Rivian announced the acquisition of Third Dimension AI on October 22, 2026, bringing a specialized 3D simulation platform into its autonomous driving operation. The technology will help the American EV manufacturer recreate difficult traffic situations using real-world driving data, supporting the development of both consumer driver-assistance features and future robotaxis.
The financial terms of the acquisition were not disclosed. Third Dimension AI, founded in 2024 by Tolga Kart and Piotr Sokolski, specializes in converting real-world information into detailed, controllable three-dimensional environments.
Rather than relying exclusively on additional miles of road testing to encounter unusual situations, Rivian wants to reproduce those events virtually and explore how its driving software responds when conditions change. The acquisition adds both the underlying technology and its developers to Rivian’s autonomy research and engineering organization.
Turning Real Driving Footage Into Controllable 3D Environments
Third Dimension AI’s platform is designed to reconstruct realistic environments from real-world data. Engineers can then manipulate those environments to create variations of an original driving scenario.

For example, a recorded intersection could become the basis for multiple simulated tests involving different traffic movements, visibility conditions, or unexpected obstacles. Instead of treating each recording as a single fixed event, developers can use the reconstructed environment to examine how automated driving software responds to different circumstances.
This addresses a persistent difficulty in autonomous vehicle development: the most challenging situations are often too uncommon to appear frequently in ordinary driving data. A system might encounter thousands of routine intersections before experiencing an unusual combination of road users, obstacles, and weather conditions.
Rivian says the acquisition will accelerate synthetic data generation and full-world simulation, allowing its engineers to recreate and vary complex driving scenarios at scale.
Rivian Collects More Than 400,000 Driving Sequences Daily
Rivian already gathers substantial quantities of driving information from vehicles whose owners have enabled external camera data sharing. The company uses aggregated and anonymized information to support development of its automated driving systems.
James Philbin, Rivian’s senior vice president of autonomy and AI, told TechCrunch in an exclusive interview that the manufacturer’s connected fleet uploads more than 400,000 driving sequences per day.
The challenge is extracting useful training and validation scenarios from that information. Routine driving footage can help characterize everyday road conditions, but collecting enough examples of rare and dangerous situations is considerably harder.
Third Dimension AI’s technology gives Rivian another way to use the information it already receives. A recorded event can provide the foundation for multiple virtual scenarios, potentially increasing the value of individual driving sequences without requiring every variation to occur naturally on public roads.
A Simulation Team With Autonomous Driving Experience
According to TechCrunch’s original reporting, Third Dimension AI has a 15-person team, with most employees based in the United Kingdom. Its personnel will join Rivian’s autonomy and AI organization, while the acquisition also creates an opportunity to expand Rivian’s existing AI presence in London.
The startup’s founders bring experience spanning simulation technology, software development, and autonomous vehicles. Before establishing Third Dimension AI, Tolga Kart worked on Tesla’s Autopilot simulation efforts and held engineering positions in the gaming and synthetic-data industries. Cofounder Piotr Sokolski previously worked on neural simulation technology at autonomous driving developer Wayve.
Those backgrounds are particularly relevant because creating convincing virtual environments is only part of the challenge. For autonomous driving development, simulated vehicles, road users, and physical interactions must behave realistically enough to produce meaningful training and testing results.
Rivian has not disclosed how much it paid for the company or provided a timetable for integrating the simulation platform into its development workflow.
The Acquisition Supports Rivian’s Consumer and Robotaxi Programs
Rivian intends to apply the technology across two related areas: driver-assistance systems for privately owned vehicles and autonomous driving technology for commercial robotaxi operations.
Its existing Universal Hands-Free system allows drivers to operate without holding the steering wheel under supported conditions, but they must continue monitoring the road and remain responsible for driving.
The manufacturer is also developing more capable point-to-point driving software intended to handle a broader range of road environments while remaining under driver supervision. Simulated testing could help engineers evaluate how that software handles complicated intersections, unusual traffic behavior, and changing environmental conditions.
Beyond consumer vehicles, Rivian plans to develop autonomous versions of its R2 SUV for deployment through Uber. The companies are targeting robotaxi operations beginning in 2028, making the reliability and validation of Rivian’s automated driving technology important to its commercial ambitions.
The Third Dimension acquisition does not establish that Rivian has achieved unsupervised driving capability or completed the safety validation required for driverless commercial operation. It strengthens the development infrastructure intended to support those goals.
Why Realistic Simulation Matters for Self-Driving Safety
Simulation offers an important advantage over collecting road mileage alone: developers can repeat difficult situations under controlled conditions. They can modify one variable at a time, compare software responses, and examine failures without exposing other road users to the same experimental risks.
However, simulated performance is only as useful as the accuracy of the underlying virtual environment. Differences between simulated behavior and actual road conditions can produce misleading results, particularly when software must interpret unusual objects or predict unpredictable human actions.
Philbin emphasized this challenge in his TechCrunch interview, explaining that developers need confidence that simulated testing accurately represents real-world performance before using it to validate software for release.
That distinction matters for Rivian owners. Acquiring a more advanced simulation platform does not automatically make an existing vehicle capable of autonomous driving, nor does it change the supervision requirements of currently available driver-assistance features. The immediate development benefit is the ability to train and evaluate software against a wider variety of reconstructed driving situations before those systems are deployed on public roads.
