Can AI Productivity Outrun the U.S. Debt Spiral?
Sources 17
Primary documents, filings, and reporting cited in this episode. Click any source to highlight where it comes up in the transcript, or press ▶ to jump to that moment.
Modelling AI’s macroeconomic impact ↗
Oxford Economics baseline: generative AI makes the US economy about 3.5% more productive over the next decade, at the cautious end of published estimates ranging from ~1% to more than 12%.
Scenarios for our Economic Future ↗
Anthropic Economics team's interactive scenarios for US GDP in 2030: modest +1.6%, substantial +8.3%, extreme +32.4% (extreme likely requires recursively self-improving AI); labor share 59.4%, 56.1% and 45.2%.
Economic Scenarios for Transformative AI (technical report) ↗
Technical report behind the scenarios: in the extreme scenario the labor share of income falls from 60% today to 45% in 2030.
Global Evidence on Business Use of AI ↗
Survey of nearly 6,000 executives (US, UK, Germany, Australia): they forecast AI will raise labor productivity 1.4% over the next three years while cutting employment 0.7%, a net output gain of ~0.8%.
AI, productivity, and work: Evidence from US firms ↗
Survey of 734 US financial executives: AI-attributed labour productivity growth ~0.6% in 2025, expected ~1.8% in 2026, largest in high-skill services and finance.
Can AI restore fiscal sustainability in the US? ↗
Brookings working paper (July 2026): a once-in-a-generation productivity shock could cut deficits ~5 points of GDP by 2036, but five AI-specific forces could claw back more than half.
No, AI Will Not Fix Soaring Budget Deficits ↗
Riedl's own analysis: a sustained 1-point productivity boost raises revenue $834bn a year by 2036, about a fifth of the projected $4.4T deficit.
Gartner Says Over $3.5 Trillion of AI Infrastructure Spending Could Be Exposed to Impairment by the End of 2030 ↗
Gartner press release, 14 October 2026: data-center owners need $8.5T cumulative revenue through 2030; attributable AI revenue no more than $3.6T.
The AI boom is nearly $5 trillion short of paying for itself, Gartner says ↗
Fortune write-up of the Gartner figures, framing the gap as nearly $5 trillion.
The AI boom hasn’t revived US investment (yet) ↗
MGI: in the three years after ChatGPT, US data center investment tripled while total productive investment held flat as a share of GDP.
Final Monthly Treasury Statement, Fiscal Year 2026 (through September 30, 2026) ↗
Official FY2026 results: $1.992T deficit; corporation income taxes $381.4bn vs $452.1bn in FY2025 (-16%) out of $5,403bn total receipts (~7%).
Does the US Tax Code Favor Automation? (Brookings Papers on Economic Activity, Spring 2020) ↗
Effective US labor taxes 25.5–33.5%; effective taxes on software and equipment ~10% in the 2010s and 5% after the 2017 reforms.
How potential AI futures would play out in the current tax system ↗
Rapid AI growth scenario: federal revenues up to $246bn (3.7%) above CBO’s 2026 baseline by 2030; nearly twice as large if capital and labor shares were held fixed.
The Role and Design of Net Wealth Taxes in the OECD — Chapter 1: Overview ↗
OECD (2018): countries levying individual net wealth taxes fell from 12 in 1990 to 4 in 2017 (France, Norway, Spain, Switzerland); 2016 revenue ranged from 0.2% of GDP (Spain) to 1.0% (Switzerland).
Transformation de l'impôt de solidarité sur la fortune (ISF) en impôt sur la fortune immobilière (IFI) et création du PFU : un premier bilan ↗
Senate finance committee review (2019) of the 2017 replacement of the ISF by the IFI; ISF net yield €4.2bn in 2017; net departures of ISF payers averaged 597 a year in 2012–2016.
S.4246 — Ultra-Millionaire Tax Act of 2026 (introduced text) ↗
Bill text: 2% on net worth between $50 million and $1 billion, 3% above $1 billion (6% if qualifying health legislation is in effect).
Proposition 40 — Imposes One-Time Tax on Certain Taxpayers (Quick Guide) ↗
November 3, 2026 ballot: one-time 5% tax on taxpayers with assets over $1 billion; revenue primarily for health care.