British self-driving startup Wayve has raised $1.2 billion in a Series D round led by SoftBank, with participation from NVIDIA, Uber and three major automakers — one of the largest single financings yet for end-to-end learned driving, and another nine-figure validation of the "embodied AI" thesis.
End-to-End Learning Over Hand-Built Maps
Founded in Cambridge, UK in 2017 by CEO Alex Kendall, Wayve has spent years arguing that autonomous driving's bottleneck is not sensors or rules but learned intelligence. Its Embodied AI platform is hardware-agnostic and map-free: instead of relying on high-definition maps and hand-engineered stacks, end-to-end deep learning models learn driving behavior from fleet data and generalize to new environments — an approach philosophically closer to Tesla's than to the robotaxi incumbents, but packaged as a licensable product for automakers rather than a vertically integrated fleet.
The company's research output includes LINGO, a language-grounded driving model, and GAIA, a generative world model for driving simulation — both early examples of the world-model techniques now converging with the robot learning stack used in humanoids.
Cap Table and Scale
The Series D takes Wayve's total funding to roughly $2.8 billion, per StockAnalysis data. It follows the $1.05 billion Series C in 2024 — also led by SoftBank, with NVIDIA and Microsoft — which was then the largest AI fundraise in UK history. Uber's participation adds a distribution dimension: the ride-hailing platform has been assembling autonomous partnerships across multiple suppliers, while NVIDIA's check extends its now-ubiquitous position across the physical AI value chain, from GPUs to reference platforms.
Wayve has said its platform ambitions extend beyond passenger cars: the same map-free, hardware-agnostic driving intelligence is designed to transfer to other vehicle form factors and, eventually, to robotics companies building machines that move through the physical world.
Why It Matters for Embodied AI
The round lands in a crowded week for physical AI capital — with robot foundation model startups and humanoid makers also raising aggressively — and underscores a structural point: investors now treat learned physical intelligence as a general-purpose asset class, whether the embodiment is a car, a mobile manipulator or a humanoid. Wayve's bet is that a driving brain trained end-to-end at fleet scale becomes a foundation for far broader embodied applications.
