Waymo rides surge; AVs hitting massive RAM needs

Waymo’s rollout tracked city-by-city lift: the company reported about 150,000 weekly trips in its December 2024 year‑in‑review post, then publicly reported 250,000 weekly paid rides in April 2025 as it opened Austin and expanded in the Bay Area, and investor documents later showed roughly 450,000 weekly paid rides by December 2025. (waymo.com CNBC CNBC ) Waymo has publicly stated expansions into multiple U.S. metro regions — Phoenix, San Francisco, Los Angeles and Austin in operation, with launches or sign‑ups announced for Atlanta, Miami, Washington, D.C., and Dallas — as part of a strategy that the company says has produced millions of autonomous trips to date. (waymo.com The Robot Report ) Investor and media reporting shows the recent ridership figures came into public view via investor letters and earnings coverage — CNBC cited a Tiger Global letter for the December 2025 weekly figure, illustrating that some of the most current operational metrics are being disclosed to investors before wider press releases. (CNBC ) Micron’s leadership framed future automotive demand for DRAM during its March 18, 2026 earnings call, with CEO Sanjay Mehrotra linking Level‑4 vehicle platforms and advanced robots to a multi‑hundred‑gigabyte memory BOM per vehicle as part of long‑term market growth for memory suppliers. (Motley Fool The Register ) Market reporting summarized Micron’s point that today’s typical ADAS memory footprints (often on the order of tens of gigabytes) would need to scale by an order of magnitude for geofenced L4 stacks and advanced onboard models, creating a new edge‑compute segment competing with data centers for high‑performance DRAM. (Autotech.news TechSpot ) Hardware vendors are releasing centralized vehicle compute platforms that target those same demands: NVIDIA’s DRIVE Thor positions a unified AV/IVI/parking/monitoring computer with up to ~2,000 teraflops of performance to consolidate workloads, while competitors like Qualcomm are pitching scalable Snapdragon Ride solutions for centralized, software‑updateable vehicle architectures. (NVIDIA developer blog TS2.Tech comparison ) Supply and product signals from both memory manufacturers and SoC vendors imply concrete engineering consequences for vehicles: OEMs and Tier‑1s are already evaluating centralized ECUs, higher‑capacity DRAM modules, and automotive‑grade thermal solutions to house larger memory footprints and denser accelerators inside constrained vehicle envelopes. (IoT‑Automotive.News Autotech.news )