Crypto is weak because ETF flows have gone from support to drain: BTC is down 5% in seven days, ETH 10%, XRP 16%, SOL 10%, BNB 4%, with $1.2B of ETF outflows last week, outflows in 5 of the last 6 weeks, and ETH/SOL funds still bleeding. The same risk-off hit tech and semis: NVDA is down 17% in the AI-crash window, semis are down 22% in 5 days, and the rotation has already started into AMD and AVGO while fear/greed sits at 32 in crypto and 33 in stocks. Despite the selloff, the core thesis stays bullish on AI capex and compute economics: top tech is at its cheapest P/E in years, data-center demand and leasing obligations are still exploding, and the drawdown is framed as a cycle reset rather than an AI bubble burst. BTC’s near-term path is tied to ETF flows and macro correlations, with no V-shaped recovery unless a new catalyst pulls money back in.
Crypto fear and greed is 32; stock fear and greed is 33, while BTC is down 5% over 7 days, ETH 10%, XRP 16%, SOL 10%, and BNB 4%.
ETF flows are the main driver: last week’s BTC ETF outflows totaled $1.2B, outflows hit 5 of the last 6 weeks, and this week started bumpy.
BlackRock is said to have dumped $250M-$400M with great regularity over the last 7 days, described as nonstop dumpage.
ETH ETF outflows in September were often $200M a day against a $12B fund base; SOL outflows were $26M with a $1.3B fund base and 5 straight outflow weeks.
Crypto prices are framed as flow-driven: when ETF flows dry up, price goes down.
Tech had a bloodbath, with the top 20 market-cap losers tied to AI names and NVDA down $1.2T in market cap over 30 days.
The rotation out of NVDA is already visible: AMD is up 18% in 2 days and AVGO 14% in 2 days, while AMD, GOOG and SpaceX are said to show a V-shaped recovery.
Over the AI-crash window, NVDA is down 17%, ALAB 6%, and MRVL 9%; MRVL’s drop is blamed on confusion around the Lamentum Light acquisition.
PLTR is down around 5% after 4-5 months of gains; MSTR is down 9.3% over 7 days and MSTR/Block-linked Sailor is down 5.8%.
November is down 6% even though it is historically BTC’s strongest month; there are 20 days left, and the claim is that BTC can still add 6% in a day.
Ecoinometrics is cited saying drawdowns of this scale usually take 12-18 months to recover; the market is about 390 days into the cycle.
BTC would have reached $90K if $5B more had flowed into BTC ETFs, but instead there has been drainage; a V-shaped recovery is unlikely without a new fiat-collapse catalyst.
Glassnode is cited with BTC’s 30-day correlation to the S&P 500 at 0.8, the highest in nearly 5 years; BTC/NASDAQ correlation is also said to be the strongest in 5 years.
BTC is said to move in lockstep with tech and high beta, with a 24-hour lag in the AI crash: it held up on day one, then tanked on day two.
A rule of thumb is given: 10,000 BTC of ETF inflows equates to about a 3.2% monthly return; BTC treasury-company impact is put at only 0.7%.
Equity trading volume is said to have reached 126% of BTC trading volume, while traders chase SKH, SanDisk and MU instead of spot crypto.
The NASDAQ 100 forward P/E is said to be at its lowest since October 2022 and in the first percentile of the last 5 years, with no AI bubble because top tech is cheaper than during the Covid era.
The Philly semiconductor index is said to have seen 2-year forward P/E compress from over 30x to 10x even as the companies grow at 80%+ per year; semis are down 22% in 5 days, the third-largest 5-day decline since 1994.
Credit markets are pricing chip default protection above hyperscalers like ORCL, META, AMZN and MSFT, implying chips are now riskier than hyperscalers.
NVDA is down 25% from its peak, the biggest drawdown since April 2025, and is said to be at minus 2.6 standard deviations in analyst-consensus sentiment after being bought at $88.
Despite the stock drop, demand for AI compute and B200s is said to be up 74% over 12 weeks, fetching $10 per GPU hour, supporting the claim that this is not a bubble.
AMD data-center revenue surged 97% YoY to $8.6B, outpacing its other segments.
Google, Meta, Anthropic and Microsoft are said to carry about $4.5T of multi-year compute lease and purchase obligations, with Anthropic alone at roughly $600B.
Anthropic is said to be going public next week, with valuation talk cut from $2T to $1T; the trade idea is to buy the IPO and sell it 1-2 weeks later.
Capex commitments are said to stay massive until maybe 2028; the AI trade has been aggressive since 2025 and is expected to run through 2028 with casualties along the way.