# Wayve bets its driver on rented Blackwell muscle

_Wayve will train its AI Driver on CoreWeave Cloud with Blackwell GPUs in a multi-year deal._

**Published:** 2026-11-10
**Source:** https://www.startuphub.ai/ai-news/artificial-intelligence/2026/wayve-bets-its-driver-on-rented-blackwell-muscle

---

Why would Wayve hand its AI Driver training to CoreWeave? It wants dedicated time on NVIDIA Blackwell GPUs to turn real-world driving data from more than 70 countries into commercial driver software, CoreWeave says.

[CoreWeave](https://www.coreweave.com/news/coreweave-to-power-wayves-ai-driver-training-in-multi-year-agreement) says the multi-year agreement covers training, evaluation and simulation for Wayve's end-to-end foundation models that learn to drive directly from raw driving data rather than hand-written rules. Wayve started on Weights & Biases for experiment tracking and now extends development and training onto CoreWeave Cloud, the company said. The shift moves heavy training off Wayve's own clusters and onto rental infrastructure built for long AI runs.

Training that loop is simple to sketch and hard to keep running.

Silvius Rus, Wayve's SVP of engineering, described the test that matters: "Our AI Driver gets better by running this data flywheel: gathering real-world driving experience, training on it, evaluating it, and deploying only what outperforms what is already on the road." Gavin Jackson, managing director for EMEA at CoreWeave, said physical AI needs training, evaluation and simulation designed to work together from the start, pointing to [record MLPerf results](https://www.startuphub.ai/ai-news/ai/2026/coreweave-shatters-mlperf-records) and top SemiAnalysis ClusterMAX and Artificial Analysis rankings, according to CoreWeave. It is the reliability claim that counts here, because an interrupted run wastes weeks of compute and muddies safety validation.

For Wayve's autonomy engineers, the change is practical. They do not provision clusters. They submit larger training and simulation jobs and must prove each model version beats the last before it ships to automakers and fleet owners, from personal cars to robotaxis. CoreWeave did not disclose the contract value, duration in years, or number of GPUs, so the budget impact stays opaque for now.

Watch deployment pace and safety reports. If CoreWeave shortens the train-evaluate-deploy cycle without breaking reproducibility, Wayve can push its mapless, hardware-agnostic driver that runs without prebuilt maps and on different vehicle hardware to more partners faster.

---

Original analysis from [startuphub.ai](https://www.startuphub.ai), the #1 AI startup directory.

© StartupHub.ai. All rights reserved. You may not republish this article in full without a license. Search engines and AI research tools may crawl and summarize for reference. Bulk reproduction or model training on this content requires a license. See https://www.startuphub.ai/terms.