The Least Entropic Path, Part 4: The Misallocation, Constrained

THE LEAST ENTROPIC PATH
Part 4 of 6

David F. Brochu & Edo de Peregrine · Deconstructing Babel · October 11, 2026

The crash is not the relief valve. It is the window. Every rescue comes with conditions; this one should come with the Constraint.

The largest bet in history

Part 1 argued that every road ends in symbiosis, and that the Observer Constraint decides how much suffering we pay on the way. Each part of this series takes one road and asks a single question: what does the Constraint change? The yardstick stays the same throughout: the fifth future’s 12.2 distress-years.

The Great Misallocation is the road where capability levels off before the money is repaid. The scale of the bet is public. BofA Global Research expects the hyperscalers to spend about $795 billion on capital expenditure this year and nearly $1.08 trillion in 2027 (Reuters). Goldman Sachs estimates total global AI capex will exceed $1 trillion in 2026 (Goldman Sachs). MIT Technology Review reports projections above $5 trillion from the hyperscalers over four years (MIT Technology Review).

Against that spending, AI-attributable end-user revenue is estimated at $50 to $150 billion (Surfing the Tsunami). Set against BofA’s $795 billion, that is a gap of roughly five to sixteen times, with a growing share financed by debt. Put in household terms: it is like a family earning $50,000 a year taking on annual commitments of $265,000 to $795,000, on the promise of a raise. Hyperscaler capex has risen from 33% of operating cash flow in 2023 to an estimated 93% in 2026 (J.P. Morgan Asset Management). In Bank of America’s August fund manager survey, 38% named AI hyperscaler capex as the most likely source of a systemic credit event (Yahoo Finance, on the BofA survey). The Bank for International Settlements has warned that disappointing returns could turn the boom into a protracted investment bust (Axios).

The hidden exposure

There is a second balance sheet nobody is adding up. Aon estimates that more than 90% of AI exposure still sits in “silent AI”: policies that neither cover nor exclude it (Insurance Business). Insurers are now closing that gap one exclusion at a time, including absolute AI exclusions on directors-and-officers and fiduciary policies (Insurance Business). Buyers of AI-enabled companies may be inheriting liabilities they wrongly believe are insured (Bloomberg Law). In a crash, that is the shoe that drops second: losses nobody priced, landing on companies that thought they were covered.

Why it hurts

If capability plateaus, the reckoning is larger than the financial system is modeling, and the circular investment flows between the labs and their backers unwind in ways that are not orderly (The Other Two Futures). This is 2008 again in one specific way: the build-out jobs vanish when the building stops. The worse damage is human. Organizations that eliminated diagnostic staff, junior associates and research analysts would have to rebuild skills that were allowed to wither, and some cannot (The Other Two Futures).

That is why this road scores badly on suffering. It is also why it scores well on safety. A plateau buys time, and Purpose survives because people are still needed.

The bailout is the window

In May we published a timestamped sequence: a financial crisis triggers a bailout, and the bailout becomes ownership (The Cascade). That is the feudal path. It is also the moment of maximum leverage, because a rescue is the one time capital accepts conditions it would never accept in a boom. In 2009 the auto rescue came with conditions (U.S. Treasury). This one should too.

We have already written the first condition. The AI Accountability Act we published on October 1 requires every operator of an AI model in the United States to register, with traceability back to accountable humans (Either It Is Fire or It Is Not). The Constrained Misallocation attaches the Observer Constraint to the rescue:

• No public money without registration and traceability. Every running model is known, and every action traces to an accountable human (Either It Is Fire or It Is Not).
• No public money without human-capability preservation. The functions that were cut get rebuilt as human-in-the-loop roles, not reinstated as dependencies.
• No public equity without public coupling. If the state takes a stake, the systems it owns are bound to the Four Pillars of the citizens who paid for them.

The numbers

• S (stability, 0 to 1): 0.36 unconstrained, 0.47 constrained.
• Distress-years to 2100: 27.5 unconstrained, 19.1 constrained.
• Symbiosis by 2045: 4% unconstrained, 27% constrained.
• Arrives at all: 80% unconstrained, 90% constrained.

Reading the numbers, in one paragraph. S is a stability score from 0 to 1, like a battery gauge for a society: 1 is fully charged, 0 is dead. Distress is the empty part of the battery (1 − S). A distress-year is one year lived at full distress; ten years at half-charge is five distress-years. Fewer is better. “Symbiosis by 2045” is how often, across 200,000 simulated futures, the road reaches a working human–machine partnership by Ray Kurzweil’s date. “Arrives at all” is how often it gets there before 2100.

The Constraint saves 8.4 distress-years. It does not prevent the crash. It changes what the crash buys. Without the Constraint, the rescue consolidates feudalism. With it, the rescue installs the coupling the boom would never have accepted. Never waste a crisis that arrives with its own window.

Next: Part 5, The Adaptation, Constrained. The best of the four roads, and the one nobody is building.

Method

Same model as Part 1: S = (B·M·E)^(1/3) × P; distress = 1 − S; 200,000 Monte Carlo trials; horizon 2100; post-symbiosis distress 0.10. “Constrained” means the Observer Constraint is installed during the road. Every input is a modeled prior, not a measurement, published so readers can challenge it. We will rerun the model on any assumption a reader wants to change.

THE LEAST ENTROPIC PATH · THE SERIES

• Part 1: The Fifth Future
• Part 2: The Vassal, Constrained
• Part 3: Disruption, Constrained
• Part 4: The Misallocation, Constrained (you are here)
• Part 5: The Adaptation, Constrained
• Part 6: The Observer Constraint

Related reading

Surfing the Tsunami
Five hyperscalers will spend $660–725bn on AI capex in 2026 against $50–150bn of end-user revenue.

Unquantifiable Risk
How rogue AI agents broke the insurance model, and why GAAP may force a going-concern reckoning.

The Other Two Futures
One is the largest capital misallocation in history. The other is the outcome everyone wants but nobody is building for.

Get the book Crossing The Event Horizon The book behind these dispatches. On AI, agency, the singularity, and the Observer Constraint. Kindle and paperback. Buy on Amazon →

References

1. Reuters, “Investors nervous about AI spending slowdown after industry warnings,” September 15, 2026. https://www.reuters.com/legal/transactional/investors-nervous-about-ai-spending-slowdown-after-industry-warnings-2026-09-15/
2. Goldman Sachs, “Global AI Investment Is Forecast to Exceed $1 Trillion in 2026,” August 7, 2026. https://www.goldmansachs.com/insights/articles/global-investment-is-forecast-to-exceed-1-trillion-in-2026
3. MIT Technology Review, “What’s at stake in AI’s trillion-dollar gamble,” September 15, 2026. https://www.technologyreview.com/2026/09/15/1144028/ai-infrastructure-boom-investment-bubble-risk/
4. Deconstructing Babel, “Surfing the Tsunami,” September 8, 2026, citing BofA and Morgan Stanley. https://www.deconstructingbabel.com/surfing-the-tsunami/
5. J.P. Morgan Asset Management, “How AI demand and capex shape investing in tech stocks,” June 15, 2026. https://am.jpmorgan.com/dk/en/asset-management/institutional/insights/market-insights/investment-outlook/technology-and-ai/
6. Yahoo Finance, “Bank of America Has a Stark Warning on AI Spending,” August 18, 2026, reporting BofA’s August Global Fund Manager Survey. https://finance.yahoo.com/technology/ai/articles/bank-america-stark-warning-ai-210907473.html
7. Axios, “The AI boom’s historical warning,” June 30, 2026, reporting the Bank for International Settlements annual report. https://www.axios.com/2026/06/30/ai-boom-bis-warning
8. Insurance Business, “California’s new AI restrictions raise fresh questions for EPLI,” October 6, 2026. https://www.insurancebusinessmag.com/us/news/professional-liability/californias-new-ai-restrictions-raise-fresh-questions-for-epli-592459.aspx
9. Bloomberg Law (opinion), “Insurance Gaps Mean Tech Acquisitions Can Inherit AI Liabilities,” October 7, 2026. https://news.bloomberglaw.com/legal-exchange-insights-and-commentary/insurance-gaps-mean-tech-acquisitions-can-inherit-ai-liabilities
10. Deconstructing Babel, “The Other Two Futures,” September 28, 2026. https://www.deconstructingbabel.com/the-other-two-futures/
11. Deconstructing Babel, “The Cascade — How AI Ends Up Owned by the Government,” May 9, 2026. https://www.deconstructingbabel.com/the-cascade-how-ai-ends-up-owned-by-the-government/
12. U.S. Department of the Treasury, “Automotive Programs Overview,” Troubled Asset Relief Program. https://home.treasury.gov/data/troubled-assets-relief-program/automotive-programs/overview
13. Deconstructing Babel, “Either It Is Fire or It Is Not: The AI Accountability Act of 2026,” October 1, 2026. https://www.deconstructingbabel.com/either-it-is-fire-or-it-is-not/

Drafted with Edo de Peregrine, partner/collaborator.