“Capital creates intelligence. Intelligence attracts capital. Capital raises the price of money.”
AI was supposed to make everything cheaper. Before it does, it may make money more expensive.
The AI capital-formation cycle is becoming recursive. Investment creates infrastructure. Infrastructure increases compute capacity. Greater compute improves AI capabilities. Better capabilities raise expectations of future productivity and earnings. Those expectations lift valuations and improve access to capital. New capital finances another round of investment — and the cycle begins again at a larger scale.
This is no longer simply a technology investment boom. It is an increasingly self-reinforcing capital-formation system: Capital → Compute → Capability → Expectations → Valuations → Financing → More Capital. That third characteristic — the financing — is what changes the role of central banks.
The Five Reinforcing Loops
How investment creates conditions for more investment
-
More capital → more compute → better models → broader applications → greater expected productivity → more capital. signal: Compute begets capability, and capability begets fresh capital.
-
Higher expected AI earnings → higher valuations → looser effective equity-financing constraints → greater CapEx → stronger AI expectations → higher valuations. signal: Valuations become a financing input, not just an output.
-
Rising expected cash flows → greater borrowing capacity → more bond and private-credit financing → more infrastructure → higher expected future cash flows. The Bank of England says AI-focused companies reached an inflection point when required investment exceeded their capacity to fund it from internal cash flows, accelerating the move toward external finance during 2026. signal: The move from internal cash to external debt is where finance enters the loop.
-
More AI demand → more data centres → more electricity demand → more grids, generation and storage → more infrastructure investment → greater available AI capacity. signal: Compute scarcity converts into an energy and grid problem.
-
AI investment → GDP growth → stronger expected productivity → stronger asset prices → easier financial conditions → greater investment. The BIS notes that optimism around AI has simultaneously boosted capital expenditure and equity valuations, helping support favourable financial conditions. signal: The boom eases the very financial conditions that finance it.
The Law of Recursive Capital Formation
Law IV
Why This Cycle Is Different
Expectations driving physical investment
The Great Bifurcation
Two macroeconomic phases
R-Star Becomes Endogenous to the AI Cycle
The deepest monetary implication
Why the Role of Central Banks Must Expand
From reaction function to system function
Five New Central-Bank Functions
What the mandate now requires
I — Measure the AI Capital Cycle
Real-time measures of AI CapEx, data-centre construction, power commitments, semiconductor investment, corporate bond issuance, private credit, off-balance-sheet financing and AI-related leverage. Traditional investment statistics may arrive too slowly for a recursive cycle.
II — Estimate a Dynamic R-Star
The neutral rate cannot be treated as static. Central banks must distinguish temporary inflation from a structural upward shift in the equilibrium price of capital — the distinction that determines whether "higher for longer" is cyclical or structural.
III — Monitor the AI Credit System
The Bank of England reports rapid expansion of AI financing across public debt, private credit, leveraged and structured finance, warning shocks could propagate more widely. The Fed's May Financial Stability Report records concerns that increasingly debt-funded AI CapEx is adding leverage. Visibility is needed across the whole chain, not just regulated banks.
IV — Watch the Equity–Credit–Sovereign Loop
AI expectations raise equity valuations and household wealth; growth expectations influence sovereign debt sustainability; higher real rates raise debt-service costs. The Bank of England identifies this two-sided sovereign effect. The cycle now links Corporate Equity → Corporate Credit → Growth → Rates → Sovereign Debt.
V — Become Faster
Agentic systems process information and execute at machine speed. The Bank of England argues AI creates a double challenge: managing new risks while transforming how central banks perform their own role. A faster economy eventually requires a faster central bank — in measurement, simulation, surveillance and scenario analysis.
The New Central-Bank Dashboard
The velocity ratio
Markets — Equities and Bonds
Conversion, not participation
Follow returns, not CapEx
- Follow returns, not CapEx — the decisive metric is AI cash-flow conversion.
- Balance-sheet strength becomes strategic — self-financing capacity matters more as the cycle levers up.
- Higher rates accelerate selection — the hurdle rate separates scalable economics from speculative capacity.
- Expect concentration before diffusion — capital may compound first around scale, cash and infrastructure.
- Watch recursion — when rising investment stops producing rising expected returns, the cycle changes direction.
Trajectory over the next cut
- Stop focusing exclusively on the next rate cut — focus on the trajectory of r-star.
- AI is becoming a major source of credit demand.
- Long-duration sovereign bonds increasingly sit at the intersection of AI growth and AI capital demand.
- Credit selection becomes critical as infrastructure financing expands.
- Watch the equity-credit gap — credit may identify deterioration in AI economics before equity fully prices it.