The Blue•Print — home

Investment boom or historic overspend?

The free enterprise system has triggered the most massive overspend on a technology in human history.

John A Brown’s starting hypothesis. This is a question to test, not the report’s conclusion.

Scale: a scenario, not a loss

Brookings’ summary of Stijn Van Nieuwerburgh’s research describes US$10.3 trillion of projected US investment during 2025–2032, averaging 3.63% of US GDP annually. Buildings, power, networking, chips and other equipment are included. The study compares that scenario with earlier US infrastructure waves. It is neither audited expenditure nor a quantified overspend. [C6]

Demand: evidence with a narrow meaning

CBRE reports 1.4% vacancy across its surveyed North American primary markets in the first half of 2026. That supports a finding of tight available capacity there. It does not measure worldwide utilisation, GPU workload intensity, AI profitability or investor returns. [C4]

Cancellations: commitments that may never be spent

Heatmap reports at least US$260 billion of cancelled announced investments in 2026 to its October publication, following local opposition. The underlying project survey has not been independently audited here. Preventing a proposal can avoid future expenditure; cancellation values are not automatically capital destroyed. [C5]

Private enterprise and public decisions

This review does not quantify the public/private funding split or subsidies. A causal claim about free enterprise requires separate evidence about financing, government support, planning and energy infrastructure. Unreferenced tax-policy assertions from earlier drafts have been removed.

Assessment

The strongest historical overspending claim is not established. Retained evidence supports large planned investment, tight capacity in particular markets and substantial cancelled announcements. It does not measure the whole buildout’s long-term productive value or losses.

Evidence that would change the assessment

Help keep this research open

Voluntary contributions help fund research, source access, hosting and updates, so our published findings and evidence tables remain freely available to the public.

Contributing does not buy a particular conclusion, editorial control or priority. We will report how research contributions are used.

Support open research through PayPal — opens PayPal in a separate tab.

Contributed and have a research suggestion?

Use our contact form to share your proposed topic, questions, countries or sources. If you would like us to associate your suggestion with a contribution, include your PayPal transaction ID or receipt reference, payment date, amount and currency. Providing a payment reference is optional. These details remain private.

Suggestions help guide future research. Contributions do not guarantee that a topic will be undertaken, buy priority or determine our findings. You are also welcome to suggest research without contributing.

Contributions and AI assistance

Research compiled by Grok under instruction from John A Brown, The Blue•Print Research Unit. Editorial integration and source checks used ChatGPT assistance at John’s direction.

The original question and scope came from John. AI assistance produced drafts, tables and website code and helped check cited sources. John approved publication of this edition. Checking cited material is not independent verification of underlying claims.

Publishing Contributions Specification — Draft 0.1 · Publishing Contributions Registry

Published contribution statement — revision 1 · Download contribution statement as XML

Registry reference: PCR-20261031-7A5A82E2E9157692DF3F3DDA. This is a registrant declaration, not independent verification or endorsement of the research.