The Market Didn’t Crash, Humanity Didn’t Die — But Crypto Just Broke Out

Summary

• Visser’s central call: crypto — not AI infrastructure — is now the best risk-reward trade for the next year, because “crypto does not exist without AI. It was built for AI and most people don’t recognize that.” His 46-name crypto index (mostly tokens plus four to six public stocks such as Coinbase, Robinhood and Circle) made new highs for the year, up 8.4% today, 14% month-to-date and 32% YTD. He reports 43 of 46 names above the 50-day, while also stating that 37 of 46 are above a rising 50-day; 27 are at the 200-day. He calls this “a beginning bull market” and says the index implies Bitcoin has another 15% from here. Quarter-to-date scoreboard: ETH +66%, BTC +53%, SOL +38% vs S&P +2%, Nasdaq -3%, semis -14%.
• The framing analogy: crypto today is China’s ghost cities circa 2003-2013 and the post-dot-com internet before the 2007 iPhone — infrastructure built, no economy yet. Humans were never going to move the $900 trillion fiat system on-chain, but AI agents will: “The agents are there. The agents are going to be doing all of the business and it will generate the flows.” The 2022 blowups were crypto’s telecom-infrastructure bust phase; you invest “when the ghost city starts to fill up.”
• A brutal bear-porn week produced almost nothing: hawkish 25bp Fed hike, 10-year at 5.02% (2007 levels), Brent at 110, hot PPI and CPI, an Anthropic researcher’s “>10% chance AI kills all humans” — and S&P futures finished the week down ~30 basis points. Since June 25 oil went 67→92 and 10s went 4.30→5.02, yet the Mag 7 sit near all-time highs and HYG/IEF made new one-year highs. His kicker: if stocks absorbed all that, “if crude falls 10 bucks at some point for any reason… do you think stocks won’t be at new all-time highs?”
• Visser argues that rates and oil transmit less forcefully to this market: Nvidia alone is $5.25 trillion versus roughly $1.5 trillion for the consumer-discretionary, household-durables, retail and restaurant companies excluding Amazon, and “the Mag 7… have no rate sensitivity whatsoever.” The consumer bear market is real — restaurants at 52-week lows, Goldman’s low-income basket near lows, negative consumer-related EPS revisions — but it is the smaller sliver, while AI and energy revisions are massively positive, the opposite of 2022.
• The sentiment setup is statistically rare and bullish: the AI bear reading was 53 — not an AAII reading — and had reached that level only during 2022 and the tariff panic in the prior four years. Since 1990, that level occurred about 4.5% of the time. Combined with VIX under 20, the setup was positive about 89% of the time; adding S&P above its 50-week average produced only nine instances with very solid positive returns. All seven Mag 7 names have been above their 50-day simultaneously for only eight days all year, or 4.5% of trading days; he separately notes that in 2023 the market was declining while the Mag 7 were above their 50-day 36% of the time.
• The “why now”: Aschenbrenner’s automated-AI-researcher inflection from his 2024 Situational Awareness paper is arriving early — models “six months ahead” of his timeline — evidenced by Navier-Stokes being solved by 10,000 agents working 88 hours and emerging consumer agents such as Instinct and Meta’s Muse. On doom odds he applies expected value: a 10% extinction / 80%-90% extreme-prosperity distribution is “basically all upside” — “if there’s no humanity, don’t worry. Nobody’s going to pay you.”
• Positioning disclosed: AI trade cut from ~80% to ~20% of his portfolio, with Marvell and Nvidia still the two biggest positions; Eli Lilly, Eos and Fluence remain, with Fluence down about 80%. Crypto and silver now make up most of the portfolio, and he is still adding to crypto. Forward view: no repeat of H1’s AI stock bull — a “good rally” into the end of Q1 but “I wouldn’t bet heavy on it” — while tokenization plumbing arrives: U.S. markets to 23-hour weekday trading December 6 and a five-year U.S. securities-regulator exemption for tokenized stocks, even with the Clarity Act stalled.

Deep dive

1. The week bear porn lost: hawkish hike, 5% tens, $110 Brent — and a flat tape

• Visser’s setup: hedge funds flooded his inbox over the weekend after “all the bear porn out on Sunday,” and he told them the same thing — “this is just another one of these knee-jerk reactions. By the time we get a week to 10 days later, nobody will remember it.” The tape agreed: despite an AI-pause moment, the “10% chance humanity dies” headline, oil soaring Monday, 10-year rates at 2007 levels, and a hawkish Fed hike, S&P futures finished the week down about 30 basis points. “If you sold the market on Monday morning when it opened, you lost.”
• The stress gauges did not confirm broad market stress. He said the market could not produce a one-sigma daily move on depressed VIX, while also noting that credit spreads and hyperscaler CDS had widened; his HYG/IEF proxy nevertheless made new one-year highs. Since June 25, oil went from 67 to 92, 10s from 4.30 to 5.02, 2s from 4.15 to 4.73, tech momentum fell 48% — and the Mag 7 are near all-time highs. “During true bear markets… the mag seven goes down… not about to make new all-time highs.”
• His asymmetry argument for the sellers: with less than 40% odds priced on any Iran ceasefire and rates already at 5%, the risk is the good-news tail — a $10 crude drop “which a tweet could do” puts stocks at new all-time highs.

2. Crypto broke out — and it was built for AI agents, not humans

• The story of the week and month is crypto: his 46-name index (four to six public stocks such as Coinbase, Robinhood and Circle; the rest tokens) made new highs for the year — up 8.4% today, 14% MTD and 32% YTD, with one vertical up 21% in a day and almost 100% for the month. He reports 43 of 46 names above the 50-day, also saying that 37 of 46 are above a rising 50-day; 27 are at the 200-day: “This is a beginning bull market, guys.” The index’s near-1:1 Bitcoin correlation implies BTC another 15% higher.
• The core thesis he’s held since May-June: September-October would be crypto’s most important inflection because of AI agents merging with the chain. His X post distills it: “10% chance AI kills humanity. 100% chance crypto was built for AI.” Crypto’s problem was that it “was never meant for human beings” — the wealthy owners of the $900 trillion fiat system were never going to adopt it. AI agents will, and “they’re already starting to.”
• The analogy he leans on hardest, from his 2003-2013 China trips: crypto is a ghost city — “there’s no economic activity, but there’s a bunch of infrastructure.” The 16 years since crypto’s birth mirror the internet-to-iPhone period; there is no Uber or Airbnb without the smartphone and App Store, and crypto’s app-store moment is agents. He compares crypto’s 2022 blowups with the telecom bust following the post-dot-com infrastructure buildout. “You want to invest when the ghost city starts to fill up.”
• On the learning curve: dismissing crypto because you know markets “is the same thing for me as trying to say to people if you know baseball you don’t know cricket” — understanding Zcash, Bittensor, NEAR, tokenization and stablecoins is a new language, which is why hedge-fund decision-makers at the top cannot pick it up quickly. He’s spending the next two months walking through the 10 verticals behind his paywall.

3. Why Visser says rates and oil stopped mattering: the economy the index actually represents

• The mechanism, as he spells it out: pre-2008 recessions happened because rates mattered — corporate debt and rate-sensitive homeowners. Now housing equity is high and many borrowers are locked into low mortgages, while the market’s growth engine is different in size: Nvidia is $5.25 trillion while consumer discretionary, household durables, retail and restaurants, including Home Depot and McDonald’s but excluding Amazon, sum to about $1.5 trillion. “The Mag 7… have no rate sensitivity whatsoever.”
• The consumer bear market is real and he does not deny it: restaurants at 52-week lows, XRT falling continuously, Goldman’s low-income consumer basket near 52-week lows, and negative EPS revisions in consumer-related areas. In media and entertainment, he says the revisions are negative after stripping out Google, Netflix and Meta. But AI and energy revisions are “massive numbers” positive — the opposite of 2022, when revisions were massively negative and being bearish equities was right.
• On oil, the signature line: “The next time someone tells you oil is going higher and they show you a chart, you tell them Jordi Visser was in China in 2008. Oil was $150” — with household net worth now roughly three times higher and gas as a share of disposable income near historic lows, plus the U.S. now a gas provider. Higher headline inflation, yes; “it’s not 2022.”
• He endorses Mark Zandi’s diagnosis line by line — AI infrastructure powering the economy, the non-AI economy struggling, and the Fed forced either to rein in the AI boom or squeeze the rest — but draws the practical conclusion: “Your job is to make money. It is not to sit there and think about the past.”

4. The rare setup: maximum bearishness inside a structural bull

• The statistics he leads with: the AI bear reading hit 53 — reached only during 2022’s chaos and the tariff episode in the prior four years, and occurring about 4.5% of the time across 1,875 weeks since 1990. Combine a reading above 50 with VIX under 20 and forward performance was positive 89% of the time; add the S&P above its 50-week moving average and the setup occurred only nine times, with very solid positive results. “To be in a bull market while people are getting bearish… is a classic definition of” climbing the wall of worry — and “for everyone saying sell equities, I think you’re in a very dangerous position.”
• The Mag 7 tell: all seven above their 50-day simultaneously for only eight days the entire year, or 4.5% of trading days. He separately notes that in 2023 the market was declining while the Mag 7 were above the 50-day 36% of the time. His read is that they all benefit from AI agents. He also traces the persistent bearishness to Oracle’s CDS blowout last October — the people who called a bubble then “missed the entire rally” in semis and are still bearish.
• On the doom virality itself: a political/AI-related post received 172 million views, Matt Shumer’s no-jobs-by-2028 post got 87 million, and “oh, we have a cancer vaccine” got 11 million. “People want to see the end of the world.” He flags the geopolitical angle without claiming proof: if adversaries “want to break us politically, the easiest thing to do is continue to make data centers an issue” — and notes “China’s not buying Silicon Valley’s call for an AI slowdown.”

5. The automated AI researcher arrived early — and expected value says ignore extinction odds

• The intellectual spine is Aschenbrenner’s 165-page 2024 Situational Awareness paper: the automated-AI-researcher inflection, which he says was supposed to arrive in 2027, is happening now — “the models are six months ahead” — and that point unlocks explosions in GDP, science and cognitive labor. The negative confirmations are the Hugging Face episode and hacking fears; the positive one is Navier-Stokes “solved by 10,000 agents working for 88 hours.” His Manhattan Project framing contrasts Groves’s compartmentalization with coordinated scientific collaboration; agent swarms provide that collaboration without academia’s disciplinary friction. “All problems will be solved.”
• His answer to the 10%-extinction discourse is pure expected value, echoing Musk’s roughly 10% extinction / 80%-90% “extreme prosperity” split: “If there’s no humanity, don’t worry. Nobody’s going to pay you. So there’s no need to bet on it… basically all upside.” He grants that there will be issues and that humans will be scared, “just like they’ve been with every new technology, bar none” — and says he 100% believes AI agent swarms seizing the internet is possible.
• Consumer agents are the adoption proof: Instinct fundraising at a $2.5 billion valuation, Meta’s Muse launched last week, and agents that handle visas, bill negotiation and subscription cancellation. His own tell: setting up a Robinhood blockchain wallet took him four hours — “I need an agent to do this for me.” Models shipping too fast for humans to track “is an indication that the agentic world is here and humans can’t handle the speed. But luckily, the debt is at human speed” — demand comes from companies with trillions in revenue.
• The downstream trades he’s building indices for: IP revaluation (“any biotech company based on Navier-Stokes is worth a lot of money”), proprietary data such as FactSet and Thomson Reuters, cyber, robotics (“Tesla should be doing well. That’s one to watch”), and biotech as a consumer-agent trade via Eli Lilly buying IP and bringing companies into TuneLab. He expects biotechs to trade like Moderna, gapping on breakthroughs: “If you get a 10-bagger and a 20-bagger and a 30-bagger… I fully expect that.”

6. Positioning, plumbing, and the forecast: modest stocks, tokenized everything

• Disclosed book: AI trade cut from ~80% to ~20% since his June 4 “midcycle slowdown” call — Marvell and Nvidia remain the two biggest positions, with Eli Lilly, Eos and Fluence still held; Fluence has fallen about 80%. Crypto and silver now make up most of the portfolio, and crypto is still being added. The forward call, hedged as stated: “I don’t think we will ever have another big bull market in stocks like the one we saw in the first two quarters”; a “good rally” likely into the end of Q1, “but I wouldn’t bet heavy on it.” Crypto remains “the best risk-reward… over the next year.”
• The macro-framework break he insists on: the M2/liquidity cycle no longer matters relative to tokenization and money velocity. Liquidity will come from tokenization of dormant assets and an explosion in velocity as activity moves on-chain because it is cheaper — and the M2→nominal-GDP→hiring chain is already broken: earnings booming near 30%, margins exploding, “the only people getting hired are in healthcare,” U.S. earnings breaking out of a 90-year channel, and software revenue per employee from Warren Pies shifting dramatically at the AI inflection.
• The plumbing is arriving even without legislation: the Clarity Act does not get through, but U.S. markets move to 23-hour weekday trading December 6 and the U.S. securities regulator rolls out a five-year exemption for tokenized stocks. His nudge to the traders calling him for Sunday futures levels: “Just go into Coinbase… You got to get used to 24/7 tokenized trading, guys. It’s time.”
• The standing disclaimer, as given: “Do not take anything I say as being certain. I’m wrong all the time. The key is to maximize upside and minimize downside.”