A Pronit Investigation · Edition 01 · August 2026

The AI boom is not arriving everywhere at once—or on equal terms.

We traced the money, read the filings, and mapped the circular deals, the debt, the energy, the jobs, the enforcement actions, and the geopolitics. Then we let the evidence argue with itself. Every claim is sourced. Every number has a provenance. Every controversy carries its rebuttal.

Begin the investigation Open the verdict lab
$0TInfrastructure by 2030
$0THidden AI debt
0US jobs cut, AI-cited
+0%Data-centre power, 2025

Global evidence field

Hover a region to surface its latest indicators. The map redraws by domain—capital, labour, infrastructure, governance.

What this is

Not a blog post. A structured investigation into the largest capital-allocation decision in human history—separating verified facts from company claims, analyst estimates from allegations, and opinions from unresolved questions.

01Is this a bubble?→ 02The circular economy→ 03Debt & infrastructure→ 04Chips & scarcity→ 05The IPO machine→ 06Does AI actually work?→ 07Labour: boon or bane?→ 08Power, water, carbon→ 09The global race→ 10Power & control→ 11The strongest cases→ 12Verdict laboratory→ 13Source explorer→ 14Methodology & limits→

Half of the cloud revenue backlog reported by four of the largest companies on earth comes from two startups funded by those same four companies.

§ The Information / The Algorithmic Bridge, May 2026 — see Circular Economy

The timeline

From a research demo to a $5 trillion bet. Key moments in the AI boom, layered by capital, technology, and public response.

Nov 2022ChatGPT launches
Mar 2023GPT-4 · banking crisis
Nov 2023OpenAI board crisis
Jun 2024NVIDIA hits $3T
Jan 2025DeepSeek R1 shock
Sep 2025NVIDIA $100B → OpenAI
Nov 2025Circular deals exposed
Dec 2025$121B debt year
Jul 2026$1.65T hidden debt

Follow the money

The circular flow of capital between AI's biggest players. Investments become revenue. Revenue justifies more investment. Drag the nodes; hover an edge for the deal.

Evidence standards

How we decide what counts, and how confidently we say it.

Source hierarchy

  1. Court judgments & official admissions
  2. SEC filings & regulator findings
  3. Company earnings transcripts
  4. Multiple independent investigations
  5. Analyst estimates (labelled)
  6. Opinion & commentary (labelled)

Confidence labels

  • Substantiated final judicial / official finding
  • Partly core established, scale unclear
  • Unresolved credible basis, no determination
  • Claim unverified assertion
  • Opinion analysis or interpretation

What we don't do

  • Call allegations "proven scams" without findings
  • Treat spending as stolen money
  • Predict the future as fact
  • Ignore counterarguments
  • Hide our limitations