13 min left
AI bubble scored against five factors from 24 financial collapses across four centuries

Four Centuries of Collapse: Where the AI Bubble Sits Now

The Signal: Every major financial collapse since 1637 runs on the same five factors: cheap credit, crowded bets, a new-era story, fragile funding, and a tightening pin. The AI bubble scores four out of five as of July 2026. That does not tell you the date. It tells you the season, and it tells you exactly how to build a coaching business that survives the unwind.

In the film Margin Call, the bank chairman lists the crashes from memory: 1637, 1797, 1819, 1857, 1907, 1929, 1987, 2000. Then he shrugs. "It's all just the same thing over and over."

He was right, and now the question is whether the AI bubble is the next line in that speech. So I did what he implied: took 24 major financial collapses across 383 years, scored each one for what actually caused it, and ran the July 2026 AI boom through the same test.

The pattern is cleaner than you would believe. And the answer matters for you, because your coaching business is currently standing inside the biggest capital spending boom in history.

Part 1

24 Collapses, One Machine

Between Tulip Mania in 1637 and the COVID crash of 2020, there were 24 major financial panics. On average, one arrives every 16 years, and the gap has never stretched much past a working generation.

24
major financial collapses in 383 years. Average interval: one every 16 years. Longest safe stretch: about one working generation.
383 Years of Panic · Each Tick Is One Collapse
1637 1700 1800 1900 2000 10 collapses since 1974
Tulip Mania 1637 through COVID 2020. Note the clustering: financial panics arrive faster as capital moves faster.

Tulips, frontier land, railroads, stocks, Tokyo real estate, dot-coms, houses. The asset changes every time. The machine underneath does not.

That is not a metaphor. Score each collapse for the specific conditions present in its buildup and the same short list appears in century after century.

Part 2

The 80/20: Five Factors Behind Every Panic

Five factors explain the overwhelming majority of what happened across all 24 events. Everything else, the fraud, the personalities, the weather, is decoration on top of this structure.

How Often Each Factor Appears · 24 Collapses, 1637 to 2020
Cheap credit + leverage 20/24 Crowded bet on one asset 16/24 Fragile short-term funding 14/24 New-era story 13/24 Tightening pulls the pin 13/24
Factors scored per event from the standard financial-history record (Kindleberger, Reinhart-Rogoff lineage).

Here is each factor in one line.

Cheap credit and leverage is the universal fuel. Tulip futures, 10 percent margin loans in 1929, subprime in 2008, LTCM at 25 to 1. The asset changes. The borrowed money never does.

A crowded bet turns a correction into a collapse. When everyone owns the same thing, everyone sells the same thing.

Fragile funding is the mechanism that turns falling prices into system failure. Banks and shadow lenders fund long-term bets with money that can leave overnight. Panic is a funding event, not a price event.

The new-era story suspends judgment: railroads will remake the economy, stocks have reached a permanently high plateau, land never falls in Japan, the internet changes everything. The story is always partly true. That is exactly what makes it dangerous.

Tightening pulls the pin. Bubbles rarely die of old age. Someone raises the cost of money, and the borrowed-money structure can no longer roll itself forward.

Key Takeaway

Cheap credit funds a crowded bet, a story justifies the price, fragile funding holds it up, and tightening pulls the pin. The panic itself is just the unwind.

The Five-Stage Cycle Every Bubble Follows

Economists Hyman Minsky and Charles Kindleberger mapped the sequence decades ago. Every event above fits it.

The Minsky cycle is a five-stage pattern: displacement (something genuinely new arrives), credit boom (lenders expand to fund it), euphoria (the new-era story goes mainstream and latecomers buy with maximum leverage), the pin (money tightens or a big player fails), and panic (everyone heads for the same exit at once).

1
Displacement
Something real and new arrives
2
Credit boom
Borrowing gets easier every year
3
Euphoria
The story goes mainstream, skeptics look stupid
4
The pin
Money tightens or a big player fails
5
Panic
Everyone heads for the same exit at once

The cycle is not a market phenomenon. It is a human one. Markets are a collective nervous system, the same fear-and-greed loop scaled to millions of people and wired together by the fastest communication technology of the era. The telegraph spread the panic of 1857. The feed spreads this one.

Markets are a collective nervous system. The cycle repeats because the humans running it do not change.

Part 3

The AI Scorecard, July 2026

So where does the AI boom sit against those five factors? Four are present. The fifth is loading. Here is the evidence, factor by factor.

AI Boom vs the Five Factors · July 2026
Cheap credit + leveragePRESENT
Crowded bet on one assetPRESENT
New-era storyPRESENT
Fragile fundingEMERGING
Tightening pulls the pinNOT YET

How concentrated is the AI market compared to the dot-com bubble?

More concentrated than the dot-com peak, by a wide margin. The top 10 stocks now make up roughly 35 percent of the S&P 500, against 25 percent at the 2000 peak. Widen the lens to the AI Big 10 and it reaches about 41 percent.

Share of the S&P 500 Held by the Biggest Stocks
Top 10 · dot-com peak 2000 25% Top 10 · July 2026 35% AI Big 10 · July 2026 41%
Nvidia alone carries about 7.5 percent of the index, more than the energy and utilities sectors combined.

When everyone owns the same thing, everyone sells the same thing. That factor is not emerging. It is the most extreme reading on the whole 383-year timeline.

Is the AI buildout paying for itself yet?

Not yet, and the gap is the single most important number in this debate. The hyperscalers plan roughly 660 to 690 billion dollars of capital spending in 2026 alone. Annual AI infrastructure spend runs near 400 billion dollars. Enterprise AI revenue, the money actually coming back through the front door, is around 100 billion.

The Gap the Story Is Carrying · Annual, USD
Hyperscaler capex plans 2026 $660-690B AI infrastructure spend $400B Enterprise AI revenue $100B
Four dollars of infrastructure spend for every dollar of enterprise AI revenue. The new-era story carries the difference.

The story might close that gap. Railroads eventually paid, and so did the internet. But in 1873 and in 2000, the market stopped waiting before the revenue arrived.

What is circular financing and why does it matter?

Circular financing is when a supplier funds its own customers, who use the money to buy from the supplier: chipmakers investing in AI companies that spend the investment on chips. Revenue looks organic. It is partly recycled.

This is where the fragile-funding factor is emerging. The Bank for International Settlements used its June 2026 annual report to flag an AI capex bust and opaque circular financing as top risks to the global financial system, warning that some assets may be pledged multiple times. The buildout has also shifted from cash to debt: public debt issuance by the hyperscalers is heading toward 230 to 240 billion dollars this year, private AI credit has grown from about 3 billion in 2010 to over 40 billion, and Moody's counts roughly 662 billion dollars of signed data-center lease commitments sitting off balance sheet.

$1.5T
projected total debt financing for the AI data-center buildout. The boom used to run on cash flow. It now runs on borrowed money.

When a boom migrates from cash to debt, it is aging. That is the same costume change the trusts pulled in 1907 and the shadow banks pulled in 2008.

Key Takeaway

Four of five factors are present in the AI boom as of July 2026. History cannot give you the date of the unwind. It can tell you the season, and the season is late.

Why this looks more like 2000 than 2008, for now

The dot-com crash destroyed equity value but left the banking system standing, because the losses landed on shareholders rather than on leveraged institutions. 2008 was different: the losses sat inside banks funded overnight, so falling prices became a system failure.

The AI boom still leans toward the 2000 shape. Nvidia earns real profits at 53 percent net margins, and much of the spending still comes from the most cash-rich companies in history. But every month of debt funding, GPU-collateralized lending, and circular deals moves the structure toward the dangerous version. The thing to watch is not the Nasdaq. It is the funding: private credit exposure, lease obligations, and who blinks first on capex.

Your Move

What This Means for Your Coaching Business

You are not an index fund. But if you sell anything with AI in the offer, the AI bubble is your weather system. Here is how to read it.

First, separate the bubble from the technology. The Panic of 1873 killed railroad stocks, not railroads. The tracks carried freight for the next century. If AI equities correct 50 percent, coaches and creators will still need systems that produce clients. A shakeout actually helps the builders, because it clears out the hype vendors and leaves whoever delivers measurable results.

Second, know which side of the gap you are on. That 400-versus-100 chart is a warning for people selling AI hype and a map for people closing it. Every dollar of real, attributable outcome you deliver is on the survivable side. If your offer is a thin layer on someone else's model, read The Thin Wrapper Trap next, because the correction will find thin wrappers first.

Third, regulate exposure instead of predicting timing. Nobody on the 383-year timeline consistently called tops. The survivors were the ones whose position, cash, and nervous system let them stay calm while the crowd ran. That is true of portfolios and it is true of coaching businesses. Same practice.

Do these five things this week:

  1. Stress-test your revenue. Ask of every offer: does this get cut when clients tighten budgets, or does it survive because it pays for itself?
  2. Price against outcomes, not optimism. Offers tied to client results hold in any season. Offers tied to client excitement do not.
  3. Keep leverage out of the buildout. The single most common factor across 24 collapses (20 of 24) is borrowed money. Build from cash flow.
  4. Bank runway. Three to six months of expenses converts a market panic from a threat into a buying opportunity for attention, talent, and clients.
  5. Keep publishing. In every downturn, the cheapest asset is attention, because everyone else goes quiet. The Brand OS exists to make that output systematic.

The tools that make a lean operation possible are the same ones we recommend every week:

AI
Claude
The reasoning engine behind your systems
Wispr Flow
Voice-to-text for capture at speaking speed
CAL
Cal.com
Zero-friction scheduling for discovery calls

The chairman in Margin Call was right. It is all just the same thing over and over. The people who lose are the ones who believe, each time, that this time is different. The people who endure are the ones who know the machine, respect the season, and build so they never need to guess the date.

You cannot predict the timing of the AI bubble unwinding. You can regulate your exposure to it. Build accordingly.

Want a business that survives the cycle?

We build brand systems for men's embodiment and somatic coaches that run on client results, not market sentiment. Priced against outcomes, built to hold in any season.

Book Your Free Call →
← Back to all posts