Nvidia Corp. may invest $500 million in Wayve Technologies Ltd., a British startup that develops autonomous driving software optimized for its chips.

The companies detailed the potential cash infusion on Thursday. A few hours earlier, CNBC reported that Nvidia had recruited key employees of Enfabrica Inc., a data center hardware startup. The chipmaker reportedly made the hires as part of a transaction worth more than $900 million.

Wayve develops a software platform called the AI Driver that can equip vehicles with autonomous driving capabilities. Autonomous vehicles often rely on several different artificial intelligence models to navigate. The AI Driver, in contrast, uses a single neural network to process data from a car’s sensors, combine them with GPS measurements and generate driving decisions.

Wayve is currently working on a new version of the software optimized to run on Nvidia’s AGX Thor system-on-chip. The chip includes a graphics card, a computer vision accelerator and a central processing unit. Wayve says that the upcoming AI Driver release will provide Level 4 autonomy, which means it will be capable of operating without driver input under certain conditions.

Nvidia has signed a letter of intent with the company to “evaluate a $500 million strategic investment.” The capital would be provided as part of Wayve’s next funding round. Nvidia participated in the company’s previous raise, a $1.05 billion Series C round that closed last year.

The funding discussions were announced the same day the chipmaker pledged to invest £2 billion, or about $2.7 billion, in the U.K.’s AI startup ecosystem. The initiative is supported by a group of venture capital firms. It’s intended to provide local AI companies with capital and data center infrastructure.

Also on Thursday, CNBC reported that Nvidia had hired Enfabrica Chief Executive Officer Rochan Sankar and several other employees of the hardware startup. In addition, the chipmaker inked a deal to license its technology. Enfabrica has developed two hardware products called ACF-S and Emfasys that are both designed for use in AI clusters. 

ACF-S is a network chip that enables graphics cards to exchange data with one another. Traffic usually has to go through several network devices before it reaches its destination. Enfabrica says that the ACF-2 reduces the number of devices through which data must travel, which lowers latency and reduces the risk of network failures.

The company’s other product, EMFASYS, is a memory appliance. Graphics cards can use it to store the information processed by AI models. The servers that host those graphics cards can connect to EMFASYS via the ACF-S network chip.

Nvidia reportedly paid more than $900 million in cash and stock to Enfabrica as part of the deal. The transaction closed last week. 

Photo: Wayve