Where did all the metaverse money go?

Autonomous vehicles have crossed from perennial "five years away" promise into genuine, if uneven, commercial reality. Waymo is the undisputed global leader in robotaxi deployment, operating across 14 U.S. cities with over 4,000 vehicles and ~500,000 paid rides per week ^[11]. Aurora is the first mover in commercial driverless freight, running ~10 Sun Belt trucking routes with a target of 200 driverless trucks by year-end ^[24][43]. Tesla, Zoox, and Chinese operators Baidu, Pony.ai, and WeRide have all entered paid commercial service in 2025–2026. The technology has demonstrably matured enough to outperform human drivers on safety metrics in controlled geographies—Waymo reports 94% fewer serious-or-fatal-injury crashes than comparable humans over 220+ million miles ^[34]. Yet none of these businesses is profitable on a fully-loaded basis, hardware costs remain high, the "long tail" of edge cases is unsolved, and regulatory frameworks are still being written. This is an inflection point, not a fait accompli.

Getting into Y Combinator in 2025–2026 is harder by raw numbers than at any prior point in the program's history—estimated acceptance rates hover between 0.6% and 2% depending on the batch and counting method—but YC's own guidance and recent founder accounts converge on a surprisingly consistent set of requirements ^[1][2][4]. The fundamentals have not changed: YC bets on founders first, then ideas. What has changed is the context: AI-native companies now dominate the applicant pool and the funded batch alike, competition has intensified dramatically, and YC's 2026 Requests for Startups signal a sharp shift toward physical-world AI, defense, infrastructure, and government technology ^[13][26][40]. This report synthesizes YC's primary guidance, partner statements on record, and accounts from founders accepted in the 2024–2026 window to give decision-makers a clear picture of what actually earns a spot.

The short answer: Yes — but the calculus has shifted materially, and the honest answer is more complicated than either "AI will replace programmers" or "nothing has changed."