Everything About AI
NextFin News - Bitcoin fell below $74,000 on Sunday and triggered more than $1 billion in forced liquidations, but the spot move was only the visible part of a three-way unwind. U.S. spot bitcoin exchange-traded funds flipped to net outflows for 2026 after a $1.28 billion withdrawal week, and the AI compute chain that shares bitcoin's marginal buyer was already under pressure when Nvidia's board opened an independent investigation into an alleged foreign intelligence effort targeting founder and CEO Jensen Huang. The question is not whether leverage got flushed; it is whether the flush is exposing a structural crack in the trade that has funded both assets.
The three forces arrived in the same 24 hours: a break below a price level last seen in August, a fund-flow reversal that erased the year's net inflows, and a corporate-governance event at the single largest weight in the AI trade. That overlap is what separates a routine long squeeze from a regime test. Leverage flushes are cyclical and self-correcting. A repricing of the AI capex thesis is not. The rest of this piece separates the two.
The Move: Price, Liquidations, and the ETF Reversal
Bitcoin changed hands at $74,477.82, down 0.32% on the session, after dipping to $73,713.20 intraday, according to market data. That is roughly 8.9% below the token's 50-day average of $81,738.83 and about 36% below the 2026 high near $116,410. The daily relative-strength index sat at 35.07 on November 9, down from 38.19 two sessions earlier, a reading that marks oversold territory without yet confirming a capitulation low. The selloff was not isolated: Coinbase Global, the largest U.S. crypto exchange, fell 1.5% to $134.76 and is now about 58% below its 2026 high of $324.80.
The liquidation cascade was the mechanical heart of the move. More than $1 billion in leveraged positions were closed out in 24 hours, with long positions accounting for the bulk of the forced selling. A crowded long trade unwound by an exchange rather than by choice exits at any available bid, and that price-insensitive selling drives the next trigger level. The break below $74,000 was less a judgment on bitcoin's thesis than a margin call with a deadline.
History gives this number scale. On October 10, 2025, the digital-asset market recorded its largest liquidation event, with roughly $19 billion in notional positions unwound within 24 hours and total open interest falling 27.5% in a 25-minute window. Sunday's $1 billion is an order of magnitude smaller. That does not make it trivial, but it does make it a mid-cycle flush rather than a record deleveraging. The distinction matters because the market's reaction to a $1 billion flush tells you more about current fragility than about the event itself.
Behind the spot move sat a slower, more consequential reversal in institutional flows. The 12 U.S.-listed spot bitcoin funds logged $1.28 billion in net outflows for the week ended November 6, their worst weekly sum since late June. Those withdrawals pushed the group's 2026 net flows to negative $914.1 million, flipping the year to date into outflow territory for the first time. BlackRock's iShares Bitcoin Trust accounted for $403 million of Friday's gross outflows, roughly 88% of the day's total, and about $1.21 billion of the weekly withdrawals. The pattern extended beyond bitcoin: spot ether ETFs lost $365.3 million, a sixth consecutive negative week, while Solana and Hyperliquid ETFs each posted a fifth straight week of outflows, the longest such streaks since those funds began trading. When the entire crypto-ETF complex bleeds at once, the driver is usually not coin-specific news. It is a portfolio-level decision to cut risk.
Why Bitcoin and AI Chips Are Moving Together
The same week that bitcoin broke down, Nvidia's stock was trading near $186.64, up 2.78% on the day but still about 23% below its 2026 high of $243.37 and well under its 50-day average of $219.81. The correlation is not a coincidence. Crypto and AI compute share a marginal buyer: the investor willing to hold long-duration, high-volatility exposure when liquidity is abundant and confidence is high. That investor does not keep bitcoin and semiconductor stocks in separate mental accounts. They sit in the same risk book, and when the discount rate on future earnings rises or the appetite for speculative duration falls, the same hand reduces both.
The transmission runs through the balance sheet, not the thesis. The AI trade and the crypto trade are financed from the same pool of risk capital. When that pool shrinks, both assets are sold not because their fundamentals changed, but because the balance sheet that funded them got smaller. This is the second-order effect that a headline reading of "bitcoin drops on Nvidia news" misses. The board investigation did not break bitcoin's protocol or ban its use. It raised the governance risk premium at the center of the AI trade, and the shared marginal buyer responded by cutting exposure to both.
There is also a real economic link beneath the shared investor. Graphics processors built for AI training belong to the same architecture family that powers proof-of-work mining, and the capital-expenditure cycle that funds AI data centers competes with the capital available for every other high-growth bet. A slowdown in AI capex does not just hurt chip stocks; it reduces the perceived terminal value of the entire compute-intensive economy, and bitcoin is priced as a claim on that economy's future liquidity. When the anchor asset of the compute thesis wobbles, the satellites wobble with it.
"The NVIDIA Board is initiating an independent investigation in light of recent reporting that founder and CEO Jensen Huang has been the target of a foreign intelligence effort."
The board said a committee of independent directors, assisted by outside independent counsel, would oversee the inquiry, and that it had no reason to believe national security or Nvidia secrets were compromised. It added that Huang fully supports the review and has committed to cooperating, and that the board retains confidence in him and his leadership team. The statement deliberately separates three claims: the alleged intelligence effort, the board's investigation, and its confidence in the CEO. An admission that an investigation exists is not evidence that espionage occurred, and the board's language makes that separation explicit.
Even so, the governance event introduces a discount before any finding of wrongdoing. Nvidia is the largest weight in the AI trade, and a board-level inquiry into an alleged foreign intelligence effort targeting the CEO carries a cost investors can price without resolving the underlying allegation. The cost is the possibility of distraction, the risk of delayed decisions, and the chance that the inquiry itself becomes a recurring headline. None of that requires any finding of misconduct. It only requires that the question now exists, and the market prices the question.
The ETF Channel: How Outflows Become Forced Selling
The fund-flow reversal is the transmission mechanism that turns a sentiment shift into a price break. Spot bitcoin ETFs are not passive observers of the market; they are mechanical participants. A net inflow forces the fund to buy bitcoin in the spot market to back newly created shares. A net outflow forces it to sell bitcoin to meet redemptions. The fund does not take a view. It transmits the investor's decision directly into the market.
That is why a week of $1.28 billion in outflows is more than a sentiment indicator. It is a quantified measure of selling pressure that had to be absorbed by spot buyers. When those flows are concentrated in a single fund, as they were in BlackRock's IBIT, the selling is not evenly distributed across the trading day. It arrives in blocks, and it tests the depth of the order book at specific moments. A thin book gaps, and that gap is what triggered the $74,000 liquidation cascade.
The timing amplified the move. The outflow week ended November 6, and the spot break followed into the weekend, when liquidity is thinnest. A selloff that would have been absorbed on a Tuesday became a Sunday headline. Weekend markets do not forgive crowded positioning; they expose it. The same mechanism works in reverse when flows turn: forced buying into a thin book is what drives the blow-off tops that follow these flushes.
Cyclical Flush or Structural Shift?
This is the judgment that separates a buying opportunity from a regime change. The evidence points to a cyclical drawdown overlaid on a structural question that has not yet been answered, and the two must be kept separate.
The cyclical case is strong, and it rests on three historical comparisons. First, the October 2025 liquidation event unwound $19 billion in 24 hours, and the market absorbed it without a structural break in the uptrend that followed. Second, the $4.5 billion in ETF outflows in June alone did not end the 2026 rally; the funds later attracted inflows again, showing the flow channel is mean-reverting. Third, the current RSI of 35 sits in oversold territory similar to prior mid-cycle flushes, where the price recovered once leverage was cleared. The common thread is a pattern: leverage builds, a trigger forces liquidation, the flush creates the next low, and flows revert. The mechanism that forced selling today will force buying when flows turn.
The structural question sits in the AI compute chain, and it is not answered by liquidation history. If the AI capital-expenditure cycle is peaking, then the shared marginal buyer is not taking a pause; they are repricing the terminal value of the entire duration complex. A 23% drawdown in Nvidia from its high, a 58% drawdown in Coinbase, and a 36% drawdown in bitcoin from its 2026 peak are not the signature of a healthy trend taking a breather. They are the signature of a market deciding how much of the last two years of growth was real earnings and how much was multiple expansion.
The honest read is that both forces are present. The short-term leg is cyclical: leverage, thin weekend liquidity, and a flow reversal that will revert on its own. The long-term leg is structural: the AI capex cycle is being stress-tested, and bitcoin's correlation to that cycle means it will not fully decouple until the capex question is answered. Treating the whole move as one thing is how investors buy a cyclical dip into a structural wall, or sell a structural breakout as a cyclical top. The trade now is to act on the cyclical signal while watching the structural one.
The Counter-Thesis: What If This Time Is Different?
The strongest case against the cyclical-flush view is that the AI capex cycle and the crypto leverage cycle peaked together, and their shared funding source is not just rotating but exiting. In that reading, the $1.28 billion ETF outflow week is not a pause but the first visible sign that the marginal buyer has left the room. The supporting evidence is real: spot bitcoin ETFs have now flipped negative for the year, ether ETFs are in a sixth straight week of outflows, and Nvidia is trading a quarter below its high. If the buyers of 2025 are the sellers of 2026, then every historical flush comparison is a trap, because the regime that produced those recoveries no longer exists.
The answer is that the flow data alone does not settle it; the capex data does. ETF flows are cyclical by construction; they revert because the same investors who redeem today can create shares tomorrow. AI capex is structural because it is committed multi-year spending by hyperscalers, and it does not reverse on a weekly print. The two should be watched as separate signals. If ETF flows turn positive while AI capex guidance holds, the cyclical view is confirmed. If ETF flows turn positive but hyperscaler capex guidance is cut, the structural view wins and bitcoin's correlation to the AI trade becomes a headwind rather than a tailwind.
The falsifying signal for the cyclical-flush thesis is specific and observable: spot bitcoin ETF flows turning positive for three consecutive weeks while the price remains below $77,000 would show the outflow channel has reversed without a price recovery, breaking the mechanical-seller thesis. The second signal is Nvidia reclaiming its 50-day average near $219.81, which would indicate the AI compute trade has stabilized and the cross-asset headwind has lifted. If neither prints within the next month, the structural read gains force.
What to Watch Next
In the short term, watch the daily ETF flow print and the $73,713 intraday low. A daily close below that level shifts attention toward the $73,000 to $72,500 zone around August's breakout, and then the psychological $70,000 level if that fails. On the upside, the first resistance is the 200-day moving average near $76,500, then the $80,000 region. The RSI at 35 suggests the downside is extended, but oversold can stay oversold while forced selling continues.
In the medium term, watch the Nvidia investigation's next update and AI capex commentary from the hyperscalers. The board said it would provide further updates as appropriate; each one will be read for scope and timeline. A narrow, fast inquiry is priced very differently from an open-ended one. The board's statement that it has no reason to believe secrets were compromised is the best-case framing, but the market will wait for the committee's findings before treating the governance discount as resolved.
In the long term, watch whether bitcoin decouples from the AI trade. If Nvidia stabilizes and bitcoin continues to drift lower, the problem is crypto-specific: leverage, regulation, or liquidity. If bitcoin rallies while AI chips keep falling, the decoupling has begun and the shared-marginal-buyer thesis is losing force. Either outcome is tradable; confusion between them is not.
The central takeaway is that this selloff is a leverage and flow event first, and a fundamentals event second. But in a market where the same balance sheet funds both the AI trade and the crypto trade, a governance shock at the center of one is a risk event for both. The board said it has confidence in its CEO. The market will wait for the inquiry's findings before deciding whether that confidence is enough, and until then the risk premium on the entire compute complex, bitcoin included, stays elevated.
Explore more exclusive insights at nextfin.ai.
