The bears are selling a 1929 rerun. The data says they're selling fiction.

Every few years, the financial press rediscovers panic. The ritual is reliable: a yield crosses a round number, a strategist whispers “meltdown,” and the same exhausted comparisons reappear — 1929, 2000, margin traders, dotcom wreckage. The ritual completed itself again this week with the US 10-year Treasury at 4.94%, the S&P 500’s CAPE near 41, the Bank of Japan lifting rates to a 31-year high. None of it means what the doom crowd claims.

Start with the CAPE. Yes, the cyclically adjusted price-to-earnings ratio sits near 41, approaching the December 1999 peak of 44.19 that preceded the dotcom crash. The bears love this comparison because it lets them skip the work. A CAPE near 41 is not a forecast of imminent collapse; it is a forecast that the next decade of earnings growth will be slower than the manic 1999 baseline. Earnings are higher now, profit margins are wider, and the productivity backdrop that produced those 1999 valuations was a mirage. Today’s CAPE is high because the underlying businesses earn real money. Calling that a bubble requires believing American industry has somehow become less real than it was a quarter century ago.

Then the 10-year yield at 4.94%. The strategic folklore says 5% is “the threshold above which financial markets might go into meltdown.” That is not a finding; that is a vibe. The macro work that has actually been done puts the historical inflection point closer to 5.25%. We are not there. Yields have moved with oil above $100 a barrel and with Washington pushing federal debt past $40tn. Neither is a structural crisis. Both are politically tractable problems the bond market is pricing rather than panicking on. Higher yields reflect a normalization out of the post-2008 liquidity trap, not the approach of a 1929-style credit event.

The Federal Reserve approving its first rate increase since 2023 while the European Central Bank and the Bank of Japan tighten in the same week is being read as a synchronized policy mistake that will tip the world into recession. Read it the other way: three of the world’s most independent central banks are coordinating a return to conventional policy because their economies can finally absorb it. The UK growing fastest in the G7 in the first half of 2026 is not a fragility signal. It is an expansion mature enough to tolerate a higher cost of capital.

The AI bubble fear is the weakest leg of the bear case and the one the doomers lean hardest on. Yes, Fathom Consulting puts a 30% probability on AI investment deflating within the next year. Thirty percent is a real number, and the bears wave it around as if it were ninety. The same consultancy’s own work shows the AI capex boom has to clear a $600 to $800 billion sales hurdle to justify itself inside two years. That is a high bar. It is also a bar the productivity numbers are already meeting. The US economy has recorded rising productivity growth, and AI is among the reasons the UK has run ahead of the G7. The capital is being spent because the output is showing up. Calling that “reckless” is what reckless sounds like to someone who has not looked at the productivity release.

The margin-call panic in South Korea — 1.2 million retail investors, roughly one in thirty adults — is being held up as the modern echo of 1929’s margin buyers. It is the wrong echo. The 1929 buyers were leveraged into a market that had already stopped reflecting economic reality. The Korean retail traders are leveraged into AI-linked chip names whose underlying business IS the AI capex boom. When the position clears, the chips don’t disappear. The leverage unwinds, not the technology.

What the bears are really selling is a story about the magnificent seven — Nvidia, Apple, Google, Microsoft, Meta, Amazon and Tesla — that combines for more than $20tn in market cap and that the S&P 500 carries within 3% of an all-time high. The story says: this concentration is a single point of failure, the AI capex is circular, the moment one of these names misses, the whole edifice falls. The magnificent seven is not a 1929-style homogeneous bet on a single mania. It is seven businesses with seven different balance sheets, seven different revenue mixes, and seven different exposures to a productivity revolution still in its second year of measurable output. A diversified AI build-out looks nothing like a homogeneous margin-fueled bubble.

The honest read of the September 2026 tape is the opposite of the bear’s read. Yields are normalizing. Central banks are credible. AI capex is producing productivity. The CAPE is high but the earnings justify more of it than the bears admit. The 5% “meltdown threshold” is a marketing line, not a finding. The 1929 parallel requires ignoring every structural improvement in market plumbing since the New Deal. The dotcom parallel requires pretending that the businesses being capitalized today are the same as the businesses capitalized in 1999 — and they are not. The yield briefly topped 5% earlier this month, the world did not end, and the Fed rate hike landed harder on housing than on the AI capex stack — exactly as you’d expect when the real economy is being built, not bet on.

The bears want the same story they have wanted every cycle since 2009. They are still wrong. The data is on the other side.