Zhihu Article Analysis: Why Do So Many on Zhihu Deny a US Stock Crash?
Fact-checking Zhihu @小滑头's five arguments against the US stock crash thesis: CAPE, AI circular funding, Fed Put, dollar hegemony, historical probability
Zhihu Article Analysis: Why Do So Many on Zhihu Deny a US Stock Crash?
Background
In the Zhihu question Why do so many people on Zhihu deny that the US stock market will crash?, author @小滑头 opens with a conclusion: “US stocks are evergreen” never meant stocks won’t fall or won’t have bear markets — it means after every decline, the market comes back and reaches higher ground. He then rebuts the five main “no crash” arguments found on Zhihu: ① AI companies are profitable, so no crash; ② PE of only 22x is cheap; ③ the Fed will backstop; ④ dollar hegemony protects; ⑤ historical crash probability is low. Below is a point-by-point fact-check.
Verdict
I fact-checked this answer’s data. Bottom line: this is one of the few bearish/neutral answers on Zhihu with solid data verification. The core arguments (CAPE at 42x, Hartnett’s “railroad bubble”, the AI capex circular-funding chain, negative free cash flow) mostly hold up — but there are two hard errors and one logical flaw. Score: 7.0/10.
📊 Fact-check (as of Aug 25, 19:28 Beijing time)
| Claim in article | Actual data | Verdict |
|---|---|---|
| Shiller CAPE ~42x | 41.84 (Aug 24, multpl; mean 17.4, all-time high 44.19 in Dec 1999) | ✅ Accurate |
| “PE only 22x” | Current S&P 500 TTM PE 29.49 (multpl Aug 24); 22x is likely forward PE | ⚠️ The number itself is questionable |
| 2022: S&P -18.1%, Nasdaq -32.4% | S&P price -19.4% / total return -18.1%; Nasdaq -33.1% / total return -32.5% | ✅ Basically accurate |
| AI Big 5 capex Q3'25 +72.9% YoY | Big 4 combined $112.5B in Q3'25, +77% YoY (CFO Dive) | ✅ Consistent |
| “All but Microsoft under water on FCF” | Q2'26: Amazon TTM FCF -$7.6B, Alphabet negative for the quarter, Meta cash generation -91%, only Microsoft healthy (CNBC Jul 31) | ✅ Basically holds |
| Hartnett “largest since 19th-century railroad bubble” | BofA Flow Show May 22: AI Big 10 concentration ~48%, second only to the 1880s railroads’ 63% | ✅ True |
Conclusion: virtually every key figure in the article checks out — a rare quality in Zhihu’s finance section.
🔍 Logic breakdown
Three points that hold up:
- “Individual profitability ≠ systemic safety” — This is well-targeted. The lesson of 2000 is not “companies without profits will fall,” but that once the industry chain forms circular funding (A invests in B, B buys from C, C finances A), a break at any link takes everyone down. Today’s AI cross-investment structure does replicate this — and is worse than when the article was written (Alphabet’s $84.75B equity raise, Oracle downgraded to BBB- by S&P).
- Fed Put has preconditions — 2022 is a valid counterexample. But don’t over-extrapolate: the Fed Put worked in 2018 Q4, 2020, and the 2023 regional bank crisis. The real rule is “the Fed Put has a strike price — it fails when inflation constraints bind,” which the article itself concedes.
- The FCF trend is hard data — This carries the most weight. It’s not narrative; it’s the balance sheet.
Three hard flaws:
- ❌ “US market circuit breakers in 2022” is factually wrong — No circuit breakers were triggered at all in 2022 (the last ones were in March 2020). The author misattributed 2020’s circuit-breaker memory to 2022. A small detail, but it exposes the habit of “memory as data.”
- ❌ The dollar-hegemony argument is a strawman the author himself admits is weak — “Countries rotating from Treasuries to gold” is not the end of dollar hegemony; it’s marginal central-bank diversification. The dollar is still ~58% of global reserves. The author says “take these two with a grain of salt,” effectively conceding the strawman — yet still included it.
- ⚠️ “25 years to recover from 1929 vs 5 months in 2020” mixes nominal and real — 1929 didn’t regain its nominal price level until 1954 (real purchasing power took far longer); 2020’s recovery was 5 months in nominal terms. Inflation and Fed policy distort this comparison; “the market’s immunity is evolving” is half an illusion.
The logical contradiction:
The article mocks the “no crash” crowd while concluding “US stocks are evergreen; long-term investors should never let crash panic drive decisions” — which converges with the very people it’s rebutting on the long-term conclusion. The real disagreement is whether a 20%+ correction will happen, and when. The author only argues “it could crash,” without an actionable “when” — an honest but limited stance. Don’t read it as a crash prediction.
📌 My independent take
Current market position (Aug 25, Beijing time): SPX 7652.86 (-0.28%), NDX 25980 (-0.76%), VIX 15.82 (low), Fear & Greed 54.6 (neutral — but junk-bond demand at 93.8 extreme greed vs stock price strength at 26.6 fear: a clear divergence). CAPE 41.84 sits right at the article’s 42x, only ~6% below the 2000 peak of 44.19.
The article was written July 22. What’s happened since: gold crashed (5294→3986→4692), the 10Y spiked to 5.31% then pulled back, Iran sanctions, and the Treasury TGA buyback program. The market is still at highs, but the funding/credit side (extreme greed in junk bonds, hyperscalers surviving on debt and equity issuance) is indeed adding weight to the article’s thesis.
One-line verdict: This is not a “US stocks will crash” prediction piece; it’s a solid risk-management reminder — its real message is: don’t use “evergreen” as an excuse to skip risk controls. I endorse that stance.
Sources: multpl (Shiller PE, 2026-08-24), CFO Dive (2025-11-04), CNBC (2026-07-31), FactSet, BofA Flow Show (2026-05-22), moomoo, CNN Fear & Greed Index, yfinance