US economic outlook Report Interpretation

Overview

Morgan Stanley’s US outlook argues that the economy’s growth mix is shifting toward AI-related business investment rather than consumer spending. The report expects resilient but uneven activity, gradual disinflation, a stable low-hire/low-fire labor market, and delayed Federal Reserve easing in 2027.

Core views

The central thesis is “capex over consumption.” Higher gasoline prices are expected to neutralize much of the household boost from OBBBA-related fiscal measures and larger tax refunds, limiting goods spending. Morgan Stanley now expects real consumption growth of 2.1% 4Q/4Q in both 2026 and 2027. Refunds were about 17% higher year-on-year, or roughly $47 billion, and the average refund rose about $320, but the report estimates that retail gasoline prices averaging $3.60 per gallon would offset this support. Consumption has remained resilient because households reduced saving and paid down less debt, but lower- and middle-income households have limited buffers: liquid savings cover three months of spending for the bottom income quintile, versus 19 months for the top quintile. The top 20% holds about 75% of household net worth, making upper-income spending more sensitive to asset-market wealth. Business investment is the counterweight. Morgan Stanley forecasts nonresidential fixed investment growth of 7.7% in 2026 and 8.0% in 2027, led by AI infrastructure, compute, and power capacity. Hyperscaler capex is expected to exceed $1 trillion in 2027, and analysts have repeatedly revised those estimates higher, pushing the anticipated growth slowdown further into 2027. The report treats AI spending as structurally rather than cyclically driven. After accounting for imports, it estimates AI-related investment contributed 0.6 percentage points to GDP growth from 2025 through mid-2026, while AI-only activity added 0.4 points; AI-related spending and productivity are forecast to add 0.8-0.9 points to growth in 2026 and 2027. Infrastructure spending is differentiated from adoption spending: the former builds computing capacity, while software-led adoption reflects diffusion across businesses. Computers and peripherals account for 60-70% of AI infrastructure spending. The report sees housing as a persistent drag. Real residential investment is expected to decline 1.3% 4Q/4Q in 2026 before rising 1.5% in 2027. Affordability remains near four-decade lows, while the gap between market mortgage rates and homeowners’ effective rates keeps owners locked in. The FHFA estimates each one-percentage-point increase in that gap reduces sale probability by 18.1%; at a current 210bp gap, existing-home sales remain under substantial pressure. New-home sales have stagnated and inventories are elevated, so starts should remain subdued until backlogs clear. Fiscal policy supports near-term activity but enlarges deficits. Morgan Stanley projects a $2.05 trillion fiscal deficit in 2026, or 6.3% of GDP, and deficits around 6% of GDP over the forecast horizon. It estimates OBBBA adds about 0.4 percentage points to GDP in 2026 and 0.14 points in 2027, but has a negative medium-term effect. The tariff regime is viewed as increasingly durable: tariffs averaged 6.8% in April-June and the statutory effective rate is expected to approach 10% by year-end, with implementation, Section 232 actions, and country-specific escalation remaining uncertain. Inflation is expected to descend gradually. Core PCE is forecast at 3.2% 4Q/4Q in 2026 and 2.4% in 2027. Tariffs have contributed roughly 60bp to core prices so far in 2026, against estimated total pass-through of about 70bp, implying tariff pressure should fade. Continued shelter disinflation should help, while the report expects only limited oil pass-through to core inflation. Still, inflation persistence is a risk: 54% of PCE components were rising more than 3%, versus 38% in 1Q25. The baseline calls for the Fed to hold rates through 2026, followed by 50bp of easing in 2027. The labor-market baseline is steady rather than strong: slow payroll gains, low turnover, a low unemployment rate, and moderating wages. Morgan Stanley estimates payroll breakeven at 50,000 jobs per month in 2026 and 40,000 in 2027, reflecting constrained labor supply from immigration restrictions and demographics. AI is currently associated with limited aggregate disruption, although high-exposure occupations show unemployment roughly 0.5 percentage points above normal after adjusting for cyclicality, and younger workers show somewhat more disruption. The report finds high-AI-exposure industries contributed 1.7 points of the 2.4-point increase in output per employee through 4Q25, up from 0.7 points in 2024; this reflected faster output growth rather than broad labor displacement. AI’s financing and distributional effects are increasingly material. Morgan Stanley expects $3.2 trillion of issuance to support data-center spending through 2028, including $1.8 trillion from credit markets and $700 billion from private credit. Hyperscaler credit quality remains strong, so wider spreads rather than constrained debt capacity are expected to be the adjustment mechanism. Data-center debt is about 3% of the high-yield index, up from less than 1% a year earlier. The report also estimates a 47GW US power shortfall facing developers through 2028. College-educated, high-income, city-dwelling households are more exposed to both AI displacement and AI-related wealth, job-creation, and disinflation benefits; in the base case they benefit most, but faster diffusion or asset-market weakness could initially affect them and then spread demand weakness across cohorts. Alternative scenarios underscore the sensitivity to oil, demand, and AI diffusion. A 15%-probability demand-upside case produces 2.9% GDP growth in 2026 and 3.1% in 2027, with Fed hikes beginning in 4Q26. A 10%-probability AI-productivity-with-displacement scenario raises unemployment to 4.5% in 2026 and 4.8% in 2027 and brings three rate cuts in early 2027. A 20%-probability permanent oil-premium scenario keeps oil above $100 through 2027, core PCE at 3.1% in 2026 and 2.8% in 2027, and rates unchanged through 2027. A 15%-probability global-recession scenario assumes oil reaches $140-160 per barrel through 3Q26, GDP contracts in 2H26, unemployment reaches 5.5%, and the Fed cuts 200bp in 2026.

Analysis framework

The report combines a GDP expenditure breakdown with forecasts for consumption, investment, housing, fiscal policy, trade, inflation, labor, and monetary policy. It separates AI infrastructure from AI adoption, adjusts capex for imports when estimating GDP effects, compares alternative macro scenarios, and uses occupation and industry AI-exposure measures to assess productivity and labor-market transmission.

Methodology notes

• Industry AnalysisVolume-price decomposition

  Corporate pricing, costs, and profit decomposition

  Morgan Stanley attributes the 2Q26 corporate-profit surge primarily to higher selling prices while unit labor and nonlabor costs were broadly flat.

• MacroeconomicsTaylor rule

  New Keynesian AI-diffusion simulation with a Taylor-rule monetary-policy response

  The report models how AI adoption speed, displacement, task creation, wealth effects, and policy responses affect unemployment, inflation, output, and rates.

• Other

  Felten et al. (2021) AI exposure index

  The report classifies occupations by how their tasks align with current and anticipated AI capabilities, then compares labor outcomes across exposure groups.

Key data

• Real GDP growth2.2% in 2026; 2.6% in 20274Q/4Q forecast
• Personal consumption expenditures2.1% in 2026; 2.1% in 20274Q/4Q forecast
• Nonresidential fixed investment7.7% in 2026; 8.0% in 20274Q/4Q forecast
• AI contribution to GDP growth0.8pp in 2026; 0.85pp in 2027AI-related investment plus productivity contribution
• Core PCE inflation3.2% in 2026; 2.4% in 20274Q/4Q forecast
• Federal deficit$2.05tn, or 6.3% of GDP2026 projection
• Data-center financing$3.2tn through 2028Including $1.8tn from credit markets and $700bn from private credit
• Power shortfall47GW through 2028Projected US shortfall facing developers

Impact & implications

The report expects AI-related capex to make the US expansion more investment-led and less dependent on broad consumer demand. It also highlights a growing link between AI investment, power constraints, credit issuance, productivity gains, and uneven household outcomes, while inflation and energy-price risks keep monetary-policy normalization gradual.

Risks

• A sustained oil-price premium could weaken consumption and keep inflation elevated.
• Faster-than-expected AI diffusion could cause greater labor displacement, especially if task creation and policy feedbacks lag.
• Inflation may prove more persistent than forecast because of Middle East conflict or AI-related demand-side price pressures.
• Tariff implementation, further Section 232 actions, and country-specific escalation remain uncertain.
• Data-center power shortages and increasing AI-related credit issuance could raise financing and infrastructure pressures.

What to watch

• Retail gasoline prices and whether they offset tax-refund support for household spending.
• The pace of hyperscaler capex, compute and power-capacity construction, and AI adoption spending.
• Shelter inflation, tariff pass-through, and the breadth of PCE components above 3% inflation.
• Mortgage-rate lock-in, housing affordability, new-home inventories, and housing starts.
• AI-exposed occupations, especially unemployment and job-finding outcomes for younger workers.
• Data-center credit issuance, credit spreads, and the projected US power shortfall.