AI beyond tech: What Q2 earnings told investors about the widening AI trade

Key points:

• AI is becoming a broader capital-spending cycle. Data centres require heavy equipment, electricity generation, grid infrastructure, cooling, real estate and significant external financing.
• The evidence is increasingly visible in company results. GE Vernova disclosed more than $5 billion of data-centre orders in the first half, Eaton reported data-centre revenue growth of around 65%, while Digital Realty and Trane showed strong growth in leasing and commercial HVAC demand.
• Broader sector exposure does not necessarily mean broader risk diversification. These companies operate in different industries, but much of the incremental demand ultimately depends on continued spending by hyperscalers.

For much of the last few years, investing in artificial intelligence has largely meant technology: Nvidia, semiconductors, memory, networking and the hyperscalers themselves.

That definition is becoming too narrow.

An AI data centre does not run on GPUs alone. It must be constructed, connected to the grid, cooled, backed by reliable power and financed. Increasingly, those requirements are showing up in the earnings of companies that would not traditionally be considered AI stocks.

Deere and Caterpillar: Data centres require heavy equipment and construction

Deere's latest Construction & Forestry sales rose 18% year-on-year, helped partly by US infrastructure spending and AI data-centre construction.

Caterpillar showed a similar trend. Its Construction Industries sales rose 35%, while Energy & Transportation sales increased 17%. Sales of large generator sets and turbines to end users rose sharply, with management highlighting data-centre demand as an important driver.

Investor perspective: In our view, these results show that AI infrastructure spending is becoming large enough to influence traditional industrial companies. But neither Deere nor Caterpillar is a pure AI exposure: agriculture, mining, construction and the broader economic cycle remain important earnings drivers.

GE Vernova and Dominion Energy: AI growth is increasing electricity demand

The expansion of AI infrastructure is also becoming a power-generation and grid story.

GE Vernova reported more than $5 billion of data-centre orders in the first half of 2026, already more than double its data-centre orders for the whole of 2025. Overall Q2 orders rose 88% organically.

Dominion Energy provides the demand-side evidence. Its Virginia business operates in one of the world's largest data-centre markets, and Q2 adjusted operating earnings in the segment rose 22% to $670 million, with growing data-centre demand contributing to electricity-load growth.

Investor perspective: We think power availability is becoming one of the key constraints on further AI infrastructure growth. That can support demand for generation, transmission and utility investment. The counterpoint is that supplying this growth requires heavy capital expenditure, while regulation and financing costs influence how much ultimately translates into shareholder returns.

Eaton and Trane Tech: Higher computing density requires more electrical and cooling infrastructure

Once electricity reaches the data centre, it has to be distributed reliably across increasingly power-intensive servers. Those servers also produce substantial amounts of heat.

Eaton reported data-centre revenue growth of around 65% year-on-year and data-centre order growth of roughly 85% in Q2. Its equipment includes switchgear, breakers and power-distribution systems used inside data centres.

Trane Technologies' Americas Commercial HVAC bookings rose 50%, while applied-equipment bookings increased by around 130%. Total backlog reached $12.1 billion, up 70%. Trane does not disclose a standalone AI revenue figure, but data-centre cooling is an important contributor to commercial demand.

Investor perspective: The results suggest some of the less visible infrastructure around AI — power management and cooling in particular — is becoming increasingly material. The risk is that strong growth expectations may already be reflected in valuations, while order growth remains dependent on continued data-centre construction.

Digital Realty and Blackstone: Data-centre demand is supporting real estate and infrastructure assets

AI spending is also flowing into the physical real estate required to house computing capacity.

Digital Realty's Q2 revenue rose 29% year-on-year. It signed $307 million of annualised new bookings, ended the quarter with a record $1.9 billion leasing backlog, and reported 25.4% cash rental-rate growth on renewals.

Blackstone provides a private-markets example. The firm said nine of its ten largest appreciating investments in Q2 were AI-related, including exposure through its QTS data-centre platform and other digital-infrastructure assets.

Investor perspective: These results suggest strong demand for locations where both computing capacity and sufficient electricity can be secured. However, data-centre real estate is highly capital intensive and remains sensitive to financing costs, power availability and assumptions around long-term utilisation.

Goldman Sachs and Bank of America: The AI buildout requires large amounts of financing

The scale of data-centre investment also means the AI cycle is becoming increasingly relevant for banks and capital markets.

Goldman Sachs' Q2 investment-banking fees rose 55% to $3.4 billion, although the bank does not separately disclose how much came from AI-related transactions.

Bank of America has said it has helped AI-related companies raise nearly $500 billion since 2025 across debt and equity markets.

The financing opportunity extends across loans, bond issuance, equity raising, structured finance and advisory work as companies fund data centres, GPUs, power generation and supporting infrastructure.

Investor perspective: In our view, AI is evolving into a capital-markets theme as well as an equity-market theme. But the earnings linkage for banks is less direct than for equipment suppliers because profitability still depends heavily on rates, credit quality, trading activity and the broader dealmaking environment.

What earnings are telling investors

The factual evidence suggests that AI spending is already reaching much further into the economy.

In our view, this marks an important evolution of the AI investment story: from a relatively narrow technology theme towards a broader capital-expenditure cycle.

That said, broader sector exposure does not automatically mean broader risk diversification. Deere, Eaton, Digital Realty and Goldman Sachs may look very different, but part of their incremental growth increasingly traces back to the same source: hyperscalers continuing to spend heavily on AI.