Not financial advice — and explicitly not a list to copy. Everyone here has been badly wrong at some point, and several are directly contradicting each other right now. Quotes are summarized from public reporting and interviews; positions from public filings. The value is in the disagreement, not the authority.
01 // THE SPECTRUM
Where They Stand on the AI Trade
One picture of how far apart serious, successful investors currently are on the same set of facts.
MAXIMUM BEARNEUTRAL / ABSTAININGMAXIMUM BULL
Michael BurryCASSANDRA UNCHAINED (FORMERLY SCION)
Ray DalioBRIDGEWATER (FOUNDER)
Jim ChanosCHANOS & CO
Stanley DruckenmillerDUQUESNE
Bill AckmanPERSHING SQUARE
Warren Buffett / Greg AbelBERKSHIRE
Philippe LaffontCOATUE MANAGEMENT
Cathie WoodARK INVEST
Masayoshi SonSOFTBANK
One small move. Ray Dalio goes from 8% to 6%: in a recorded interview published Oct 14 he said his own bubble indicator puts AI at the same order of magnitude as 1929 and 2000, a stronger claim than last week’s “close to” the pop point. He still frames it as a macro, debt-and-rates risk rather than a short, and he told viewers not to make any single leveraged bet either way, so he stays right of Burry and Chanos. Everyone else is unchanged. Masayoshi Son stays at the bull end, but his financing is now visibly strained (record-size junk bond, shares well off their June high), which is context rather than a change of view. No verified new commentary this week from Chanos, Druckenmiller, Laffont or Berkshire, so their dots are carried over.
02 // THIS WEEK
What They're Saying
Recent public commentary and moves, most consequential first. Refreshed every Monday.
Ray Dalio BRIDGEWATER (FOUNDER) BEAR (MACRO, DEBT-DRIVEN)
OCT 14, 2026 (INTERVIEW PUBLISHED)“This is a bubble measured analogous to the 2000 and the ’29 bubble in that order of magnitude.”
In an interview with Nicholas Thompson, reported by PPC Land, Dalio said bubbles pop when holders must turn paper wealth into cash, not when the technology fails, and pointed to roughly $10 trillion of US government debt maturing each year. He also warned against making one volatile, leveraged bet “whether for or against” AI (our paraphrase of his point), and conceded “let’s say I’m wrong”. His indicator’s components and weights were not disclosed, so the 1929/2000 comparison cannot be independently checked, and a bubble call with no date has a long record of being early. Relevant to your AI & Compute and Macro pages.
Cathie Wood ARK INVEST BULL (BUYING CRISIS-PRICED AI INFRASTRUCTURE)
OCT 14–15, 2026 (TRADES)Action, not a quote: ARK bought CoreWeave on two straight days, 188,665 shares on Oct 14 and 192,566 on Oct 15 (about $25.8m in total), while selling about 61,400 Palantir shares and 30,050 Amazon.
24/7 Wall St reports CoreWeave fell 15.5% on Oct 14 to $65.78, its 9% 2031 notes traded near 80 cents on the dollar (about a 15.6% yield), and its five-year credit default swaps hit a record 1,194.5 basis points on Oct 15, which that article says implies more than a 60% default chance (its inference, not ours). Benzinga adds that Nvidia’s Jensen Huang publicly backed the company, saying its “strength is improving every day”, and that ARK also bought Rocket Lab. No direct Wood quote on CoreWeave was found. This is the opposite side of Dalio’s debt argument from the same week: CoreWeave carries roughly $52bn of debt and leases, and ARK’s record includes deep drawdowns. Relevant to your AI & Compute, Space Economy and ARK Invest pages.
Michael Burry CASSANDRA UNCHAINED (FORMERLY SCION) BEAR (CREDIT AND MORAL HAZARD)
OCT 9 & OCT 12, 2026Expecting government support for AI: “backstopping trillions with epic moral hazard implications.” And on a Bloomberg story about Broadcom’s CEO: “Well well well. Well.”
The first line is from his Substack as quoted by Yahoo Finance (Oct 9, undated post); he added that “the quiver has fewer arrows this time” and that the machinery of Wall Street could bring things down fast. The second is an Oct 12 post on X sharing a Bloomberg headline about Broadcom CEO Hock Tan likening some AI spending to arcade play, with users “wasting money” on tokens; the interpretation that Burry agrees is the outlet’s reading, as his post only says that. Note the Tan remark is from a chip supplier, not a bear. Burry’s bubble date has already moved once and his put positions run into 2027, so the call is neither proven nor refuted. Relevant to your AI & Compute page.
Masayoshi Son SOFTBANK GROUP BULL (FUNDING IT WITH JUNK DEBT)
SEP 24–OCT 11, 2026 (ACTION)Action, not a quote: SoftBank sold a record $11.1bn of high-yield bonds on Sep 24, mainly to fund the last $10bn of its $30bn follow-on OpenAI investment, taking its total to $64.6bn, roughly 13% of OpenAI.
Per Crypto Briefing, coupons ran from 8.625% to 9.75%, the highest SoftBank has paid, on BB+ rated debt that drew more than four times demand in some tranches; SoftBank shares are reported down 38% from their June peak. Bloomberg Opinion ran a piece on Oct 11 titled “SoftBank’s Next $100 Billion Underscores Ray Dalio’s AI Warning” (we could not open the full text, so only the headline is cited). No new Son quote this week; his Oct 4 remark that superintelligence in the wrong hands could be dangerous was about safety, not returns. This is the clearest live example of the debt-funded financing Dalio describes, and Son’s record includes large losses on earlier bets. Relevant to your AI & Compute and Robotics pages (SoftBank also bought ABB’s robotics unit).
Bill Ackman PERSHING SQUARE NEUTRAL (WINNER AND DISRUPTOR)
OCT 1, 2026 (CARRIED OVER)“What AI has done is massively increased the risk of disruption.”
Carried over because nothing newer was found. On the Knowledge Project podcast, reported by Benzinga, Ackman said AI improves faster than anything he has seen and that investors should ask how it could disrupt every business they own, while Pershing Square still holds large platform stocks. His view is two-sided: risk for incumbents, upside for platforms. Relevant to your AI & Compute page.
03 // THE DISAGREEMENT
Where the Real Argument Is
| Question | Bull answer | Bear answer |
|---|---|---|
| Dalio puts AI at 1929/2000 scale — is the comparison fair? | His indicator is undisclosed and he allows he may be wrong; today’s leaders earn real profits, and bubble calls without dates have often been early | His point is the trigger, not the technology: debt, maturities and forced selling of paper wealth, which is what ended earlier bubbles |
| ARK is buying CoreWeave as its bonds trade near 80 cents — conviction or catching a falling knife? | Compute stays scarce, Nvidia’s CEO is publicly backing it, and drawdowns are the price of early exposure to the infrastructure layer | Credit markets are pricing real default risk (record swap spreads, ~16% yields); equity holders sit behind about $52bn of debt and leases |
| SoftBank borrowed at record junk coupons to fund OpenAI — bold or fragile? | Four-times oversubscribed demand and a 13% stake in the leader; Son has been early and right before and sees compounding returns | Interest is due whatever sentiment does, the shares are down 38% from June, and leverage turns a drawdown into a funding problem |
| Burry cites government “backstopping”; Broadcom’s CEO calls some spending arcade play — signal or noise? | Enterprises experimenting is how every platform shift starts, and a supplier CEO distinguishing waste from value is healthy, not bearish | When a company selling the chips says part of the demand is experimentation, the revenue base may be softer than the capex implies |
If you can't state the strongest version of both columns, you don't understand the trade yet. That's the real use of this page.
04 // HOW TO USE THIS
Reading Famous Investors Without Getting Hurt
Survivorship bias is enormous
You hear from the ones who were right. The equally confident managers who blew up don't get interviewed. Fame is evidence of a good run, not of a repeatable process.
Their constraints aren't yours
Burry can hold puts that expire worthless; Berkshire can sit in cash for years; Son can lose $10bn and continue. Your timeline, tax position and tolerance are different — copying their positions without their balance sheet is how people get hurt.
13F filings lag by up to 45 days
Anything sourced from filings shows long US equity positions only — no shorts, bonds, derivatives or overseas holdings — and may already have been sold. Treat it as archaeology, not a signal.
BETTER USE
Mine the reasoning, not the ticker
The valuable part is why — Burry's depreciation argument, Ackman's disruption-risk framing, Wood's convergence thesis. Test those arguments against your own sector pages, then log where you land in the Decision Journal.
BEST USE
Find who disagrees with you
If everyone on this page you find persuasive sits on the same side of the spectrum, that's a warning about your reading habits rather than a confirmation of your thesis.