“Technology creates possibility. Capital creates capacity. Political permission determines what can be built.”
The AI Supercycle is not losing its technological momentum. But the November 3 U.S. midterm elections have changed the financial conditions under which its next generation of infrastructure may be built. For the first time in this cycle, the political legitimacy of large-scale AI infrastructure has become a major market variable.
Democrats have secured control of both chambers of Congress, while candidates campaigning on electricity affordability and data-centre restrictions won important state-level contests. Demand for accelerated computing remains powerful, but the ability to turn that demand into approved, financed and commercially productive infrastructure is becoming increasingly selective. The next rotation could therefore occur within the AI value chain, rather than away from it.
The Big Take
From technological capacity to economic legitimacy
The midterms transformed the AI debate from technological capacity into economic legitimacy and investment discipline. Computing scarcity is broadening—electricity, permitted land, construction certainty and public acceptance now matter alongside semiconductors.
Financial markets may distinguish more sharply between operating infrastructure and speculative development pipelines: a megawatt of commissioned capacity has different economics from a megawatt announced for future construction. The next cycle may not reward the largest AI investment commitments. It may reward the most credible conversion of commitments into cash-generating capacity.
Five Signals Shaping the Next Economy
What the results change
-
The AI Supercycle has entered the economics of political permission. The result cancels no existing data centres and creates no federal ban, but state permitting and utility regulation can lengthen commissioning without new legislation. The transmission runs from political opposition to approval uncertainty to deferred revenue and lower returns. signal: Political permission is becoming a scarce complement to computing capacity.
-
Legrand raised 2026 guidance; data-centre work is now ~one-third of revenue and energy-saving products ~40% of data-centre sales. Skanska's orders reached SEK 68B (from SEK 39.9B), backlog a record SEK 328.3B, construction margin up from 4.2% to 5.0%. Suppliers are paid during construction; owners wait years to earn their return. signal: Industrial demand is stronger than some developer valuations suggest.
-
Microsoft will not exercise an option to more than double its Nscale Texas agreement, preserving optionality; the existing contract stands. Google is providing credit support for TeraWulf's ~$19B Kentucky lease, delivering ~401 MW of critical IT load for Anthropic. Financing credibility, not asset value alone, now shapes which projects proceed. signal: Financial credibility is becoming a productive resource.
-
Brent fell sharply, trading around $96/bbl on November 5. Asian markets diverged—KOSPI −5.3%, with Japan and Taiwan also down—while European equities advanced. Roughly 600,000 Nvidia H200 chips await China import approval, deepening U.S.–China fragmentation. President Trump retains veto power, limiting assumptions that every proposed change becomes law. signal: Permitting uncertainty and cheaper energy pull project economics in opposite directions.
-
The STOXX Europe 600 rose 1.8% on November 4, approaching a record. Yet the Bank of England raised its policy rate to 4.00%, showing falling oil does not immediately reverse prior inflation. Europe's opportunity is to pair infrastructure demand with industrial AI adoption—capturing productivity rather than hyperscale ownership alone. signal: Higher sustainable cash per unit of capital matters more than more AI capex.
Three Scenarios for the Next Six Months
FDK central assessment: Selective Capital Concentration
Disciplined Acceleration
Political scrutiny improves project selection while AI demand stays robust: stronger project economics, better returns on operational assets, improving credit quality.
Selective Capital Concentration
Financing and permitting advantages favour established developers and hyperscalers: greater valuation dispersion, consolidation and concentration of productive infrastructure.
The Infrastructure Squeeze
Regulatory delays, energy constraints and financing risk reinforce one another: project postponements, rising credit premiums and weaker infrastructure investment.
Reprice the new constraints
- Reprice political permission as a financial variable, alongside electricity and financing costs.
- Distinguish AI demand from AI infrastructure profitability—strong orders do not guarantee owner returns.
- Treat financing optionality as a competitive advantage: invest when returns justify it, withhold when they do not.
- Anticipate value migration toward scarce complements—powered land, grid access, efficiency, credible contracts, execution.
- Prepare for capital concentration without assuming it is inevitable; better policy and efficiency can broaden access.
The next trillion dollars will not simply determine how much intelligence the world produces—it will determine where it can be deployed, who controls the productive assets and who captures their returns. The future will belong to those who can turn investment into productive assets before the financial and political cost of building them overwhelms the returns.