AI Spending Is Starting to Reshape the Economy Around It

The AI trade is no longer confined to semiconductors and cloud companies. The scale of spending is now large enough to influence financing markets, industrial demand, power infrastructure, and even sector leadership across the broader economy.

Hyperscaler capital expenditures are now projected to approach roughly $800B in 2026, with spending continuing to rise into 2027 as companies race to secure compute capacity, energy access, and data center expansion. The market is beginning to treat AI less like a software cycle and more like a multi-year industrial buildout.

That shift is becoming visible across earnings and capital flows.

Infrastructure-linked companies continue seeing accelerating demand tied to electrical systems, cooling, transmission equipment, and power management. Eaton, Vertiv, Quanta Services, Emcor, and Constellation Energy have all benefited from growing investor focus on the physical constraints tied to AI deployment rather than the applications alone.

At the same time, financing conditions are becoming increasingly important.

AI companies are increasingly relying on more borrowing and outside financing to fund the massive cost of building data centers, buying chips, and expanding power capacity. As spending rises, investors are paying closer attention to which companies can afford these projects without putting too much pressure on their balance sheets.

That dynamic is beginning to create a split inside the AI trade itself.

Companies showing clear infrastructure demand, backlog growth, or visible monetization continue outperforming, while areas viewed as crowded, overfinanced, or dependent on continued multiple expansion are seeing more volatility beneath the surface.

The result is a market increasingly rotating away from “AI concept” exposure and toward the companies supplying the power, networking, cooling, electrical equipment, and industrial capacity required to support the buildout itself.

AI Is Becoming an Infrastructure Cycle

This earnings season reinforced that some of the strongest growth tied to AI is no longer coming only from software and semiconductor companies. Increasingly, companies tied to power infrastructure, cooling systems, networking equipment, storage, and data center construction are reporting accelerating demand as the physical buildout behind AI continues expanding.

Vertiv, which builds cooling and power systems used inside AI data centers, reported 30% revenue growth and a 64% increase in adjusted operating profit as demand for higher-density AI infrastructure accelerated. The company also raised guidance and pointed to growing demand from hyperscalers trying to deploy AI capacity faster.

Eaton, which supplies electrical equipment like switchgear, transformers, and power management systems, reported record backlog growth and continued expansion tied to data center and grid demand. Its Electrical Americas orders rose sharply as utilities and AI developers continued upgrading power infrastructure.

Quanta Services, one of the country’s largest transmission and grid contractors, reported record backlog of $48.5B as utilities and power providers continue expanding transmission networks and electrical capacity to handle rising energy demand.

EMCOR, which handles electrical and mechanical construction projects for data centers and mission-critical facilities, raised guidance after nearly 20% revenue growth and record project obligations. Management highlighted continued strength in network infrastructure and large-scale construction demand.

The networking side of AI infrastructure is also accelerating.

Lumentum, which supplies optical networking components and photonics systems that move data between AI servers and data centers, reported 90% revenue growth YoY and significant margin expansion. Management pointed specifically to rising demand tied to cloud computing, optical networking, and AI infrastructure buildouts.

The impact is now spreading into utilities and power generation as well.

Constellation Energy highlighted growing electricity demand tied to data centers and AI facilities, while Vistra announced additional long-term power agreements connected to rising data center demand and noted strong load growth across its markets.

Even storage companies are benefiting. Western Digital reported 45% revenue growth and expanding margins as AI workloads continue generating enormous amounts of data that must be stored long-term. Management said nearly every major AI workload is increasing demand for persistent storage infrastructure.

The broader point is that the AI trade is evolving beyond software and semiconductors into a much larger physical infrastructure buildout.

• More electricity — AI data centers require enormous amounts of power to train and run large models.
• More cooling — high-performance AI servers generate extreme heat and need advanced cooling systems.
• Larger transmission networks — utilities must expand grid infrastructure to deliver power to new AI facilities.
• Optical networking equipment — AI workloads require ultra-fast data transfer between servers and data centers.
• Data storage — AI models generate massive amounts of data that must be stored and accessed efficiently.
• Industrial construction — companies are rapidly building new data centers and expanding existing facilities.
• Power management systems — AI infrastructure needs advanced electrical equipment to distribute and stabilize power usage.

Strong Earnings Are Becoming Increasingly Concentrated

The broader earnings season has been exceptionally strong, with S&P 500 earnings growth now tracking above 27% YoY and roughly 84% of companies beating EPS expectations.

But a growing share of those upward earnings revisions continues coming from companies tied directly to AI infrastructure spending.

AI infrastructure-related companies — including semiconductors, data centers, networking, and power infrastructure — have seen 2026 earnings estimates revised higher by roughly 55% since late 2024. Meanwhile, earnings estimates for the S&P 500 excluding those AI infrastructure names have actually moved slightly lower over that same period.

The same trend is showing up in earnings growth itself. The Magnificent Seven are now expected to post roughly 61% earnings growth this quarter versus about 16% for the rest of the index.

That helps explain why leadership continues clustering around:

• semiconductors
• hyperscalers
• networking
• power infrastructure
• utilities
• data center construction

rather than the broader market.

The important distinction is that this is no longer just speculative enthusiasm around AI. Many of these companies are now reporting real revenue growth, expanding margins, rising backlog, and improving cash flow tied directly to the ongoing AI infrastructure buildout.

Image Below: That demand is also beginning to expose a growing constraint: power availability. Morgan Stanley estimates U.S. data centers could face a potential 55 gigawatt power shortfall through 2028 as AI deployment accelerates faster than grid expansion and generation capacity. The chart highlights why investors have increasingly rotated toward utilities, natural gas infrastructure, nuclear power, electrical equipment, and grid expansion names like VST, CEG, ETN, and PWR as the market focuses on the physical limits behind the AI buildout.

AI Electricity Demand Is Now Showing Up Across Nuclear Earnings

The connection between AI infrastructure and nuclear power is no longer just a long-term theme — companies across the nuclear supply chain are now beginning to reference rising electricity demand and domestic fuel expansion directly in earnings results.

Cameco (CCJ), one of the world’s largest uranium producers, reported strong uranium earnings growth and highlighted increasing global demand for secure and reliable nuclear fuel supply as electricity demand rises and countries continue prioritizing long-term energy security. The company also pointed to growing interest in Westinghouse reactor technology as utilities and governments expand nuclear development plans.

Centrus Energy (LEU), which focuses on uranium enrichment and HALEU fuel production for next-generation reactors, raised full-year revenue guidance while accelerating expansion projects in Tennessee and Ohio. Management highlighted efforts to rapidly scale domestic enrichment capacity, including new partnerships with Fluor and Palantir, as the U.S. pushes to rebuild nuclear fuel infrastructure.

Bloom Energy (BE), which provides on-site fuel cell power systems for data centers and large industrial facilities, reported 130% revenue growth and raised full-year guidance as demand accelerated for reliable electricity outside the traditional grid. The company has increasingly benefited from AI-driven power demand, including Oracle’s agreement to purchase up to 2.8 gigawatts of Bloom fuel-cell power for AI-focused data centers — one of the clearest signs yet that hyperscalers are directly securing dedicated power infrastructure to support the next phase of AI expansion. Management described Bloom as becoming a “go-to choice” for digital power infrastructure as electricity demand tied to AI continues rising.

The broader takeaway is that the AI buildout is increasingly flowing beyond chips and software into the energy systems required to support it. That continues driving investor focus toward uranium, enrichment, reactor infrastructure, and alternative power generation across names like:

• CCJ
• LEU
• BE
• CEG
• BWXT
• DNN
• NXE
• GEV

‍