AI Spending Now Lifts All 11 S&P 500 Sectors: What That Means for Allocation
For the first time since late 2021, analysts project positive earnings growth across all eleven sectors of the S&P 500 when results are tallied for the fourth quarter. The common interpretation is that AI spending is finally paying off. The structural interpretation is that the spending moved from concentration to distribution.
The data center buildout is ending
Between 2023 and early 2025, most of the capital earmarked for AI went to semiconductor manufacturers and cloud hyperscalers. Nvidia sold chips. Microsoft, Amazon, and Alphabet built data centers. The spending was enormous, cumulative AI-related capital expenditure by the major hyperscalers will cross $1.0 trillion by the end of 2026, but it flowed through a narrow set of companies. The S&P 493, which is the index minus the Magnificent Seven, posted mid-single-digit earnings growth while the top firms printed 20% gains quarter after quarter.
What changed in 2026 is that the spending shifted from building the rails to running trains on them. Utilities are seeing double-digit earnings growth as data centers in Ontario, the U.S. Northeast, and the Pacific Northwest strain electrical grids. In some regions, energy providers have become the bottleneck. New capacity takes years. Demand showed up in eighteen months.
Financial services firms are using generative models for credit underwriting and fraud detection. The efficiency gains are large enough to show up in net interest margins, which had been under pressure since rates began climbing in 2022. Healthcare companies are deploying AI to compress drug discovery timelines and identify clinical trial candidates faster. Industrial manufacturers have rolled out predictive maintenance systems that cut unplanned downtime costs by margins wide enough to offset rising labor expenses in tight markets.
The adoption is broad, but the impact is uneven. The sectors seeing the largest uplifts are the ones where AI addresses a structural cost or a capacity ceiling that was already binding. Utilities face energy demand they cannot meet. Financials face regulatory pressure to improve risk models. Healthcare faces patent cliffs and lengthening approval timelines. In these cases, AI is relieving a constraint that was already blocking growth.
The "phase two" trade is valuation-sensitive
The catch is that the market has priced in much of this already. Price-to-earnings ratios in Utilities and Financials are now at the high end of their historical ranges. The S&P 500 as a whole is trading near 21 times forward earnings, which is above the ten-year average and leaves little room for disappointment. If the productivity gains plateau or if the capital expenditure required to maintain them starts climbing faster than the returns, the repricing will be sharp.
The 10-year U.S. Treasury yield is hovering in the 4.0% to 4.5% range as of mid-2026, which raises the discount rate applied to long-term growth projections. A company that was pricing in ten years of margin expansion at 2% rates now faces a higher hurdle. The math works if the productivity miracle narrative holds. It stops working the moment the narrative cracks.
Canada's major banks and energy firms are following a similar path. Royal Bank and TD are integrating AI into back-office operations and client-facing platforms. Enbridge and TC Energy are positioning themselves to broker power agreements between U.S. data centers and Canadian electrical suppliers for cross-border expansions. The spillover effect is real, but it is derivative. If U.S. hyperscalers pull back on capital spending, the Canadian plays are second in line for the repricing.
Broad-market sectors are now seeing earnings growth outside technology for the first time since 2021. Whether this justifies current valuations depends entirely on whether the cost savings materialize at the scale the forward estimates assume. The market is no longer betting on who builds the system. It is betting on who saves the most money using it.
For the first time since late 2021, analysts project positive earnings growth across all eleven sectors of the S&P 500 when results are tallied for the fourth quarter. The common interpretation is that AI spending is finally paying off. The structural interpretation is that the spending moved from concentration to distribution.
The data center buildout is ending
Between 2023 and early 2025, most of the capital earmarked for AI went to semiconductor manufacturers and cloud hyperscalers. Nvidia sold chips. Microsoft, Amazon, and Alphabet built data centers. The spending was enormous, cumulative AI-related capital expenditure by the major hyperscalers will cross $1.0 trillion by the end of 2026, but it flowed through a narrow set of companies. The S&P 493, which is the index minus the Magnificent Seven, posted mid-single-digit earnings growth while the top firms printed 20% gains quarter after quarter.
What changed in 2026 is that the spending shifted from building the rails to running trains on them. Utilities are seeing double-digit earnings growth as data centers in Ontario, the U.S. Northeast, and the Pacific Northwest strain electrical grids. In some regions, energy providers have become the bottleneck. New capacity takes years. Demand showed up in eighteen months.
Financial services firms are using generative models for credit underwriting and fraud detection. The efficiency gains are large enough to show up in net interest margins, which had been under pressure since rates began climbing in 2022. Healthcare companies are deploying AI to compress drug discovery timelines and identify clinical trial candidates faster. Industrial manufacturers have rolled out predictive maintenance systems that cut unplanned downtime costs by margins wide enough to offset rising labor expenses in tight markets.
The adoption is broad, but the impact is uneven. The sectors seeing the largest uplifts are the ones where AI addresses a structural cost or a capacity ceiling that was already binding. Utilities face energy demand they cannot meet. Financials face regulatory pressure to improve risk models. Healthcare faces patent cliffs and lengthening approval timelines. In these cases, AI is relieving a constraint that was already blocking growth.
The "phase two" trade is valuation-sensitive
The catch is that the market has priced in much of this already. Price-to-earnings ratios in Utilities and Financials are now at the high end of their historical ranges. The S&P 500 as a whole is trading near 21 times forward earnings, which is above the ten-year average and leaves little room for disappointment. If the productivity gains plateau or if the capital expenditure required to maintain them starts climbing faster than the returns, the repricing will be sharp.
The 10-year U.S. Treasury yield is hovering in the 4.0% to 4.5% range as of mid-2026, which raises the discount rate applied to long-term growth projections. A company that was pricing in ten years of margin expansion at 2% rates now faces a higher hurdle. The math works if the productivity miracle narrative holds. It stops working the moment the narrative cracks.
Canada's major banks and energy firms are following a similar path. Royal Bank and TD are integrating AI into back-office operations and client-facing platforms. Enbridge and TC Energy are positioning themselves to broker power agreements between U.S. data centers and Canadian electrical suppliers for cross-border expansions. The spillover effect is real, but it is derivative. If U.S. hyperscalers pull back on capital spending, the Canadian plays are second in line for the repricing.
Broad-market sectors are now seeing earnings growth outside technology for the first time since 2021. Whether this justifies current valuations depends entirely on whether the cost savings materialize at the scale the forward estimates assume. The market is no longer betting on who builds the system. It is betting on who saves the most money using it.
Read Next
7 Essential Checks Before You Buy in 2026: Building Confidence Through the Numbers
Getting Lenders to Count 80% of Your Gross Rent: A Checklist
820 Low-Rise Homes, 129 Condos: The GTA Slump Belongs to Speculators
200 Basis Points or 2 Percentage Points: Same Stress Test, Clearer Wording