The AI Builders Have Had Their Day. Now Watch the Adopters.

September 16, 2026

6 minute read

The first chapter of the AI investment story was easy to write. Buy the builders: the chip makers, the hyperscalers, the data center REITs, the power infrastructure plays. That trade has delivered. The S&P 500 has logged 24 new all-time highs year-to-date, with technology and communication services posting earnings growth of roughly 50% year-over-year in Q1 alone, a figure that overshadows every other sector by a factor of two or three. ¹ But the builders are no longer cheap, the infrastructure buildout is now consensus, and the more consequential question for the next five years is the one most advisors have not yet asked: who benefits from using all of this?

The answer is not a single sector. It is a restructuring of competitive advantage across the entire economy, and the early evidence suggests the map looks very different from the one Wall Street has priced.

The AI Factory Is Already Running

Understanding who benefits from AI adoption requires first understanding what the infrastructure actually produces. Irena Petkovic, CFA, Equity Analyst at Mawer Investment Management, reframes the data center not as a cost center but as an AI factory: a facility that converts electricity into tokens, with incremental cost per token that is "basically just electricity" once the fixed capital is sunk. ² The four largest hyperscalers are spending $700 billion on capital expenditure this year alone, a figure that dwarfs the peak telecom buildout of the dot-com era. The early return signals are credible: AWS is growing at 37% at twice its prior scale; over 90% of Microsoft's cloud revenues now come from customers outside the frontier labs; Amazon's server payback period is under three years. The buildout is not speculative capex. It is producing returns.

What the factory produces, beyond raw compute, are four distinct monetization channels: GPU rental, AI-enabled productivity software, core business enhancement (Meta's ad targeting is the clearest example), and direct token sales. Each channel addresses a different layer of the adopter ecosystem. The enterprise software layer, in particular, is where the next durable value creation is likely to concentrate.

Productivity Is Already Showing Up in the Data

J.P. Morgan Asset Management's Dr. David Kelly and Gabriela Santos make a point that should anchor every advisor's AI thesis: the economy is not being driven by broad cyclical expansion. Real GDP growth is running at roughly 3% in Q2 and Q3, but employment gains are decelerating sharply to between 50,000 and 75,000 jobs per month. The gap is filled by productivity, now running at 2.5% output per hour, a rate not sustained in a generation. "If wage growth remains relatively moderate in a tight labor market with strong productivity gains," Kelly and Santos write, "the danger of sticky inflation is much reduced." ³ For investors, this is the signal: the companies capturing that productivity gain are the adopter beneficiaries.

The scale of agentic AI adoption makes this structural, not cyclical. Sandy Kaul, Head of Digital Assets and Innovation at Franklin Templeton, projects that by 2028, 38% of organizations will have AI agents as active team members, collaborating on complex, multi-step tasks without constant human oversight. ⁴ Agentic commerce, the autonomous purchase and exchange of goods and services by AI agents, could reach $3 trillion to $5 trillion by 2030. IBM's warning in July that corporate technology spending is shifting rapidly toward AI infrastructure, sending its stock down 25.2% in a single session, was a preview of what conventional IT vendors face and what AI-native replacements stand to capture.

Asset Management: A Case Study in Institutional Adoption

Boston Consulting Group's July 2026 analysis of the asset management industry offers the sharpest sector-level case for AI adoption as a competitive wedge. ⁵ The industry has grown AUM to $147 trillion while margins have remained flat at 30% for fifteen years; costs have outpaced revenues for fifteen consecutive quarters. BCG's verdict: AI-first firms are already generating 3.8 times higher total shareholder returns and deploying 20 times more agentic solutions than laggards. The mechanism is specific: agentic AI compresses the alpha decay cycle from months to hours, expands investment coverage by two to five times, and unlocks 55% to 65% of operations capacity. "300 to 500 basis points of value across the P&L," BCG writes, "the difference between a 30% margin and a 40%-plus margin business." The 70% of firms still on the sideline are not safe; they are simply later to a race that has already started.

The Physical Economy Dividend

Not every AI beneficiary is a technology company. AllianceBernstein portfolio managers Snezhana Otto and Justin Moreau make the counterintuitive case that large-cap value stocks are the smarter AI-era allocation for much of a diversified portfolio. ⁶ Physical-asset-heavy sectors, from aerospace components (RTX, Hexcel) to agricultural machinery (Deere, CNH Industrial), either sit outside AI's near-term disruption radius or benefit directly from the AI infrastructure buildout. Capital Group's equity portfolio managers, surveying the H2 2026 landscape, extend the same logic: banks are capitalizing on higher interest rates, healthcare is growing on innovative therapies, and energy is benefiting from higher crude. ⁷ The AI freight train, in Capital Group's framing, "potentially is the highest impact technology in a generation," but its total addressable market reaches across every sector of the physical economy, not just software.

The corollary is equally important. McKinsey estimates AI will automate 30% of all hours worked within four years; the World Economic Forum projects 92 million displaced workers by 2030. Companies with heavy exposure to white-collar labor and few physical assets, including certain software vendors, office REITs, and clinical research services firms, face a different reckoning. The adopter dividend and the disruption tax are two sides of the same ledger.

The Competitive Coexistence Framework

Capital Group's Mark Casey, Winnie Kwan, Sugi Widjaja, and Jared Franz offer the clearest lens for thinking about the geopolitical dimension of AI adoption. The likely outcome over five years is not winner-takes-all but competitive coexistence: the U.S. pursues artificial superintelligence and frontier model dominance, while China focuses on industrial applications, robotics, and autonomous vehicles at lower compute cost. ⁸ For investors, this means the total addressable market for AI adoption is global, the competitive dynamics are bifurcated, and broad U.S. hyperscaler exposure is only part of the story. The industrial AI wave rolling through Chinese manufacturing, even if it runs on less sophisticated models, will produce a generation of operationally efficient manufacturers that advisors with undiversified, U.S.-centric AI exposure may not be positioned to capture.

5 Key Takeaways

1. The AI infrastructure trade is maturing; the adoption trade is beginning. With hyperscaler CapEx at $700 billion annually and server payback periods under three years, the buildout is producing real returns. The next five years belong to the sectors that deploy what has been built.

2. Productivity is the signal. Output per hour is running at 2.5%, a generational high, in an environment of decelerating job growth. The companies capturing that productivity gain are the durable AI adopter beneficiaries; the ones failing to capture it are the displacement casualties.

3. Asset management is a leading indicator for institutional AI adoption. BCG's finding that AI-first firms generate 3.8 times higher total shareholder returns is not limited to asset managers. Any knowledge-intensive industry with high fixed costs, wide performance dispersion, and data-rich client relationships faces the same competitive bifurcation.

4. Physical economy sectors are underappreciated adopter plays. Healthcare, aerospace, agriculture, banking, and energy are neither disrupted by AI nor merely adjacent to it. They are direct beneficiaries of AI-driven efficiency, and their physical asset base insulates them from the displacement risk hitting software, office REITs, and white-collar-intensive businesses.

5. Competitive coexistence, not winner-takes-all, is the correct geopolitical framework. U.S. frontier AI and Chinese industrial AI serve different markets, command different price points, and will likely grow in parallel. Advisors with U.S.-only AI exposure are structurally underweight the industrial AI wave building in global manufacturing and emerging markets.