Wayve trains camera-first self-driving AI on Azure to speed global rollout
The UK autonomous driving start-up says treating self-driving as a data problem, rather than a hardware one, allows its system to adapt rapidly across vehicles and markets.
Wayve is training its artificial intelligence Driver on Microsoft Azure as it develops a camera-first approach to self-driving that it says can be deployed quickly across different vehicle models and geographies.
The company said it uses a combination of Azure Storage, Azure Databricks, Azure AI infrastructure and Azure Kubernetes Service to link thousands of graphics processing units into what it described as a flexible supercomputer for model training and validation. The goal is to support end-to-end deep learning for driving, rather than relying on rule-based systems or heavy sensor stacks.
“We’re really approaching autonomous driving as an AI problem and building a data-driven stack with end-to-end deep learning,” said Alex Kendall, co-founder and chief executive of Wayve.
Wayve’s system centres on a single computer installed in the trunk of the vehicle and relies primarily on cameras to interpret road signs, traffic lights and the surrounding environment. By avoiding complex sensor fusion and expensive hardware, the company argues its software can generalise more effectively and be transferred between vehicles with fewer modifications.
Cars equipped with the Wayve stack are already operating in the UK, the US, Germany and Japan. Since its founding, the company has raised $1.3 billion, reflecting renewed investor interest in autonomous driving approaches that prioritise software scalability over bespoke vehicle platforms.
Engineers at Wayve said Azure’s scale has been critical to that strategy. Alex Persin, a principal engineer at the company, said improvements in Azure Kubernetes Service have reduced operational overhead. “One concrete example is AKS used to only support 1,000 nodes,” he said. “Now it supports 5,000 nodes, which reduced the need for us to run our own Kubernetes clusters.” Persin described large-scale pre-training as essential groundwork for autonomy, comparing it to years of human learning before driving independently.
Wayve has begun translating the Azure-backed training effort into commercial partnerships. In June, the company said it plans to run a limited passenger trial with Uber in London this year. It has also signed an agreement with Nissan to bring Wayve-equipped vehicles into mass production in the 2027 fiscal year.
The relationship with Microsoft has also deepened. In October 2025, Wayve and Microsoft announced an expanded agreement and signed a strategic framework that formalised Azure as the backbone for Wayve’s training and validation workflows.
As the autonomous driving sector recalibrates after years of missed timelines, Wayve is betting that cloud-scale training and camera-led generalisation offer a faster path to real-world deployment than building bespoke autonomy stacks for each vehicle and city.
The Recap
- Wayve uses Azure to train its deep-learning AI Driver.
- Wayve has raised $1.3 billion in funding so far.
- Limited Uber passenger trial planned to start in London this year.
