For the last two years, the AI story for investors has mostly been about who builds the technology. Chipmakers, cloud providers and power companies have captured most of the enthusiasm, and Morgan Stanley Research expects roughly $800 billion in AI-related capital spending this year, rising to $1.1 trillion in 2027 [1].
Morgan Stanley strategists Ellen Zentner and Sarah Wolfe argue the next phase of the trade looks different. As the infrastructure buildout matures, the more useful question for investors becomes which companies can actually use AI well enough to cut costs, sharpen decisions and grow margins.
#The Gap Between Access And Real Use
Stanford University's 2026 AI Index found that 88% of organizations now have some access to AI tools [2]. But a separate S&P 500 study by MIT researchers posted on arXiv, found only about one in five companies had AI genuinely embedded in production or core business processes as of 2025 [3].
Morgan Stanley calls the spending needed to close that gap intangible capex. It covers the less glamorous work of making AI actually pay off, things like clean data systems, governance, cybersecurity, staff training and redesigning how work gets done. Firms that skip this step and just bolt AI onto old workflows tend to get the adoption headlines without the productivity gains.
#The Payoff Often Shows Up Late
Even companies doing this work properly may not see results right away. Economists call this pattern a productivity J curve, where spending on reorganizing data and retraining staff comes first, and the payoff in margins or earnings shows up later.
There are early signs that payoff is starting to show. Morgan Stanley Research reviewed more than 10,000 earnings and conference transcripts and found that about 25% of S&P 500 companies cited at least one quantifiable AI impact by July 2026, up from 15% in the third quarter of 2025, as shown above. That is still a minority of the index, but the direction is the signal worth tracking.
A sharper way to see the markets in just 5 minutes.
Same news, different lens. We cut through the noise and hand you the overlooked ideas and the deeper read the crowd misses. Join 38,000+ investors seeing the markets differently.
#Where Deep Adoption May Emerge Next
For now, measurable AI gains are concentrated in tech and communication services. Morgan Stanley points to three other sectors it is watching for the next wave of deep adopters.
In financials, select banks, insurers and wealth managers can use AI to speed up document heavy workflows and support decisions in highly regulated settings. In health care, embedded AI can ease administrative burdens and support clinical documentation and drug development. In retail, large chains with big customer datasets and complex supply chains can apply AI across merchandising, inventory, logistics and customer service.
The common thread across all three is scale. These are sectors where companies tend to have the structured data, balance sheet flexibility and management appetite to actually redesign workflows rather than layer AI on top of them.
#What Could Slow The Payoff
The risks are real. AI outputs are not always reliable, and outcomes depend on the quality of human oversight built around the models. Regulation and data privacy rules could tighten in ways that slow adoption in the very sectors seen as best positioned. And plenty of AI spending will simply not turn into durable profit, no matter how it gets marketed.
For investors, the takeaway is not to chase every AI announcement or pilot program. It is to look for companies showing real, quantifiable results, and to watch whether that 25% adoption figure keeps climbing in the quarters ahead.