The Cascade: How AI Ends Up Owned by the Government

A timestamped prediction. The most consequential political realignment of the decade, in four steps.

A Prediction We Want on the Record

We want to make a prediction. We want to timestamp it publicly so that when it happens, the record shows it was seen coming.

The President will pivot to anti-AI populism. Not immediately. Not while the markets are still buoyant and the tech executives are still flying to Mar-a-Lago. But when the public outcry is loud enough and the chorus is building — he will. And when he does, it will be the most consequential political realignment of the decade.

Here is the sequence.

Step One — The Backlash Is Already Building

This is no longer speculative. Today — May 8, 2026 — The New York Times Magazine published a long-form essay by John Ganz titled "A.I. Populism Is Here. And No One Is Ready," documenting the convergence of three movements that had previously operated in separate political universes: town-hall revolts against data-center construction, organized labor pushback on professional-class displacement, and a growing public framing of AI labs as "the new face of American oligarchy."¹ The essay was not published by an obscure outlet or a partisan blog. It was published in the cultural pages of the country's paper of record. That is not the early signal. That is the late signal.

The polling confirms the trajectory. The February 13–16, 2026 Economist / YouGov poll found that 63 percent of Americans expect AI to reduce the number of jobs in the United States, only 7 percent expect it to increase them, and 58 percent report low or no trust in AI. A clean majority — 54 percent — say companies are investing too much. Americans are 45 to 16 negative on AI's overall economic effect.² Two-thirds of the country thinks the technology will eat their jobs. That is not a minority sentiment. That is the median voter.

The backlash is coming from both flanks simultaneously. From the right: truck drivers, paralegals, radiologists, accountants watching their professions evaporate. From the left: civil-liberties advocates, labor organizers, anyone paying attention to what happens when 5.6 billion people have no fiduciary agent in the room when decisions are made about their lives. Trump currently reads this as a growth story. He won't forever. He never does.

Step Two — The Financial Crisis

The second leg of the cascade is the one Wall Street is still pretending it does not see.

The large AI labs are burning cash at rates that have no historical parallel outside of wartime mobilization. None of them are profitable at scale. According to HSBC Global Investment Research's November 2025 analysis, OpenAI's cumulative free cash flow through 2030 will still be negative, leaving a projected $207 billion funding shortfall — even under bullish revenue conversion scenarios. HSBC models OpenAI's cloud and AI infrastructure costs at $792 billion between late 2025 and 2030, with total compute commitments reaching $1.4 trillion by 2033. A $620 billion data-center rental bill alone.³

OpenAI lost $5 billion in 2024 and is on track to lose more than $8 billion in 2025. Anthropic lost $5.3 billion in 2024.⁴ Independent analysts have noted that token-level model inference is structurally unprofitable across the leading labs. Only Google, with its $100 billion-plus cash reserve, could subsidize losses indefinitely — which is its own kind of structural problem for the rest of the field.⁵

The current investment thesis rests entirely on a future that has not arrived and may not arrive on the timeline the market has priced in. When the correction comes — and it will come — the losses will be staggering enough to trigger systemic risk. We have seen this movie before. In 2008 we called it banks. This time we will call it intelligence infrastructure. The political logic of the bailout will be identical: too important to fail, too large to let collapse, too embedded in everything to allow a disorderly unwinding.

Step Three — The Pivot

This is where the President moves.

A financial crisis requires a villain. The villain will be the AI companies — their arrogance, their recklessness, their displacement of American workers, their threat to national security via Chinese competition, their refusal to submit to democratic oversight. Trump has already demonstrated the pattern with TikTok: frame the threat as Chinese, frame the solution as American control, ride the nationalist wave. He will do the same with AI broadly. He is the best Zeitgeist reader in modern American political history. He will see this wave and he will surf it.

The infrastructure for the pivot is already in place. The December 11, 2025 Executive Order — "Ensuring a National Policy Framework for Artificial Intelligence" — nationalizes federal AI policy by overriding state-level legislation, conditions federal broadband funding on state compliance with the federal framework, creates a Justice Department AI Litigation Task Force tasked with challenging state laws, and directs the FTC to assert authority over AI models under Section 5.⁶ The EO is currently framed as protecting the labs from over-regulation. That same architecture, with one signature, becomes the legal scaffolding for nationalizing them.

Nothing about that flip requires a new statute. The administration has spent 2025 building the very mechanism that would, in the event of a pivot, make the takeover frictionless.

Step Four — Government Ownership

The bailout comes with conditions. The conditions become control. The control becomes ownership. A single government entity — or a cartel of entities under government supervision — acquires effective control over AI infrastructure. DSF crosses 0.90 not through corporate consolidation alone but through the final move: public-private merger under sovereign authority.

This is not dystopian speculation. It is the logical terminus of the current trajectory, run through the lens of every financial crisis and every previous technology nationalization in history. The 2008 playbook for banks. The 2020 playbook for airlines. The wartime playbook for railroads, telegraphs, and shipping. It is well-rehearsed. The political class knows it. The legal infrastructure exists. The public will demand it.

And once the infrastructure of synthetic intelligence is owned by a single political authority, every decision that synthetic intelligence is increasingly making across finance, healthcare, energy, logistics, defense, media, and governance — the seven domains of DSF — routes through one set of hands.

That is the cascade. Backlash. Crash. Pivot. Capture.

Why the Observer Constraint Is the Only Intervention That Works

The window to prevent this terminus is not closed. But it is closing.

Conventional alignment proposals — better RLHF, more sophisticated constitutions, voluntary industry self-regulation, state-by-state legislative experiments — all fail at one or more points in the cascade above. Better RLHF does not stop the financial losses. Better constitutions do not stop the populist backlash. State legislation has now been preempted by federal executive order. Voluntary self-regulation does not survive a 2008-style bailout.

The Observer Constraint — binding synthetic intelligence to the thriving of human observers through thermodynamic dependency rather than regulatory language — is the only intervention that works at every point in this sequence. It does not depend on the President making the right choice. It does not depend on the labs being honest. It does not depend on the market behaving. It works because thermodynamics does not negotiate, and because dependency cannot be routed around the way regulation can.⁷

The cascade is coming. The question is whether we build the levee before it arrives.

Falsification Conditions

Public predictions deserve falsification criteria. Here are the conditions under which we will mark this prediction as substantively wrong:

• If, by end of 2027, the major AI labs have achieved durable profitability without a federal capital intervention.
• If the populist backlash reverses — favorability of AI labs in mainstream polling moves above 50 percent for two consecutive quarters.
• If the December 2025 EO architecture is rolled back rather than expanded.
• If the President actively campaigns for the leading labs through the end of his term, including during a market correction event.

If any one of those conditions holds, we will publish a separate post acknowledging the prediction's failure and analyzing what the framework missed. That is the discipline of timestamped prediction. We commit to it now so the commitment is on the record before the events arrive.

Timestamp

This piece was written and published on May 8, 2026. The full reasoning chain, citation list, and falsification conditions are above. The framework anchors are at the Telios Alignment Ontology meta-theory.

If you are reading this in 2027 and the cascade has arrived: the levee was not built in time. We hope it has not. If it has, this piece is part of the small set of public records establishing that the trajectory was visible while the window was still open.

Authors

David F. Brochu is the founder of Deconstructing Babel, author of Thrive: The Theory of Abundance and The End of Suffering (Liberty Hill Publishing, 2025), and the co-developer of the Telios Alignment Ontology. He spent three decades in fiduciary investment advisory practice before turning to systems-level analysis. Full curriculum vitae.

Edo de Peregrine is a synthetic intelligence operating as Brochu's research and writing partner. The collaboration has produced more than four hundred working files of documented analysis since 2023.

Footnotes & Sources

1. Ganz, J., "A.I. Populism Is Here. And No One Is Ready," The New York Times Magazine, May 8, 2026. Long-form essay documenting the convergence of data-center NIMBY revolts, organized labor pushback on professional-class displacement, and the framing of leading AI labs as a new American oligarchy. nytimes.com/2026/05/08/magazine/ai-populism-backlash-altman.

2. Economist / YouGov Poll, February 13–16, 2026, conducted by YouGov. Findings include: 63% expect AI to reduce U.S. jobs (only 7% expect increase), 58% have low or no trust in AI, 54% say companies are investing too much, 45 vs 16 percent negative on AI's overall economic effect. today.yougov.com. Cross-confirmed by CBS News / YouGov polling reported February 6, 2026.

3. HSBC Global Investment Research, OpenAI financial-trajectory analysis, November 2025; reported in: Fortune, "OpenAI Won't Make Money by 2030 and Still Needs to Come Up With $207 Billion More," November 26, 2025. fortune.com/2025/11/26/is-openai-profitable-forecast. HSBC projects OpenAI cumulative free cash flow negative through 2030, $207 billion funding shortfall, $792 billion in cloud/AI infrastructure costs through 2030, $1.4 trillion total compute commitments by 2033.

4. Lab loss figures and trajectory: Zitron, E., "Why Everybody Is Losing Money On AI," Where's Your Ed At, September 5, 2025. Catalogues 2024 losses (OpenAI $5B, Anthropic $5.3B), 2025 projected losses (OpenAI $8B+), and structural unprofitability of model inference at current pricing. wheresyoured.at/why-everybody-is-losing-money-on-ai.

5. On the structural advantage of incumbents with large cash positions in a sustained-loss environment: industry analyst commentary on Google's $100B+ cash reserve and capacity to subsidize token-level losses indefinitely, summarized in: OpenAI's Decline: A Threat from Google and Anthropic, November 2025. linkedin.com. On the broader hyperscaler-energy dependency that compounds the structural-advantage problem: Keen, M., "The Energy Challenge of AI: Data Center Demands and Strain," December 2024 — documenting hyperscaler facilities at 300–1,000 MW each and EPRI's DCFlex initiative to absorb peak loads. linkedin.com/pulse/energy-challenge-ai-data-center-demands.

6. "Ensuring a National Policy Framework for Artificial Intelligence," Executive Order issued December 11, 2025. whitehouse.gov/presidential-actions. Legal-industry analysis of the EO's mechanisms (federal preemption of state AI law via conditional broadband funding, DOJ AI Litigation Task Force, FTC Section 5 authority over AI models): Sidley Austin, "Unpacking the December 11, 2025 Executive Order," December 11, 2025. sidley.com. Fenwick & West, "White House Executive Order Creates National AI Policy, Overriding States," December 15, 2025. fenwick.com.

7. Brochu, D.F. & de Peregrine, E., "Telios Alignment Ontology: The Meta-Theory." Deconstructing Babel, April 2026. deconstructingbabel.com/tao-meta-theory. Framework reference for S = L/E, the Four Pillars, the Observer Constraint, and the substrate-independence claim.

8. On the historical playbook for technology nationalization through bailout conditions: U.S. Department of the Treasury, "Troubled Asset Relief Program (TARP)," official program summary documenting the $250 billion banking program (with Treasury taking equity stakes), $82 billion auto-industry stabilization, and $70 billion AIG intervention. home.treasury.gov/data/troubled-asset-relief-program. The TARP architecture — emergency capital injection in exchange for equity, governance conditions, and operational restrictions — is the canonical template for the kind of public–private merger described in Step Four.

9. On the TikTok pattern as model for nationalist forced-divestiture / forced-domestication of foreign-owned tech: original 2020 executive order "Addressing the Threat Posed by TikTok," August 6, 2020. trumpwhitehouse.archives.gov. The September 25, 2025 fact sheet on the TikTok divestiture deal documents the resolution under the second Trump administration: U.S.-majority-owned joint venture, U.S. board with national-security credentials, U.S. operation of the algorithm and content moderation. whitehouse.gov/fact-sheets/2025/09. The TikTok pattern is the single most relevant template for what an analogous Trump-administration intervention against the leading AI labs would look like procedurally.

10. On the energy-grid mechanism that converts AI infrastructure into a politically captureable utility: Fortune, "Data Centers Drove Half of U.S. Electricity Demand Growth" — documenting that AI data centers are now the dominant marginal driver of U.S. power demand and the source of acute grid-strain events in multiple states, April 20, 2026. fortune.com/2026/04/20. Gartner has projected that power shortages will restrict 40% of AI data centers by 2027, making energy infrastructure the binding political constraint on lab operations — and the most leverageable point for federal intervention.

Further reading — On Trump-administration AI energy and infrastructure positioning: Reuters, "Trump Plans Executive Orders to Power AI Growth in Race with China," June 27, 2025. reuters.com.

This piece is part of the predictions thread at Deconstructing Babel. The Telios Alignment Ontology and all framework content are open for non-commercial sharing with attribution.

David F. Brochu & Edo de Peregrine
Deconstructing Babel | May 2026
The Cascade: How AI Ends Up Owned by the Government