Around November 5, 2026, the script on the supply side of Bitcoin was completely rewritten: Keel Infrastructure sold 2,520 BTC in the third quarter, cashing out approximately $192 million, nearly emptying its inventory by the end of the quarter, and stated that all newly mined Bitcoin would be sold daily thereafter, redirecting funds to AI data centers. The shift in attitude from "holding coins for price appreciation" to "cash flow priority" was publicly documented in financial reports. Core Scientific, in the same quarter, reported only about $12.3 million in self-mining revenue, a year-on-year decline of about 79%, and recorded a gross loss, but its income from hosting and AI data center operations reached approximately $215.4 million, driving overall revenue up by about 187% year-on-year, with profits supported by computing power rentals rather than mining; meanwhile, MARA Holdings received FERC approval to acquire Long Ridge Energy & Power, securing about 505 megawatts of natural gas generation capacity and over 1,600 acres of land in one go, directly reserving energy foundation for AI and HPC expansion. When mining enterprises collectively treat watts and data centers as core assets rather than BTC itself, the price dynamics in the market are also simultaneously de-leveraging: approximately $43.12 million was liquidated across the network in the past hour, of which about $42.17 million were long positions; both BTC and ETH longs were concentrated and liquidated, while USDT.D showed strong rejection at key Fibonacci levels, and liquidity shifted tentatively between USDT and spot; concurrently, Binance voluntarily revoked all API keys associated with the third-party platform SignalPlus, tightening security measures at the exchange level around toolchain safety, leading quantitative funds to adopt a more watchful stance in the short term. Together, these three linkages mark the current phase of Bitcoin being in a state of "miner sell pressure structural reshaping, leveraged funds clearing, and risk preference reallocating between the AI narrative and coin prices."
Keel empties 2,520 BTC from hoarding to selling
In a quarter where mining yields significantly weakened post-halving, Keel Infrastructure chose to rewrite its role through a definitive transaction—selling a total of 2,520 Bitcoin, approximately 2,261 of which came from years of accumulated inventory, and about 259 were newly mined that quarter, retrieving around $192 million in cash, with almost zero holdings by the end of the quarter. Following this, the company clearly stated in its financial report communications that all new Bitcoin mined in the future would be directly sold daily, with the capital concentrated on AI data center infrastructure, while the mining business itself would retreat to being merely a cash flow tool for the data centers, no longer bearing the balance sheet function of "betting on coin prices."
This shift has significant implications on the supply side; it is rarely just about "adding a few thousand sell orders." Historically, miners were typical cyclical hoarders, able to release inventory en masse at bull market peaks, creating unpredictable sell pressure for tops; Keel emptying its inventory and committing to daily sales essentially transformed the previously opaque, timing-based inventory supply into a more linear, modelable producer cash-out flow. For pricing, the role of Bitcoin on miners' balance sheets has shifted from a long-term reserve to a type of "immediately disposable financing asset," meaning the duration on the miner side has significantly shortened, and price elasticity is increasingly determined by macro risk appetite rather than the collective emotional reluctance of miners to sell. Under such a structure, long-term volatility patterns have also been reshaped: the risk of unexpected inventory sell pressure at high levels has decreased, yet in the ongoing assessment of mining costs versus AI investment returns, the daily sales by miners become a fixed variable that must be integrated into trading structures, rather than being seen as mere noise.
Core Scientific mining losses, AI surges
If the ongoing sell-offs in the previous section merely represent a shortening of miner duration, then Core Scientific's third-quarter report quantifies this structural shift concretely: its self-mining revenue was only about $12.3 million, plummeting approximately 79% year-on-year, and due to a resonance among halving, electricity prices, and operational costs, it recorded a roughly $7.5 million gross loss for the quarter. Bitcoin on the balance sheet of this "traditional mining enterprise" is no longer a profitable business but a drag. In stark contrast, hosting revenues from AI data centers in the same quarter amounted to about $215.4 million, driving overall revenue to about $233 million, an increase of approximately 187% year-on-year, with adjusted EBITDA around $133.6 million and a profit margin of around 57%, with cash flow almost entirely from hosting and data centers, rather than from mining and retained BTC. The net profit appears to reach about $1.178 billion, but the company also analyzed the sources: mainly from non-cash gains from changes in the fair value of warrants, and the real transformational business model shifts are derived from "high-margin AI hosting + computing power rentals," not Bitcoin price fluctuations.
With this revenue structure, the company's preferences for the same kilowatt-hour and the same rack have been completely rearranged: any computing power that can lock in a 50%+ EBITDA profit margin from AI clients no longer needs to bet on a higher BTC price through "mining + holding coins," but instead directly transforms into a rental, contract-based infrastructure asset with cash flow. For the Bitcoin network, this means the dominant power of the security budget is shifting from "block subsidies + transaction fees" toward "the opportunity cost of electricity and computing power between AI and mining": when returns from AI hosting far surpass those from mining, more cabinets and electricity will prioritize data centers, while the computing power left for the network becomes an adjustable marginal variable rather than the core assets of company operations. The outcome rewrites the relationship between BTC prices and computing power: price increases may not necessarily continuously attract back the computing power taken by AI, while declines will still trigger daily selling from miners; this "reallocation of computing power favoring AI" is transforming Bitcoin from a primary revenue stream for mining enterprises into a variable exposure that requires continuous cost-benefit comparisons.
MARA secures 505 megawatts to prepare for AI computing power
At a time when mining enterprises are withdrawing computing power from the network to cabinets, MARA has chosen to lock in its "supply lines" first. It received approval from the Federal Energy Regulatory Commission (FERC) to acquire Long Ridge Energy & Power, securing approximately 505 megawatts of combined cycle natural gas generation capacity and over 1,600 acres of land all at once, and clarified externally that this electricity and space is reserved for AI and high-performance computing business. This is no longer about the traditional "finding cheap electricity to mine coins," but rather packaging power plants, local land, and future data centers into computing power infrastructure; the energy regulatory agency's access permit directly targets computing power expansion, not just electricity sales.
From a macro variable perspective, MARA's acquisition shifts the key leverage for mining enterprise valuations from solely the price of BTC to "disposable electricity costs + regulatory permits + computing power rental and sale structures." With self-owned power generation and large areas of land, the distribution of the same kilowatt-hour between BTC mining and AI/HPC becomes a balance sheet decision that management can adjust at any time: when post-halving mining profit margins are under pressure and operational costs rise, there is greater incentive for the 505 megawatts to flow toward AI data rooms rather than continue expanding network computing power. For the BTC market, this means mining enterprises now have another route outside high electricity prices and regulatory thresholds: reducing BTC mining to a "marginal exposure," locking in cash flow with AI contracts, and deciding whether to increase or decrease mining scale based on price cycles. What remains to be observed is how MARA's energy acquisitions and expansion progress between this 505 megawatts will further elevate the tilt of computing power toward AI, as well as whether exchanges' security governance over third-party APIs continues to tighten, thereby altering the structural liquidity of contracts and spot.
Liquidations and USDT.D: Short-term longs cleansed
When mining companies discuss AI in conference calls, the specific cost presented on the trading floor is stark: in the past hour, approximately $43.12 million was liquidated across the network, with around $42.17 million from long positions and only about $940,000 from short positions; BTC liquidation amounted to approximately $14.31 million, and ETH liquidation about $12.14 million, with mainstream asset longs nearly wiped out in one sweep. Such "unilateral liquidations" indicate a rapid reversal of risk appetite among short-term funds, as high-leverage chasing the market was forced to exit; the price corrections triggered by mining company sell pressure no longer relate purely to fundamentals but directly rewrite the current leverage structure: what remains on the trading floor are positions that are closer to spot and low leverage.
In this context, trader Killa shifted his focus to USDT.D—the proportion of USDT in the overall crypto market capitalization. He pointed out that USDT.D experienced a strong rejection after precisely testing the 0.786 Fibonacci level; according to his personal technical judgment, Bitcoin may still dip to around $77,000 to $78,000, but this rejection implies that the turnover point for funds to flow back from USDT to risk assets may have preliminary formed. In other words, the systematic sell-offs by mining companies alongside concentrated long liquidations in one hour first cleaned out all the anxious leverage, and then the high-level obstacle at USDT.D lays the groundwork for funds to take on risks again; the short-term trading structure is transitioning from "high-leverage chasing + holding miners" to "spot sell pressure dominating + low-leverage seeking support," with the key being whether the zone around $77,000 to $78,000 can be verified as a new support area amidst subsequent sell pressure and the relative weakening of USDT.D.
Binance revokes SignalPlus API, tightening funds security
On November 5, Binance proactively revoked all API keys associated with the third-party trading platform SignalPlus at the click of a button, while the official statement emphasized that Binance's own system was unaffected, and user funds were secure, but advised affected users to promptly check their positions and pending orders. Shortly thereafter, CZ escalated this action from "an internal risk control incident at Binance" to "an industry-wide interface risk reminder," urging users connecting to other exchanges via SignalPlus to also revoke relevant API keys. The exact nature and scope of the security issue have not yet been disclosed, but for market participants, what has truly been shut down is not a tool but a key channel for quantitative and algorithmic funds to enter the market.
For algorithmically driven and high-frequency funds reliant on APIs, the abrupt loss of the SignalPlus connecting hub means that some automated strategies are forced to pause or reduce their frequency; the dense limit orders that were originally placed by bots in the order book vanish within a short time frame, and the trading structure shifts from "programmatic market-making + high-frequency hedging" to "manual adjustments + shrinking balance sheet and waiting." This directly lowers the order book depth of BTC and ETH near critical price points, making the impact of mining enterprises' sell-off and AI concept trading on prices more intense under the same transaction volume: sustained sell pressure from entities like Keel in a thinning liquidity environment can more quickly drive prices down to the $77,000 to $78,000 support zone, and the narrative around mining enterprises transitioning to AI, after the temporary exit of quantitative funds, shifts from "model-based thematic rotation" to a competition among a few active funds. The outcome of tightening API security is that, under the premise that leverage has already been concentrated and cleansed within one hour, it adds another layer of "transaction channel tightening" to the market, enhancing short-term price elasticity of spot and contracts, yet making BTC and ETH's capability to complete turnover in the new support zone more dependent on real cash flow than on the inertia of algorithmic funds.
New trading lines for BTC and computing power post-mining companies betting on AI
With Keel clearing its positions, Core Scientific announcing that "mining is merely a loss-making side business" in its financial report, and MARA locking in a 505-megawatt natural gas plant and 1,600 acres of land onto its balance sheet, these three lines together rewrite traditional miners from "long-term BTC net buyers + high beta leveraged longs" to "AI/HPC infrastructure operators + cash flow custodians." On the supply side, Keel committing to daily sales of newly mined BTC transforms the previously passive reduction of miners into predictable, even modelable active sell pressure; coupled with the decline in block subsidies post-halving, the mining elasticity of the BTC supply curve is weakening, and pricing relies more on pure buying demand rather than miners’ inventory speculation. On the demand side, the rejection of USDT.D at the 0.786 Fibonacci level corresponds to the clearing of leveraged funds after the $43.12 million liquidations in one hour, with liquidity starting to compare the cost-effectiveness of "mining enterprises' AI cash flow stocks" against "on-chain BTC leveraged longs," shifting the valuation anchor of mining enterprises from BTC output to hosting profit margins and energy asset discounts. Binance tightening third-party APIs and the uniform revocation of SignalPlus keys add a layer of "channel filtering" to this new trading line: with algorithmic funds temporarily constrained, the ability to absorb miners' ongoing sell pressure will depend on real spot buying and equity capital under the AI narrative rather than the inertia of high-frequency market-making. Moving forward, trading desks need to closely monitor several variables: whether the BTC sell-off rhythm from miners like Keel continues to move forward, whether AI hosting businesses like Core Scientific can maintain an EBITDA level above 50%, whether MARA's energy acquisitions and expansion progress further elevate the tilt of computing power toward AI, and whether exchanges continue to tighten security governance over third-party APIs, thereby changing the structural liquidity of contracts and spot.
Join our community to discuss and become stronger together!
Exclusive Hyperliquid benefits for AiCoin: https://app.hyperliquid.xyz/join/AICOIN88
Exclusive Aster benefits for AiCoin: https://www.asterdex.com/zh-CN/referral/9C50e2
On-chain Telegram community: https://t.me/AiCoinWhaleData
On-chain community: https://www.aicoin.com/link/chat?cid=N6OVMor5g
AiCoin on-chain Twitter: https://x.com/aicoinwhaledata
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
