HomeFull speed ahead. Keep track of advancements in every frontier

TheTechno Optimist.

Wayve AI DriverWayve—Wayve AI DriverWayve—Wayve—Wayve AI DriverWayve—
Wayve AI Driver
Photo: Department for Digital, Culture, Media and Sport · Public domain (opens commons.wikimedia.org)

Wayve AI Driver

by WayveGB

PilotStage 3 of 5

London robotaxi rides with Uber since Sep 2026; Nissan driver assist in production cars from FY2027.

Updated 3 Sep 2026Checked 25 Sep0 updates this week

Milestones

Next · Nissan ProPILOT with Wayve in production cars
  1. Production partnership with Nissan2025Complete.
  2. Robotaxi service with Uber in LondonSep 2026Complete.
  3. Nissan ProPILOT with Wayve in production carsTarget 2027Not yet reached.

Most important updates

  • 3 Sep 2026
  • 21 May 2026
  • Feb 2026
  • 10 Jun 2025
  • 9 Apr 2025

Upcoming

  1. Q4 2026Robotaxi pilot in Tokyo with Nissan and Uber, aimed for late 2026 (next)
  2. 2027Nissan ProPILOT with Wayve in production cars
  3. 2028Stellantis to ship Wayve-based driver assistance in US vehicles

Current obstacles

  • Proving a black-box driver safeOne big AI model is hard to inspect, so proving it safe needs lots of testing, including AI-generated simulations.

Physics limits

  • Learned drivers can't be fully verifiedA neural network's behaviour in situations it never saw can't be proven in advance. Safety must be shown with testing and statistics, not formal guarantees.
  • Each new city is new dataDriving models generalise imperfectly: road rules, signs and local habits differ, so each city needs fresh data and validation before cars can drive alone.
  • Proving safety takes vast mileageUS drivers average about one fatal crash per 100 million miles. Showing a robot driver is safer with statistical confidence takes hundreds of millions of driverless miles (RAND estimated ~275M).

How it works

4 parts
A Wayve test car in London: the slim roof rack holds the cameras whose images one neural network turns into driving
A Wayve test car in London: the slim roof rack holds the cameras whose images one neural network turns into drivingCourtesy of Source: Wayve · Press kit, editorial use (opens wayve.ai)
End-to-end

One model drives

A single neural network learns to turn camera (and optional radar) input directly into driving, without hand-coded rules.

Mapless

No HD maps

It drives with ordinary navigation maps, so a new city doesn't have to be surveyed in detail first.

Portable

Same driver, many cars

The model is adapted to different carmakers' sensors and computers, so one driver can ship in Nissans and in robotaxis.

Simulation

Generated test worlds

Wayve's GAIA generative models create realistic driving video to test rare situations the fleet hasn't recorded.

Update log

5 updates

Thu 3 Sep

  • Major: BlogCommercial

Thu 21 May

  • Major: BlogPartnership

Feb 2026

  • Minor: PressPartnership

Tue 10 Jun 2025

  • Minor: PressPartnership

Wed 9 Apr 2025

  • Major: PressPartnership

About Wayve

The team behind Wayve AI Driver

Wayve

End-to-end AI driver, Nissan, Uber

London-based Wayve builds an 'AI Driver' that learns to drive from data and works across car models and cities without HD maps. It licenses it to carmakers for driver assist and to Uber for robotaxis.

  • Raised $1.2B (up to $1.5B) Series D at $8.6B in February 2026.
  • AI Driver to ship in Nissan ProPILOT from fiscal 2027.
  • Robotaxi service with Uber launched in London in September 2026; Tokyo with Nissan planned.

wayve.ai (opens wayve.ai)Company profile →

Founded
20179 yrs
Headquarters
United Kingdom
Status
Private
Valuation
$8.6BprivateFeb 2026
Raised
$2.3B2 rounds
Last round
Series D · $1.2BFeb 2026$8.6B post
Works in
VehiclesSelf-driving softwareAI
Coverage
1 program · 7 updateslatest 3 Sep 2026checked 25 Sep
Lead investors
EclipseBaldertonSoftBank Vision Fund 2SoftBank
Customers
Nissan (opens nissan-global.com)
Partners
Uber (opens uber.com)
Investors
NVIDIAMicrosoft (opens microsoft.com)
People
Alex KendallCo-founder and CEO
Primer