AI’s Expanding Footprint

Author: Samuel Miller
CFA®, CFP®, CAIA®
Executive Vice President of Investment Strategy
June 2026

For most of the past two years, AI as an investment theme meant one thing: a handful of mega-cap technology companies whose capital spending was reshaping global markets. That concentration served early investors well. June offered a reminder that it will not always work that way.

The five major hyperscalers just logged one of their worst months on record as a group, with the basket falling more than 12% from its late May peak. Microsoft alone is on pace for its steepest monthly decline since 2000. The proximate cause is not a slowdown in AI demand. Revenue and earnings at these companies have continued to beat expectations. The market’s concern is narrower and more specific: capital spending has climbed so quickly that it is now consuming nearly all of operating cash flow, leaving little room for the buybacks and dividends that have long supported these stocks.

The more notable detail is what did not fall alongside them. Semiconductor stocks climbed to new highs over the same stretch, and the broader market outside the largest technology names outperformed by a wide margin. That divergence is itself the clearest evidence yet for the thesis this series has been building: the AI investment cycle and the hyperscaler stock price are no longer the same trade.

The hyperscalers’ difficult month is not a verdict on AI. It is a reminder that the capital they are deploying does not stay with them. It radiates outward through semiconductors, power systems, industrial equipment, and ultimately the companies that put the technology to work.

The Infrastructure Wave Is Far From Over

None of this changes the underlying picture. Microsoft, Alphabet, Amazon, and Meta have each disclosed plans to sustain or accelerate data center investment through the balance of the decade, and the numbers involved are of a scale that tends to concentrate supply chains around a relatively narrow set of suppliers. That dynamic produced the first generation of AI beneficiaries: the GPU manufacturers and cloud platform operators whose earnings reflected the initial surge in demand.

What June made clear is that the hyperscalers themselves are now absorbing the cost of that build-out, while the companies supplying the picks and shovels are capturing more of the upside. Each dollar spent on a data center pulls through demand for advanced semiconductors, specialized networking equipment, cooling systems, electrical infrastructure, and the land and steel that underpin it all. What started as a technology spending story is increasingly visible in earnings reports far outside of Silicon Valley.

Where the Cycle Is Broadening

Several categories of beneficiaries are becoming more visible in portfolio analysis:

• Semiconductors beyond GPUs. The initial AI trade centered on graphics processing units. The more durable opportunity spans custom silicon, high-bandwidth memory, and advanced packaging, categories where the barriers to entry are high and the demand outlook is extending well beyond 2026.
• Power and utilities. Data centers are among the most power-intensive facilities ever built at scale. Grid operators, independent power producers, and utilities serving high-density computing corridors are seeing demand growth they had not modeled for this decade. The electricity constraint is real, and the companies positioned to address it are attracting capital accordingly.
• Industrial enablers. The physical infrastructure of AI includes cooling systems, electrical switchgear, fiber networks, and construction materials. These are not glamorous categories, but their order books reflect the scale of what is being built.
• End-user adopters. Perhaps the most underappreciated dimension of the broadening is what happens when AI reaches enterprises at scale. Companies in healthcare, financial services, logistics, and professional services are beginning to generate measurable productivity gains. Some are converting those gains into margin improvement, others into competitive differentiation. In either case, AI is becoming a fundamental driver, not a line item.
• Beneficiaries outside the US. The build-out is not confined to American balance sheets. Asian and European semiconductor manufacturers, electrical equipment makers, and industrial suppliers are capturing a meaningful share of hyperscaler capital spending, and several have outperformed their US technology counterparts this year. A theme this large was never going to respect borders.

What This Means for Portfolios

To be clear, none of this is an argument for abandoning the hyperscalers. They remain among the most profitable, well capitalized companies in the world, and they continue to control the platforms through which most enterprise AI demand will ultimately flow. A rough June, even a historically rough one, does not change that. These companies may continue to play an important role in client portfolios.

What June does argue for is making sure that role is sized appropriately rather than left to grow by default. Investors who framed AI exposure as a concentrated bet on a small number of household names were not wrong in 2023 or 2024. The trade worked precisely because capital initially concentrated where the opportunity was most visible. The risk now is treating that early concentration as a permanent feature of the theme rather than a temporary one.

A cycle of this magnitude, driven by structural demand for compute, power, and connectivity, has historically produced its most durable returns not in the initial wave of infrastructure spending but in the broader adoption that follows, and not in any single country. Diversifying into the semiconductor, power, industrial, and adopter categories described above, alongside the international suppliers capturing their own share of this spending, gives portfolios more ways to participate in the AI cycle and reduces reliance on any single group of stocks absorbing the next bout of capex anxiety.

Active management has a meaningful role to play here, as the dispersion of outcomes across sectors and geographies widens and the ability to identify second-order beneficiaries becomes an increasingly important source of return.

Bottom Line

The hyperscaler capex chart is the right starting point for understanding AI as an investment cycle, because it shows the scale and durability of the build-out in concrete terms. What the chart cannot show is that the companies funding that spending and the companies benefiting from it are increasingly different ones, a distinction June made impossible to ignore. The AI opportunity is becoming an economy-wide, increasingly global investment cycle, and portfolios built to participate in that broadening are better placed than those still anchored to the narrower thesis that defined the theme’s early chapters.

---

This commentary is for informational purposes only and does not constitute investment advice. Past performance is not indicative of future results.

Signature Estate & Investment Advisors, LLC (SEIA), an SEC-registered investment adviser, notes that such registration does not imply specific skill or training; no contrary inference should be drawn. This material is provided for general informational purposes only and does not constitute individualized investment advice or a recommendation of any specific security, strategy, or course of action. Past performance is not indicative of future results, nor do asset allocation or diversification guarantee profit or prevent loss. Investors should consult with their financial advisor before making any investment decisions.

 

---