When the Cost of Intelligence Becomes the Price of Money

“The AI Supercycle is beginning to raise the price of capital before it has delivered its full productivity dividend.”

The AI Supercycle has crossed an important monetary threshold. Artificial intelligence is no longer simply a prospective source of productivity growth: its investment requirements are now explicitly influencing central-bank inflation assessments, while regulators warn that concentrated capital formation could weaken the industries expected to benefit from the technology.

Three developments make November 3 consequential. The Reserve Bank of Australia has raised rates to 4.85%, explicitly identifying AI investment as a contributor to rising technology prices. A senior Chinese regulator has warned that excessive concentration of AI investment is increasing leverage and crowding out traditional industries. And a Bundesbank speech supplies the European counterpart: adoption is spreading rapidly, but the capital to convert it into productivity remains insufficiently mobilised.

The Great Split is becoming a contest between the inflationary cost of building intelligence and the economic capacity to convert intelligence into sustainable productivity.

The RBA's Monetary Policy Board raised the cash rate by 25 basis points to 4.85%, with six members supporting the increase and three favouring unchanged rates. The statement identifies several sources of persistent inflation: elevated energy prices, tighter-than-expected domestic capacity, rising business costs and strong demand associated with AI-related investment.

Importantly, the RBA explicitly recognises that AI investment is increasing prices for technology-related goods, and expects particularly strong data-centre investment to support business spending even as higher rates weaken household consumption and housing. Inflation is projected to remain elevated through much of 2027.

The paradox is becoming clearer. AI capital formation raises demand for semiconductors, electricity and infrastructure, lifting input costs and adding inflationary pressure. The offsetting mechanism — industrial-AI deployment driving organisational transformation, higher capacity and lower unit costs — operates on a different timetable. The first is increasingly observable; the second remains conditional. RBA business intelligence confirms productivity results are still uneven, and most contacts do not expect AI to change employment materially over the next year.

The most intellectually consequential element is the RBA's research box, Developments in the Neutral Rate of Interest. It finds that estimates of Australia's neutral rate have increased since the pandemic, explicitly identifying stronger expected productivity and AI-related investment as potential contributors.

The mechanism is straightforward: higher expected productivity raises anticipated returns, encouraging investment; greater investment demand relative to savings can lift the equilibrium real rate. But the RBA draws a critical distinction — a structural rise in the neutral rate does not capture the additional restraint required when AI investment creates immediate demand pressures. If AI raises near-term investment substantially, policymakers may need to increase rates by more than the estimated rise in long-run neutral rates. The analysis provides no precise estimate of how many basis points AI has added.

Speaking on November 3 at the HKMA's Global Financial Leaders' Summit, Xiao Yuanqi of China's National Financial Regulatory Administration warned that global AI investment is becoming excessively concentrated. His argument identifies two mechanisms: the extraordinary flow of capital into leading technology companies can increase leverage and encourage investment beyond sustainable levels; and concentrated investment may raise borrowing costs for traditional industries, weakening their capacity to expand production.

These are supervisory concerns from a senior regulator — not a new PBoC decision, nor proof that AI investment has already caused economy-wide crowding-out. Nevertheless the implications are substantial. If concentrated financing raises capital costs sufficiently, the financial system could undermine the diffusion of the very technology it is financing, just as traditional businesses need to invest in automation and workforce transformation.

The Great Split would therefore extend beyond technology leadership into a structural divergence between enterprises able to attract abundant capital and those whose transformation is constrained by the financing demands of the leaders. For central banks, this would complicate transmission: higher rates might reduce demand broadly while leaving the largest, most cash-rich AI investors comparatively insulated. That asymmetry requires monitoring, not a presumption that monetary policy has become ineffective.

Sabine Mauderer, First Deputy Governor of the Deutsche Bundesbank, delivered the European counterpart in an October 29 speech published by the BIS on November 2. Bundesbank survey findings show 68% of German businesses already use generative AI or intend to by end-2026 — but only 7% use it at larger scale. Technological accessibility is advancing far faster than organisational transformation.

Mauderer also identifies a financing imbalance. Euro-area households hold more than €36 trillion in financial assets, yet European companies struggle to access the growth capital required to scale innovation. Outstanding euro-denominated bonds from major American technology companies have reached roughly €40 billion. The Bundesbank does not conclude these investments necessarily crowd out domestic business; its central message is the need to mobilise Europe's substantial savings more effectively.

The next European competitive disadvantage may arise less from insufficient access to AI than from insufficient access to patient, productive risk capital. This is particularly important for industrial SMEs, whose investments in organisational change, software and workforce capabilities often lack the tangible collateral preferred by conventional bank lending. Without more equity and growth financing, adoption may remain superficial.

Today's developments reinforce three structural forces central banks must weigh together. Investment-driven inflation is becoming measurable through technology prices, energy requirements and specialised equipment demand. Productivity-driven disinflation remains possible but depends on deployment and complementary investment. Financial crowding-out could emerge if AI infrastructure absorbs capital on a scale that restricts investment elsewhere.

Restrictive policy can contain excessive demand, but monetary tightening cannot resolve shortages of electricity infrastructure, insufficient risk capital or the organisational barriers to productive adoption — nor should central banks allocate credit preferentially among sectors. For Europe, monetary stability and productive-capital formation must be treated as complementary objectives pursued through different instruments.

Five Takeaways for Investors and Policymakers

Where capital productivity decides the outcome

• Cost of capital is becoming endogenous to the Supercycle. Reassess duration-sensitive valuations, refinancing assumptions and discount rates under scenarios where stronger AI investment lifts equilibrium rates.
• Industrial-AI deflation cannot yet be assumed. The RBA reinforces the gap between rising technology expenditure and sustainable reductions in unit production costs.
• Capital concentration deserves greater scrutiny. Monitor relative borrowing costs and investment trends across AI-intensive and traditional sectors.
• Europe's priority is productive-capital mobilisation. Deeper equity markets, stronger venture and growth financing, and institutional structures supporting the move from experimentation to deployment.
• Energy infrastructure becomes a financial-stability variable. Data-centre investment depends on timely grid access; delays raise costs and refinancing risk, so coordinate grid investment, permitting and financing.

November 3 marks a significant step in the global central-bank response to AI. Australia has incorporated AI-related price pressures into a policy decision and acknowledged they may alter equilibrium rates. China's regulator warns that capital concentration can impose costs on the broader productive economy. The Bundesbank highlights Europe's inability to mobilise its financial resources as effectively as its industrial transformation requires.

Together these suggest the AI Supercycle is entering a more demanding phase. The decisive question is shifting from how much capital AI can attract to how much sustainable value that capital can produce. For investors, the relationship between returns on invested capital, financing costs and realised cash flows becomes the essential valuation test. The economic winners will not necessarily be those investing the most — they will be those capable of converting investment into productivity while preserving financial resilience.

The next Great Split will be determined by who can finance the transition — and who can generate returns sufficient to justify its rising cost of capital.