The AI Capital Cycle Is Splitting the Market

In Brief 

• The AI buildout is pulling capital away from financial engineering and toward the physical economy needed to power compute.
• Hardware companies have outperformed as demand outstrips supply, but history shows that capital-intensive cycles eventually attract new supply.
• The market is punishing many data and workflow businesses too broadly, creating openings for active investors who can distinguish disruption risk from durable value.

Of the many factors behind Allied victory in World War II, one was found in the steel plants of Pittsburgh, the machine shops of Worcester and the foundries of Cleveland. Germany built highly advanced weapons, but America’s edge was the ability to make, ship, fix and replace equipment at overwhelming scale. Superior technology can win battles. However, superior productive capacity helps win wars.

For decades after the war, the US had both — and prosperity rose. Starting in the 1990s, globalization changed the model. Capital shifted toward intellectual property, buybacks, dividends, M&A and outsourced production rather than toward physical assets. Even the big capital cycles that followed — telecom in the 1990s, housing in the 2000s — did little to rebuild industrial capacity.

This matters now because AI may live in the cloud, but the cloud lives on the ground. Every model, query and data center depends on electricity, semiconductors, memory, copper, steel, concrete, cooling, skilled labor, permitting, logistics and capital. After decades of underinvestment and offshoring that helped push profit margins higher, the US has less spare capacity than in prior eras. You can see it in earnings, where some hardware companies are taking meaningful price, and in rates markets, where inflation remains sticky.

The result is a capital-cycle shift — away from financial engineering and toward the physical economy needed to deliver compute. That shift is already showing up in market and portfolio performance in two ways.

First, Scarcity Is Outperforming 

The market’s initial response has been straightforward: reward companies closest to the physical bottlenecks.

Hardware, compute, memory, power equipment and other capital-intensive beneficiaries have outperformed because demand exceeds supply. In many areas, that gap has allowed meaningful price increases and stronger profits. Customers need scarce inputs and have few near-term substitutes, giving suppliers real pricing power.

But this is still a hardware capital cycle. High prices attract capital. Capital builds supply. Supply changes pricing power, returns and valuations. While today’s winners may keep growing, investors should be careful about assuming scarcity lasts forever. In capital-intensive industries — and history is clear on this — exceptional profitability typically plants the seeds of margin normalization.

Second, the Market Is Using a Blunt Instrument on Potential Disruptees 

Companies tied to software, data, information services and mission-critical workflows have been punished — in some cases even when they’ve met earnings expectations and shown little sign of revenue decline.

The market has treated AI disruption as if the verdict were simple: software bad, compute good, disruption everywhere. We think that’s too blunt.

AI will likely commoditize some generic software and workflows that are repetitive, low-trust or easy to copy. But not all software, data or workflow businesses face the same risk. Those whose value comes mainly from process friction or basic information retrieval may be vulnerable. Those that manage proprietary data, embedded workflows, regulated systems, audit trails, compliance processes or mission-critical records may be different. In those cases, value isn’t just software code — it’s the trust, governance, domain knowledge, history and accountability built into the workflow. AI can break down weak moats, but it can also reinforce strong ones.

Why Trusted Answers May Become More Valuable 

This is where Jevons’ Paradox comes in. When technology cuts the cost of something, usage often rises. Cheaper compute didn’t produce less computing. Cheaper storage didn’t produce less data. Lower-cost AI-generated analysis may drive more demand for answers, scenarios, workflows and decision support. But as generic answers get cheaper, trusted answers may become worth more.

In an AI-enabled world, the bottleneck may not be producing an answer. The bottleneck may be knowing whether the answer is accurate, compliant, auditable and fit for purpose. That should matter for companies with proprietary datasets, systems of record, embedded customer relationships and domain-specific workflows that a large language model can’t easily recreate. It’s especially relevant for business services, database companies and other information-rich firms whose value rests on trusted data and workflow integration. Some are being priced as if AI will erode their economics. In some cases, that may be right. In others, AI may deepen customer dependence and make their data more strategically important.

Conclusion 

Passive owns both sides of this transition. It holds beneficiaries of the AI capital cycle, but also companies whose margins, moats or valuations may prove vulnerable. It doesn’t distinguish between firms that can fund, absorb and earn attractive returns on this new capital cycle and those that can’t. It doesn’t distinguish between software that’s easily replicated and mission-critical workflow businesses whose data, auditability and trust grow more valuable as AI adoption rises.

The AI capital cycle is already splitting the market. Hardware scarcity is real, but capital cycles are finite. AI disruption is real, but it isn’t universal. The task is to separate temporary scarcity from durable economics, and true disruption from market overreaction. The market is using a hammer where a scalpel is needed. In that environment, active management should matter more. The opportunity isn’t simply to own AI winners and avoid AI losers — it’s to identify which businesses have economic power that AI will erode, and which have economic power that AI may make more important. 

Keep in mind that all investments carry a certain amount of risk, including the possible loss of the principal amount invested. 

The views expressed are those of the author(s) and are subject to change at any time. These views are for informational purposes only and should not be relied upon as a recommendation to purchase any security or as a solicitation or investment advice. No forecasts can be guaranteed. Past performance is no guarantee of future results.