US midterms put Congress in play

Podcast: Signal over Noise: Finding quality beyond AI. Listen on Apple or Spotify (5 min)
Podcast: What next for tech after the tech sell-off? Plus the US midterms arrive (5 mins)
Video: AI sector positioning as capex, volatility pick up (3 mins)

After a historically expensive campaign season, voting in the US midterm elections begins today, with all 435 House seats and 35 Senate seats being contested. Consensus expectations appear to be building around a “blue sweep” outcome, in which Democrats take control of both chambers.

The latest polls show that President Trump’s overall net approval rating stands at -20 percentage points (pps), according to RealClear Polling’s (RCP) average. Democrats also lead the Republicans by 8.9pps in RCP’s average of generic congressional voting polls, which ask voters whether they would rather vote for a generic Democrat or Republican in the next election. On specific issues, President Trump's net approval rating is -32pps on the economy and -43pps on inflation. Early voting data suggests turnout has so far favored Democrats over Republicans more than in the last election cycle in 2024 in several key states. Nonetheless, these figures do not tell the whole story: Around half of the national vote has come on Election Day in recent elections.

The blue sweep remains our highest probability outcome, though we note the Senate remains closely fought and the eventual size of any majority is uncertain. Polling in a number of Senate races shows an outcome that lies within the margin of error. Past elections show that polls can misjudge turnout or miss important drivers that only become evident after voting concludes.

Although voting ends on Tuesday, the result may not be clear in the first hours after polls close. High volumes of mail-in ballots, several closely contested races, and Alaska’s ranked-choice system could delay decisive calls, while election-related litigation is already under way. Even if Democrats win both chambers, narrow majorities, Senate rules, party divisions, and presidential veto power would constrain major legislative change.

We would not treat the midterms as a binary election event. While they may influence the medium-term outlook, the outcome is unlikely to displace the primary market drivers of the moment: AI investment, economic growth, corporate earnings, inflation, and interest rates. Below, we set out our expectations and reasoning:

Gridlock would likely constrain major policy shifts. A change in congressional control would increase oversight and make fiscal negotiations more difficult, but it would not automatically produce sweeping changes in tax or spending policy. President Trump would retain veto power, while the executive branch would preserve broad authority over tariffs and foreign policy, leaving room for volatility. A material surprise could shift perceptions of the 2028 race and bring future policy priorities into view, although we think it is too early for markets to price such implications sustainably. The size and composition of the eventual majority will therefore be important to watch.

Debt talks remain the clearest risk for bonds. The next Congress will soon confront the 2027 debt ceiling negotiations, which we think could result in scheduled social program cuts being softened and defense spending rising. Together, these measures could reverse an expected narrowing in deficits and increase Treasury supply. Political brinkmanship could also heighten bond market volatility. Still, we expect inflation to moderate, the Federal Reserve to finish raising rates, and Treasury yields to settle lower. Current yields offer attractive income, although persistent deficits and heavy issuance underpin our caution toward the longest maturities.

AI demand is strong, but selectivity matters. Greater congressional oversight would not necessarily derail the AI investment cycle. Broad federal restrictions would face legislative and presidential constraints, while state and local rules remain more important for data center timing and location. AI infrastructure demand remains exceptionally strong, and we have raised our aggregate AI capex forecasts to USD 1.0tr in 2026 and USD 1.4tr in 2027. Still, global investors are placing greater weight on the risks surrounding overinvestment, financing, and monetization. Higher volatility could become the new normal, favoring financially strong platforms and infrastructure providers, as well as diversified exposure across the AI value chain.

So, regardless of which outcome investors anticipate or prefer, we suggest avoiding large portfolio positions based on a binary political result. Outside politics, strong earnings and AI investment support our constructive view on equities. That said, we think investors should actively look to reduce single-name concentration in their portfolios, rebalance oversized technology exposure, and within tech, tilt toward quality companies across the AI value chain.

We also see attractive income and diversification potential in high-quality bonds, particularly short- to medium-term maturities, while remaining cautious on the longest maturities. Investors can use volatility to rebalance or build exposure gradually, adjusting portfolios only if policy changes materially alter the fundamental outlook.