The Velocity Edge
In Focus · 08:00

Central Banks and the Great Split

Artificial intelligence is moving from the margins of central-bank research to the center of monetary policy, creating one interest rate across two capital regimes — the Great Capital Divide.

Francesco De Leo Kaufmann August 19, 2026 8 min read Macro
9%share of euro-area corporate investment expected to go to AI
0.6pprise in expected AI investment per 1pt of perceived competitor adoption
$100Bhyperscaler gross bond issuance in 2025
€440Beuro-area household exposure to major U.S. tech equities

“What kind of economy is this interest rate actually acting upon?”

What happens when central banks must conduct one monetary policy across an economy increasingly separating into two different capital, productivity and inflation regimes? The answer is beginning to emerge.

Over the past several days, the Federal Reserve, the ECB and other major institutions have begun treating AI less as a distant productivity possibility and increasingly as a current macroeconomic force — one capable of simultaneously affecting investment, electricity demand, productivity, employment, asset valuations, credit formation and potentially the neutral interest rate.

This matters because the AI Supercycle may be creating something conventional monetary frameworks were not designed to manage: one interest rate, two capital regimes. The underlying structure is the Great Split. Its financial-market manifestation is becoming the Great Capital Divide.

01

The AI Investment Paradox

Two phases, one economy

02

The Signals

Where AI is entering the reaction function

  1. The July 28–29 FOMC meeting shows participants expecting AI investment eventually to raise productivity and potential output, while emphasizing uncertainty over timing and magnitude. More significantly, the discussion reached financial stability: AI infrastructure financing increasingly interacts with borrowing, nonbank finance and parts of the banking system.

    As long as AI investment is funded through the free cash flow of enormously profitable technology companies, systemic leverage stays contained. Once financing expands into corporate bonds, private credit, project finance and banking exposure, AI moves closer to the traditional domain of central-bank financial stability. signal: AI is no longer Technology → Productivity, but Technology → Capex → Credit → Financial Conditions → Monetary Policy.

  2. Christine Lagarde's August 19 intervention noted euro-area companies expect to direct roughly 9% of total investment toward AI. The more important question is what happens after: Europe must convert AI investment into diffusion, industrial AI and productivity. If that conversion fails, Europe absorbs the costs of Phase I without capturing the dividend of Phase II.

    ECB evidence reveals a self-reinforcing mechanism: a one-point rise in perceived domestic competitors investing in AI is associated with roughly a 0.6-point rise in a company's expected AI investment. But those effects remain heavily constrained by national borders. signal: Europe's fragmentation reduces the propagation velocity of AI.

  3. The first phase was visible through equities — semiconductors, hyperscalers, memory, networking, utilities, data centers, robotics. The next phase runs through balance sheets. Hyperscaler gross bond issuance already exceeded $100 billion in 2025, with external financing expanding across corporate debt, private markets and project finance.

    Equity absorbs uncertainty through valuation; debt transforms uncertainty into obligations. Central banks could confront two opposite risks at once: too much capital chasing returns above the Investability Frontier, and too little capital available below it — financial excess and capital exclusion at the same time. signal: Equity absorbs uncertainty through valuation. Debt transforms uncertainty into obligations.

  4. Euro-area households hold roughly €440 billion of exposure to major U.S. technology equities, much of it through funds and ETFs, with insurers and pension funds also exposed. An AI-related correction would not stay a Silicon Valley event — it could propagate from U.S. valuations through European funds to household wealth, redemptions, asset sales, financial conditions and European demand. signal: The AI financial system increasingly runs through nonbanks — funds, ETFs, private credit, insurers, pensions.

  5. A central bank sets one interest rate, but that rate operates across companies with radically different returns on capital. A 4% policy rate can be deeply restrictive for a legacy company earning a 6% return, yet barely constrain an AI infrastructure investment expected to generate dramatically higher returns. Monetary transmission becomes asymmetric — and, paradoxically, tightening can sometimes widen the divide. signal: The investable world compounds capital inward; the world below the frontier bleeds it out.

03

The Number Central Banks May Be Misreading: r*

The neutral rate under an AI boom

04

The Employment Paradox

Slack that isn't slack

05

Five Hidden Patterns

The map beneath the map

AI Is A Monetary Variable Before A Target

Central banks do not need an AI mandate, but they cannot understand inflation, investment or financial stability without understanding AI.

From Valuations Into Balance Sheets

The Supercycle is moving from equity into credit — which makes credit increasingly important.

Electricity Enters The Dashboard

Not because central banks should manage energy, but because energy scarcity increasingly determines whether AI capex produces inflation or productivity.

Employment Is Harder To Interpret

Weak hiring can coexist with strong productivity and output.

The Divide Can Become A Sovereign Divide

Technological underperformance can become capital outflow, weaker potential growth and ultimately higher sovereign risk premia.

06

The New Central-Bank Dashboard

Signals for the Next Economy

07

The Governance Question

Specialization × synchronization

08

The European Consequence

On the fault line

Five Consequential Takeaways

Who must do what

  • For central banks: the competitive advantage of the next generation of institutions will be regime recognition — identifying what kind of inflation, unemployment and capital formation they are actually observing.
  • For investors: follow credit, electricity and productivity dispersion as closely as AI equities; the Supercycle is broadening into the entire capital structure.
  • For CEOs and boards: AI adoption is a cost-of-capital issue, not an IT project — falling behind can raise WACC and reduce strategic optionality.
  • For EU policymakers: Capital Markets Union, electricity infrastructure and the Single Market are increasingly components of Europe's AI — and indirectly monetary — architecture.
  • For the ECB: the greatest future risk may be operating one interest rate across increasingly divergent Investability Regimes inside the same monetary union.

Francesco De Leo Kaufmann · The Velocity Edge

More from The Velocity Edge

Jump straight into another note — no need to head back.

Keep reading

More field notes, from dawn to close.

Day signals and night briefings, published like a journal for board-level AI decisions.

Back to The Velocity Edge