial Times report cast fresh doubt on how much money OpenAI is actually earning. The report put OpenAI's annualized revenue, a projection that stretches one month's sales across a full year, near $50 billion at the end of September. That sits well below the $70 billion figure reported earlier, leaving a $20 billion gap that traders treated as a warning about the economics behind the AI buildout. The shortfall does not mean OpenAI lost $20 billion in actual sales. It largely reflects differences in how revenue is calculated between the two readings, and OpenAI's revenue has reportedly grown more than 70% since July. Markets sold first and asked questions later. Broadcom (AVGO) fell nearly 4%, Nvidia (NVDA) dropped more than 2% and Microsoft (MSFT) lost over 1% in Thursday trading. Oracle has more reason than most to react. The company supplies a large share of the computing power behind ChatGPT, and its stock has re-rated on the promise of AI cloud demand, so any question about OpenAI's finances lands directly on Oracle's valuation. COINOTAG data shows Oracle (ORCL) last trading at $136.72, down 4.51% over 24 hours, after touching $134.90 at the low and $144.26 at the high. For a company valued on future cloud contracts rather than current cash flow, a one-third markdown in its largest customer's revenue run rate was enough to force a repricing across the AI supply chain.
A $300 Billion Contract Ties Oracle to OpenAI
A single agreement explains the sensitivity. In September 2025, Oracle and OpenAI signed a $300 billion cloud computing deal that runs five years, a contract equal to roughly 45% of Oracle's $664 billion backlog of contracted future revenue. The backlog is the total value of contracts already signed but not yet billed, and Oracle has to build the data centers before it collects most of that money. Its latest quarterly results showed $28.5 billion in capital expenditure and $5.4 billion in negative free cash flow, a profile that leaves the company spending heavily now against income it will book over years. Timing is the core risk. The Wall Street Journal reported Thursday that Oracle, Broadcom and SpaceX are seeking large debt deals to pay for AI chips. Oracle's own position is that AI cloud demand “continues to grow faster than supply,” and the revenue report changes nothing about the volume of compute OpenAI has committed to buy. That concentration is what separates Oracle from its peers: nearly half of its contracted future revenue rests on one customer's ability to pay. The exposure, however, extends well beyond one stock. CoreWeave (CRWV) holds announced cloud contracts with OpenAI worth as much as $22.4 billion. Broadcom is developing custom AI chips with the company and is reportedly arranging more than $50 billion in financing. AMD carries a six-gigawatt chip agreement, while Nvidia holds a 10-gigawatt infrastructure partnership. SoftBank (9984.T) faces exposure through its multibillion-dollar investment in the ChatGPT developer. Each of these names would come under fresh pressure if OpenAI's growth disappoints, and the damage in each case depends on how much future business investors have already priced in.
The selloff landed alongside fresh macro pressure in the same session. The S&P 500 closed down 0.47% at 7,765.36 and the Nasdaq lost 1.25% to 27,193.30, while the Dow held a 0.10% gain at 51,231.60, per CNBC. AMD fell about 3%, but Palantir climbed roughly 2% after Goldman Sachs raised its rating, citing sovereign AI demand. Oil added another headwind: Brent jumped about 4% to near $104 a barrel and WTI rose 4% to around $91 after President Trump said he does not want a deal to end the war with Iran, though he ruled out an attack before next month's midterm elections. The 10-year Treasury yield eased more than 3bp to 5.244% on strong auction demand, and strategist Ross Mayfield of Baird warned that sustained high energy prices could push the Fed toward further rate hikes.
The technical picture shows how thin the cushion now is. COINOTAG's support and resistance scoring places the strongest floor for Oracle at $136.05, rated 71 out of 100 and reinforced by a Fibonacci retracement level, while the heaviest resistance sits at $148.56 with a 100 out of 100 score. The daily RSI reads 40.7 and the daily trend stays sideways. In the stock perpetual market, open interest stands near $30.6 million and the funding rate at 0.0061% per interval, so the drop came without crowded leverage behind it. On the pre-IPO OpenAI instrument, the last price read $1,644.28, down 3.77% over 24 hours. A daily close below $136.05 would deepen Oracle's slide; a recovery through $148.56 would put the pre-report range back in play (as of 21:12 UTC).

• Un groupe de liquidations s’étend de 81 700 à 83 300 $, la cotation du Bitcoin se trouvant à l’intérieur.
• Les séances de trading américaines sont passées en territoire net vendeur, les gains nets venant désormais en dehors des heures US.
• La moyenne mobile sur sept jours du volume au comptant et des ETF s’élève à 6,8 milliards de dollars, en dessous de la plupart des jours depuis janvier 2024.
$81,798 Entre deux barrières d’ordres
Le prix du Bitcoin (BTC) s’établit à 81 798 $, encadré entre un groupe d’ordres acheteurs à 81 000–81 250 $ en dessous et un mur d’offre à 86 500–86 750 $ au-dessus. D’après les données en chaîne de Glassnode, une société d’analytique dont les jeux de données sont compilés à partir des relevés de nœuds de blockchain, les deux bords ont été cartographiés après que la devise a franchi les 85 000 $. Elle a atteint le mur puis s’est repliée. Le recul s’est arrêté juste avant la bande inférieure. À partir de la lecture du jeudi à 18 h 40 UTC, la cote a légèrement progressé de 0,3 %, la laissant toutefois en baisse de 1,9 % sur les 24 dernières heures. La valeur de marché se situe près de 1,64 billion de dollars. Les analystes interprètent la structure comme un prix comprimé entre une zone d’offre supérieure et une zone de demande inférieure, sans qu’aucun des deux bords n’ait encore été franchi.

• Type 2 keeps book-entry settlement and could extend to the Bank of Japan's own ledger.
• The 10-year JGB yield hit 2.95% in August, the highest since September 1996.
• Fiscal 2027 budget requests hit a record ¥143 trillion, with debt servicing at ¥36.64 trillion.
Three types of on-chain handling
Japan's Ministry of Finance convened the first meeting of its study group on the on-chain handling of Japanese Government Bonds (JGBs) on Thursday, October 8, and published the explanatory materials the session worked from. The panel brings together academics from the University of Tokyo and Waseda University, private-sector strategist Chotaro Morita, and officials from the Bank of Japan and the Financial Services Agency. Its brief is to test how blockchain technology could carry a bond market that has so far settled entirely on legacy rails. The ministry's explanatory paper sorts the options into three types, framed as a provisional map for discussion rather than a settled decision. Type 1 does not touch the bond itself: the beneficiary rights of an investment trust, in practice a money market fund (MMF) holding JGBs, would be transferred on a blockchain. Type 2 keeps the current book-entry settlement system and moves the transfer account ledgers on-chain, in three variants: one account-management institution, several institutions acting in concert, or an extension as far as the Bank of Japan's own ledger, where the central bank acts as transfer agent. Type 3 goes furthest: a new form of government bond issued directly on a layer 1 blockchain protocol outside the existing settlement framework, with the scope of change depending on where the technology is inserted. On the benefit side, the paper lists more efficient collateral and liquidity management, a broader investor base with overseas holders as the main gainers, and a stable place to park idle funds until they are spent. Policy gains include faster response to market disruption and more stable absorption of JGBs. The costs it names are market fragmentation across venues, a larger pool of funds needed to support trading, harder containment of sharp price moves under round-the-clock trading, and the expense of adapting systems, rules and operations. Everything, including the three types, is provisional; the ministry plans hearings with firms before compiling a report around January 2027.
Weak auctions and a record budget
The urgency behind the study group is fiscal. Demand for JGBs has plateaued, and the two most recent 10-year auctions drew weak bids. In August the 10-year yield climbed to 2.95%, a mark unseen since September 1996, while the 2-year hit 1.75%, its highest in 31 years, and the 5-year set a record 2.21%. Budget requests for fiscal 2027 came in at a record ¥143 trillion, about $918 billion, a fourth consecutive annual record, and debt-servicing costs are heading toward ¥36.64 trillion, roughly $234 billion, with the assumed interest rate lifted from 3% to 3.8%. Washington supplies the reference case. Under the GENIUS Act, stablecoin issuers must hold safe and liquid reserves, Treasury bills chief among them, which has turned a payments product into a steady buyer of government debt. A San Francisco Fed study found the United States has added short-term Treasuries faster than Japan since 2023, even though Japan remains the largest foreign holder of US government debt. That is demand created inside traditional finance, with no crypto asset required, and Tokyo's officials have watched the shift closely as they weigh the same mechanism for the yen. Japan also has a homegrown version. JPYC, the first yen-pegged stablecoin, is backed by domestic deposits and JGBs, and its issuer earns from the interest on those holdings rather than from transaction fees. As it issues more tokens, it buys more JGBs. The ministry's materials note that an investor already holding assets on-chain gains a stable option for parking idle cash until it is used for payment or investment, exactly the behavior that channels stablecoin float into bond demand.
Fragmentation is the open question
The three types are not equal bets, and the difference lies in where the market could split. Type 1 is the narrowest and quickest to trial, since it moves fund shares rather than the bond itself. Type 2 stays inside plumbing the Bank of Japan already supervises, so the ledger remains accountable to existing institutions even as its form changes. Type 3 invites fragmentation most directly, because it creates a parallel venue outside the current system, and the ministry itself lists fragmentation and 24/7 price-swing risk among its open costs. Connecting an off-chain ledger with an on-chain one also requires plumbing comparable to cross-chain bridges, which have been recurring exploit targets in crypto, so security design will weigh on which type advances. With hearings ahead and a report due around January 2027, this meeting fixed the agenda rather than the outcome.

• Bitcoin tested the $83,500 support early on Wednesday, 8 October, before slipping below it.
• Open interest climbed while Bitcoin's price fell, showing new leveraged positions entering derivatives markets.
• COINOTAG scoring puts the strongest support at $80,880, rated 92 out of 100.
Thursday Morning Test of $83,500
Early trading on Thursday, 8 October, pushed the Bitcoin (BTC) price down toward the $83,500 mark, a level Bitcoin market commentary had flagged as a key liquidity zone heading into the session. Through the morning hours the market defended the line, and sellers could not force a decisive close beneath it, leaving the pullback modest by the standards of recent weeks. The picture recorded at 05:12 UTC still showed the zone holding, with buyers preventing any extension toward lower levels for the moment. What made the session stand out was the behavior of the derivatives market at the same time. Open interest, the total value of positions traders keep open in futures and perpetual markets, rose while the price drifted lower, a combination that points to new money and higher leverage entering rather than capital leaving. None of that flow shows panic selling yet; it shows positioning. Traders read that mix as a volatility setup: with both sides committing fresh capital, a forced unwinding of crowded positions can move the price sharply in either direction. It also framed the question that dominated discussion through the session, whether this dip is an entry point worth repositioning on or the first leg of a deeper correction. Since that morning test, the market has leaned toward the bearish answer. The live reading puts Bitcoin near $81,830, roughly 2% lower across the past 24 hours, which leaves the price below the $83,500 zone that held early in the day. The level has flipped from defended support into a ceiling sellers now control, and the next meaningful test sits lower on the ladder rather than above it. For spot buyers that shift is a modest discount, while for anyone carrying leveraged long exposure built near the recent highs it is exactly the scenario the climbing open interest warned about. Our composite support and resistance levels frame where the next tests sit. The strongest support under the market is $80,880, scored 92 out of 100 where several indicators converge, and the nearest overhead resistance at $84,088 carries an 80/100 score. A daily close back above $84,088 would neutralize the intraday breakdown, while losing $80,880 opens a deeper shelf further down. Positioning still leans long: the long/short account ratio reads 2.05, with 67.2% of accounts on the long side, and aggregate open interest stands near $15.65 billion. That long tilt is the fuel a continued slide would burn first, and it is why the leverage build-up deserves more attention than the few percent of price drift itself.
Two Scenarios, One Leverage Question
Analysts tracking the session laid out two scenarios that decide Bitcoin's direction over the coming days. The constructive path requires the price to close and consolidate above the $83,500 zone alongside a pickup in spot volume, which would invalidate the downside case and reopen the ladder of overhead resistance; if the reclaim arrives through a short squeeze, a rapid forced buyback by traders who sold the breakdown, the move toward fresh highs can extend quickly. The cautious path treats the break as confirmation. A failure to hold $83,500 on a closing basis extends the correction toward the next support zones, which offer patient buyers a wider margin of safety, and elevated open interest under sustained selling pressure raises the risk of a fast liquidation wave among leveraged longs that deepens the dip before it stabilizes. The intraday action has already traveled part of that second path. Larger holders matter here as well, because a crypto whale absorbing supply would show up first in spot volumes rather than in derivatives positioning, and no such absorption was visible in the data behind the morning test.
For long-term investors the framing differs from the one facing leveraged traders. Participants who prefer to HODL through pullbacks, and who judge the market by cycle gauges such as the Bitcoin Rainbow Chart or structural drivers such as the Bitcoin halving, can treat a few percent of drift as an opportunity to build positions gradually rather than as a reason to exit. The retreat toward $83,500 is therefore a decision point in the price path: attractive for gradual accumulation, demanding caution from anyone running futures exposure while leverage and open interest remain elevated. The next checkpoint is the daily close relative to $83,500 and the spot volumes behind it.

• Jacob Creech a déclaré le 8 octobre 2026 que les utilisateurs de Solana n’ont pas besoin de mouvements urgents de portefeuilles liés au quantique.
• Solana signe les transactions avec l’algorithme Ed25519, dont les clés de signature dérivent d’une graine secrète.
• Creech a déclaré qu’une preuve de concept existe, mais que rien n’est en production sur le réseau principal (mainnet) de Solana.
Une assurance rassurante d’un développeur face aux menaces quantiques
Le développeur de Solana (SOL) Jacob Creech a publié jeudi 8 octobre 2026 une réponse publique rassurante, indiquant aux utilisateurs qu’ils n’ont pas besoin de prendre des mesures d’urgence pour protéger leurs avoirs contre les menaces liées à l’informatique quantique. Dans un post sur X, Creech a soutenu que ni l’accélération de la technologie quantique ni les attaques classiques assistées par l’intelligence artificielle ne nécessitent, à ce stade, que les utilisateurs de Solana prennent des précautions extraordinaires. Son propos est arrivé en réponse directe à Justin Drake, chercheur chez Ethereum, qui a exhorté le secteur crypto à se préparer aux menaces quantiques de manière calme mais planifiée, et recommande personnellement de déplacer des actifs de façon contrôlée vers de nouvelles adresses dont les clés publiques ne sont pas encore visibles sur la blockchain. La logique de Drake est préventive : si une clé publique n’apparaît jamais on-chain, une attaque future qui déduit des clés privées à partir des clés publiques n’a rien sur quoi s’appuyer. Creech parle en tant que développeur au sein de l’écosystème Solana, avec une connaissance directe de la cryptographie du protocole, et, en tant que défenseur, un intérêt à maintenir sa base d’utilisateurs sereine. Le cœur technique de son argument est Ed25519, l’algorithme de signature numérique que Solana utilise pour la signature cryptographique des transactions. Un ordinateur quantique suffisamment puissant, ou une attaque classique efficace reposant sur les mêmes bases mathématiques, pourrait en théorie déduire la clé de signature privée d’un utilisateur à partir de la clé publique exposée par chaque transaction. Rien dans le post de Creech n’évoquait une faille ou une exploitation sur le réseau ; l’échange concerne la préparation à une capacité future, pas un incident en cours. Le désaccord porte sur le calendrier et la méthode, pas sur le caractère réel de la menace. Le cours de Solana a reculé de 5,8 % au cours des dernières 24 heures : un mouvement que notre suivi en direct relie à des ventes massives sur l’ensemble du marché plutôt qu’au débat sur la sécurité.

• JPMorgan strategists led by Eduardo Lecubarri warned October 8 that rising yields threaten small-cap stocks.
• Only 9% of US small and mid-cap stocks yield above the 30-year Treasury, down from 19%.
• The 30-year Treasury yield touched 5.70% on Wednesday, a 24-year high.
A 24-Year Low in Dividend Cover
JPMorgan's small- and mid-cap strategists, led by Eduardo Lecubarri, have told clients that government borrowing is turning long-dated US bonds into direct competition for equity income. The bank put a number on the squeeze: only 9% of US small and mid-sized company stocks now pay dividends yielding more than the 30-year Treasury, down from 19% two years ago and the lowest share in 24 years. A dividend is the cash a company hands to its shareholders, and a 30-year Treasury is a loan to the US government that pays a fixed return for three decades. When the safe bond pays more, the riskier equity looks less attractive to anyone buying an income stream. The strategists traced the pressure to sovereign debt: roughly 60% of global GDP now sits in countries owing more than a full year of output while still running budget deficits, a combination they described as a first in history. Two years ago the same screen captured nearly twice as many payers, a shift the team attributes to deteriorating government finances rather than weak corporate payouts. Investors "seem to be ignoring" the risk, the note warned, flagging "nasty surprises" ahead. The document moved across trading desks after the summary posted October 8 reported that surging yields now threaten small- and mid-cap returns. The market gave the argument a hearing the same day. The 30-year yield touched 5.70% on Wednesday, its highest in 24 years, and stood just shy of that peak at the time of writing. Small caps took the harder hit: the Russell 2000, an index of about 2,000 smaller US firms, fell 1.31%, while the S&P 500 slipped 0.22% at the close, closing data shows.
Dashenlin and Befesa Make the Cut
The same note that carried the warning also carried its exceptions. JPMorgan screened smaller companies worldwide for dividends that beat their home country's 30-year bond and added two names to its model portfolio, the bank's sample list of favored holdings. One is Dashenlin Pharmaceutical, a $3 billion Chinese pharmacy chain. The other is Befesa, a €1.4 billion German-listed firm that recycles waste from the steel industry. Both carry an Overweight rating, the bank's label for holdings it expects to beat their peer group. The screen explains the choice: each company pays out enough to clear the long bond in its home market, the exact test the 9% figure shows most US small and mid-caps now fail. The strategists conceded that today's dividend share resembles the 1990s, when long yields were falling and smaller companies prospered. Their rebuttal to that comparison was blunt: "These are not the 1990s." Not every market voice shares the alarm. Jim Bianco, a longtime bond bear, has turned bullish on bonds and calls 5% yields fair value, a reversal that puts him on the opposite side of the JPMorgan team over whether current levels are a threat or a floor. Policy adds another layer: Federal Reserve minutes released Wednesday showed most officials backing another rate hike in 2026, with the next decision scheduled for October 27-28.
The Duration Call Behind the Screen
Read as a whole, the note is a duration call dressed as a stock screen. When only one in ten smaller US companies out-yields the long bond, income-seeking capital has little reason to take single-stock risk, and the gap widens if the Fed delivers the hike most officials backed in Wednesday's minutes. The two model-portfolio additions show where the bank still finds value: in firms whose payouts clear local long yields, not in the index at large. Markets now look to the October 27-28 meeting to see whether 5.70% marks a peak for the 30-year yield or the start of a higher range.

• XRP whale-retail spread fell from 64% to 46.7% between September 30 and October 8, a 27% drop.
• Binance whale-retail spread declined from 68% to 54.9%, a 13.1 percentage-point drop over eight days.
• XRP revisited its lowest price level in over a month amid market-wide volatility.
Whale-Retail Spread Loses 17.3 Points
The gap between whale and retail participation in XRP has narrowed fast, with the whale-retail spread across all crypto exchanges dropping from 64% on September 30 to 46.7% on October 8. That is a decline of 17.3 percentage points, roughly 27%, in eight days, according to the latest CryptoQuant reading. The metric measures the relative difference between whale and retail activity in XRP exchange outflows, so a falling reading means large holders account for a shrinking share of the coins leaving crypto exchanges compared with smaller traders. In practice, a wide spread means whale transactions dominate the measured outflow, while a compressed one means retail tickets fill the gap. The shift comes at a delicate moment: market-wide volatility has pushed the XRP price back to its lowest level in more than a month, and our live monitoring shows the token down 6.2% over the past 24 hours. Whale activity, in other words, has gone quiet while smaller traders keep the exchange outflow pipe moving. Across centralized venues as a whole, whale dominance in XRP flows has weakened notably over the last day alone, making the October 8 print the softest of the stretch. Readings above 50% still indicate whales dominate the flow, and the current 46.7% figure sits just under that line, which is why the speed of the change matters as much as its level. Because the metric is an aggregate of outflow behavior rather than a holdings snapshot, it can move quickly when a handful of large wallets simply sit on their hands.
Binance Still Carries the Higher Reading
Venue-level numbers show the retreat is real but uneven. Over the same eight days, the whale-retail spread on Binance fell from 68% to 54.9%, a 13.1 percentage-point decline, or 19.3%, on the world's largest cryptocurrency exchange. That slide is proportionally smaller than the market-wide 27% drop, which suggests part of the overall compression came from smaller venues. Even so, Binance's whale reading sits well above the 46.7% all-exchange average, so the platform still hosts a heavier concentration of large-holder flow than the market overall, and its depth of XRP order books is why the venue is watched as a proxy for global whale behavior. Attribution matters here: dormancy is not distribution. The spread records participation, not direction, and the dataset says large holders are barely trading at the moment, not that they are unloading XRP into the drawdown. Outflows themselves can reflect several behaviors at once, from self-custody moves to treasury rebalancing; a recent transfer of 99.86 million XRP to Binance linked to Ripple showed how a single large movement can swing the tape. For an asset of XRP's market cap, the depth normally supplied by institutional desks and market makers carries extra weight during drawdowns, because retail-heavy flow tends to be faster on the way down and thinner on the way back up. Chart work has leaned cautious as well: one widely followed analyst maps XRP below $1.50 if a 2017-style Adam and Eve base completes, while longer-horizon model-based XRP targets still stretch toward $25 and beyond by 2027. Traders who want the level-by-level picture can follow our XRP technical analysis, and newer holders reviewing entry points can start with how to buy XRP.
Dormancy, Not Distribution
Our reading is that this print is a participation story first and a price story second. Whales stepping back during a drawdown removes a layer of large bids that would otherwise absorb retail selling, which can stretch a decline even with no whale distribution behind it; the same dormancy, though, means the pressure that dragged XRP to a one-month low is not whale-driven either. The signal to watch from here is the spread itself: if whale participation re-widens while the price stabilizes, the retreat reads as patience rather than exit, and if it keeps compressing, retail stays the marginal flow in both directions. The most recent figures on record put the all-exchange spread at 46.7% and Binance's at 54.9%, both from the October 8 print.

• Dragonfly's Haseeb Qureshi proposed a cryptographic recovery mode on X on October 8.
• Justin Drake warned ECDSA could break within months and advised bunker-mode migration.
• Validators would vote to trigger an emergency switch, forcing the chain into recovery mode.
Qureshi's Answer to Bunker Mode
Dragonfly managing partner Haseeb Qureshi has put forward a plan to make blockchains survivable if artificial intelligence ever cracks the signatures that secure wallets. In a long-form post published on X on Thursday, October 8, titled “Against Cryptographic Doomerism”, he proposed what he calls a cryptographic recovery mode and asked every major chain, Ethereum (ETH) foremost among them, to adopt it before a crisis forces the issue. The proposal has landed as the most developed counterpoint in a debate that spread across the developer community within a day. The immediate trigger is a warning from Ethereum Foundation researcher Justin Drake, who argued on Wednesday, October 7 that ECDSA, the elliptic-curve signing scheme behind most crypto wallets, could be broken within months. Drake's advice was personal: move holdings to fresh addresses that have never signed a transaction, a posture he calls bunker mode.
Qureshi disagrees with that answer on its own terms. Shifting keys to an unused address protects one holder only for as long as the rest of the chain holds together, he wrote; if other accounts are compromised and their coins sold in bulk, a spared balance loses its value anyway, so the fix has to operate at the system level. He also endorsed Vitalik Buterin's position that whipping up alarm drives holders into rushed migrations, where a FOMO-driven error can cost more than a key theft that has not actually occurred. The risk he worries about is not AI speeding up quantum computers but an unexpected mathematical breakthrough that lets ordinary machines compute discrete logarithms, the problem on which modern public-key cryptography rests. Banks and web certificate authorities could re-verify identity and migrate to new standards in that scenario, he argued, while a blockchain cannot, because anyone could derive anyone else's private key and ownership itself would become unknowable.
Hash Backup Keys and a Validator-Voted Switch
The blueprint layers several mechanisms. A future protocol upgrade would carry a minimal hash-based signature scheme, which Qureshi freely describes as slow, expensive and awkward for everyday use; it exists only as a fallback. Every address would be bound in advance to one of these backup keys, voluntary at first and mandatory a few months later, after which any transaction from an address without a registered backup key would simply fail. Alongside the signatures, an emergency switch would be pre-deployed. If a working ECDSA break ever appears, validators vote to pull the switch, and every address flips into recovery mode with no separate upgrade required. Qureshi concedes the chain would become close to unusable in that state, but balances would be intact and developers would have room to design a proper conversion plan.
Addresses that never registered a key get their own path out. Holders who still control a seed phrase can prove ownership through a zero-knowledge proof and secure a new address. Everyone else must submit an ECDSA signature and solve a proof-of-work puzzle at the same time, with difficulty set by the amount of native assets the address holds and doubled once a day, so genuine owners can outrun attackers; validators can adjust both the difficulty and its growth rate as an attack unfolds. His crisis posture is explicit: in an emergency, speed outranks decentralization, decision norms can be loosened and validator authority expanded, and wallets, exchanges, RPC providers and asset issuers, including teams tokenizing real-world assets, would need to mobilize together. He reports that large investors have already begun asking which chains take the problem seriously. Qureshi also retracted a position from weeks earlier, when he argued Zcash should stop updating once its quantum-resistant migration was complete. He closes with caveats: he is not a cryptographer or a protocol designer, the details carry gaps, and the most likely outcome is that nothing happens, because AI will conclude that the cryptography is rock solid.
A Kill Switch That Concentrates Power
Our reading is that the hard question here is not cryptographic but political. Qureshi's own post grants validators, through a simple vote, the power to degrade the entire chain on judgment alone, a switch that would be irresistible in a panic and contested in every other moment. He is also right on one point the bunker-mode camp underweights: a coordinated migration, whether voluntary or validator-ordered, would revive replay attack risk, where a signature meant for one context gets reused in another, and recovery-mode design would have to solve that from day one. Ethereum price was not the catalyst; the catalyst is protocol security, and the practical test will be which chain's governance process can actually pass a plan like this.

• Micron stock rose 4.06% to close at $1,088 on Wednesday, leading the S&P 500 technology sector.
• Siebert CIO Mark Malek called Micron a buy, citing Nvidia's reliance on Micron memory.
• Micron reported $54.23 billion fiscal Q4 revenue on September 30, beating its $50 billion guidance.
Malek Backs Micron After a 4.06% Session
Mark Malek, chief investment officer at Siebert Financial, called Micron Technology (MU) a buy on Wednesday, while rejecting the idea of owning the memory maker forever. His call landed on the session the stock did the heavy lifting for its whole sector: MU gained 4.06% to close at $1,088 and led the S&P 500's electronic technology group, while Advanced Micro Devices (AMD), Nvidia, Texas Instruments and Qualcomm all finished in the red. Micron has added roughly 280% since the start of 2026.
Malek framed the position as a trade with an expiry date rather than a permanent holding, and justified the entry with Micron's grip on a key layer of the AI buildout. “This is a company that will continue to outrun the avalanche,” he said. He also pointed to Nvidia's dependence on Micron's memory: “Everyone's favorite company, NVIDIA, mine included... is beholden to Micron. It's part of their chipset. So without a boring company like Micron, they can't sell a chipset.”
The fundamentals behind the call came from the company itself. In its fiscal fourth-quarter report, published September 30, Micron disclosed revenue of $54.23 billion, ahead of its own $50 billion guidance, with gross margin reaching 87%. Management guided fiscal first-quarter 2027 revenue to $61.5 billion and said it has already contracted the vast majority of its 2027 high-bandwidth memory (HBM) supply at higher prices. HBM sits beside AI processors: Micron produces HBM4 for Nvidia's Vera Rubin platform and is co-developing a custom HBM4E version with the chipmaker. Micron has also said it has no line of sight to when DRAM supply and demand rebalance, yet the stock trades at about six times expected fiscal 2027 earnings. Malek's argument leans on that scarcity: as long as AI accelerator demand outruns memory supply, the supplier keeps its pricing power.
The rally is already being tested. The stock last changed hands at $1,038.24, down 4.52% over the past 24 hours and inside a $1,034.86 to $1,095.33 range, according to COINOTAG data. The nearest marked support sits at $1,059.86, rated 67/100 on COINOTAG's composite scoring, and the session has already traded through it; the stronger floor underneath is $1,000.72, scored 83/100 where the SMA 50, the 0.618 Fibonacci and the Ichimoku Kijun converge. Momentum has cooled with the price: the daily RSI prints 53.0 and the MACD signal has flipped bearish on the daily timeframe, while open interest in Micron's stock perpetual contracts stands at $157.2 million.
The Capex Avalanche and the $1,516 Chart Target
Malek's endorsement carried a warning: a slowdown in AI capital expenditure would hit the entire sector, and in his view memory suppliers like Micron would struggle the most. That scenario has a recent precedent. Memory names slid into a bear market in July, with MU losing about 30% from its peak, and Temasek has separately flagged an unwind of the AI trade as the top risk facing markets. Micron is not pulling back on the buildout either: the company plans roughly $25 billion of capital expenditure in the first half of fiscal 2027, outlay that only pays off if data-center demand keeps absorbing memory at current prices.
The chart behind the call has been constructive since early September. MU broke out of a symmetrical triangle on September 3 and retested the breakout on September 14 near the 0.618 Fibonacci level at $894.58, a retest that held as support. The stock has since reclaimed the 0.786 Fibonacci area near $1,053, which previously acted as resistance, and on Wednesday it dipped to $1,011 intraday before closing at $1,088 near the session high as the broader market pulled back from record levels. Volume broke above a descending trendline on October 1, and the RSI stood near 60 and rising when Malek made the call. The first target on that roadmap is the all-time high at $1,255, about 15% above Wednesday's close, while the triangle's measured move points to $1,516 near the 1.272 Fibonacci extension at $1,511.63, implying roughly 39% upside. The caveat is a daily close below $1,040, which would void the reclaim and open the path to $1,000 and then $895; Thursday's session was trading below that line at the time of writing, leaving the reclaim on trial at the close.