Big Tech lost big in the 2026 election

What It Covers

The Vergecast analyzes the 2026 midterm elections, in which Democrats won decisively across the US, with 40 of 41 state ballot measures blocking data center construction passing, Elon Musk's $500 million political investment yielding no results, and AI companies facing mounting evidence that consumer and enterprise products are failing to generate sustainable revenue or public trust.

Key Questions Answered

• •Data Center Ballot Measures: 40 out of 41 data center-related ballot measures across 10 states failed in the direction pro-AI industry forces wanted. The sole exception was Hardwood, North Dakota. Voters who oppose AI infrastructure are not a fringe group — they represent a broad, cross-partisan majority. Companies planning data center expansion should treat state-level ballot opposition as a primary risk factor, not a regulatory footnote, when modeling infrastructure timelines and capital deployment.
• •AI Agent Adoption Reality: Uber fulfilled exactly four rides via AI agents last quarter. DoorDash processed a few hundred AI agent orders. Uber Eats handled roughly 500. Combined, these numbers suggest the entire US AI agent economy for consumer commerce may fit inside a single QA team's testing log. Companies projecting agent-driven revenue growth should stress-test those projections against actual adoption curves, not analyst forecasts or lab demonstrations.
• •Google's Self-Cannibalization Problem: Google search query volume is growing at its fastest rate in five years, but AI Overview monetization lags standard search ads significantly. Chief Business Officer Philipp Schindler confirmed the monetization gap publicly, triggering a stock selloff. Google is actively replacing its highest-margin product with an unproven one. Advertisers and publishers dependent on Google traffic should model scenarios where AI mode becomes default within 18 to 24 months.
• •OpenAI Search Claims Require Scrutiny: OpenAI announced it processes roughly one-fifth of Google's search volume, but declined to disclose the source of Google's baseline figure or clarify how it defines a "search query." The metric counts any user input triggering a web lookup, including tasks that differ fundamentally from traditional search intent. Treat this figure as a competitive positioning statement rather than a verified market share data point when evaluating AI search competitive dynamics.
• •Enterprise Token Spending Contraction: Apple reduced its daily AI token cap for engineers from $300 to $200 per person, signaling that enterprise customers are beginning to audit AI tool ROI rather than expand usage uncritically. This pattern — initial adoption followed by cost rationalization — mirrors historical SaaS adoption cycles. Enterprise AI vendors should expect renewal conversations to shift from capability demos to productivity outcome documentation within the next one to two quarters.
• •Meta Muse Adoption Curve: Third-party app tracking data indicates Muse download velocity and App Store review volume have declined sharply since launch. User behavior shows a pattern of one to two sessions followed by disengagement. The novelty of agentic task completion — ordering food, making phone calls — does not appear sufficient to build habitual daily use. Product teams building consumer AI agents should prioritize identifying recurring, high-frequency tasks rather than impressive but infrequent showcase capabilities.
• •X Platform Political Irrelevance: Elon Musk spent approximately $500 million supporting Republican candidates and used X as a primary influence amplification tool. Democrats won decisively up and down the ballot. X's user base was systematically wrong in its political predictions, while platforms like Instagram comments and Reddit threads more accurately reflected broader public sentiment. Brands and political strategists should weight X engagement metrics at a significant discount relative to Instagram, TikTok, and Reddit signals when gauging public opinion.