Jensen Huang has just invested in Robotaxi, the "next trillion-dollar opportunity".

Recently, Huang (Jensen Huang), who has been generously investing, has just invested in another self-driving car company, injecting a staggering $500 million (approximately 3.56 billion yuan).

This company is none other than the UK startup Wayve, which secured the largest autonomous driving financing at that time last year.

Moreover, financial support alone isn't enough. After a test drive on the streets of London, Huang highly praised Wayve:

Wayve is going to be the next trillion-dollar company.

Considering the recent trends, Huang's investment in autonomous driving companies actually hides deeper intentions.

All signs indicate that NVIDIA is also getting involved in the Robotaxi business.

Huang Bets 3.6 Billion on a Self-driving Car Company

After making frequent moves in the capital market, NVIDIA has recently set its sights on a new target - a UK AI startup.

Jensen Huang promised to invest £2 billion (approximately $2.7 billion, 19.2 billion yuan) in UK AI startups. About one-fifth of this capital will flow into the autonomous driving company Wayve.

According to Wayve, the company has signed a letter of intent with NVIDIA, and NVIDIA plans to invest $500 million (about 3.56 billion yuan) in its next round of financing.

Other investors joining the investment include venture capital firms Accel, Air Street Capital, Balderton, Hoxton Ventures, and Phoenix Court.

Actually, this isn't NVIDIA's first bet on Wayve. In May last year, NVIDIA participated in Wayve's Series C financing of approximately $1.05 billion (about 7.47 billion yuan).

Other investors at that time included capital giants like SoftBank and Microsoft, jointly making it the largest financing in the global autonomous driving field in 2024.

Additionally, NVIDIA and Wayve have maintained a cooperative relationship since 2018.

Wayve's second-generation autonomous driving platform, integrated into the Ford Mach E test vehicle, uses NVIDIA's GPU. Its newly released third-generation autonomous driving platform is also equipped with NVIDIA Thor's development kit, expected to reach L3/L4 level capabilities.

△

According to foreign media TechCrunch, Wayve will soon release a video of Huang sitting in a Wayve vehicle during a test drive in central London.

Not only did Huang highly praise Wayve in the video, but he also straightforwardly said:

Wayve is going to be the next trillion-dollar company.

So, what makes Wayve so special that it has won the favor of NVIDIA and other capital investors?

What Makes Wayve Attractive?

Wayve was founded in 2017 and is headquartered in London, UK.

Its two co-founders, Alex Kendall and Amar Shah, are both Ph.D. holders in computer vision and robotics from the University of Cambridge, UK.

△

In June this year, Wayve's core team welcomed three new senior executives - Rob Flenniken, Uri Wolfovitz, and Dennis Jackson.

Notably, before leaving their previous positions, Rob and Dennis worked for another fallen autonomous driving giant Cruise, while Uri was the former vice president of performance optimization and optical path at Mobileye.

Regarding Wayve's origin, current CEO Kendall recalled that while studying at the University of Cambridge, he firmly believed that the best way to solve the problems of autonomous driving was to enable AI to independently learn the rules and patterns of human driving behavior.

His main research focused on how to use end-to-end deep learning to achieve safe and real-time scene understanding, laying the technical foundation for Wayve.

However, the technical route proposed by Kendall seemed rather out of place at that time.

Back then, the industry trend was to set rules for autonomous vehicles to handle various special scenarios. End-to-end deep learning was even regarded as "wishful thinking".

From a technical perspective, the end-to-end model may be more difficult to debug and verify due to its "black-box" nature.

However, Kendall and Shah believed that systems relying on handwritten rules had limitations. Eventually, they decided to establish Wayve in a residential building near the University of Cambridge, UK.

Wayve's core technical route AV2.0 is similar to Tesla's FSD. It doesn't rely on high-precision maps, preset rules, or multiple sensors (such as lidar). Instead, it uses "end-to-end deep learning" and "Embodied AI".

Its core idea is to enable AI to learn and adapt from the actual driving environment through in-vehicle cameras, just like humans.

Wayve's system directly processes the raw image data from the cameras and then outputs control instructions, such as changing the steering wheel angle, throttle, and brake intensity.

To address the "black-box" decision-making problem, Wayve has specifically developed the visual language model LINGO, which has now evolved to its second generation, LINGO - 2, as shown in the following image:

△

Performing driving actions while explaining driving behavior in natural language - this is exactly the mainstream VLA this year.

Wayve has also launched the controllable multi-view generative world model GAIA - 2, which can simulate and generate video data of various extreme driving scenarios.

Moreover, GAIA - 2 supports multiple inference modes, including generation from scratch, context-based prediction, and selective content editing.

△

Theoretically, this AI system that doesn't require pre - built, centimeter - level high - precision maps can quickly adapt to new cities and traffic rules by learning local driving data. It has already been tested in the UK, the US, Germany, and Japan.

Furthermore, Wayve's system is compatible with different types of sensor suites and isn't limited to specific vehicles, ranging from passenger cars to trucks.

This flexible combination method that automakers can choose according to their needs may be more appealing to car companies.

At least for now, Wayve has secured a major Japanese client - Nissan. In April this year, an agreement was reached. Nissan will launch the next - generation ProPILOT technology equipped with Wayve software in the fiscal year 2027.

The ride - hailing giant Uber, which has been actively involved in the Robotaxi field, has also seized the opportunity to cooperate with Wayve and intends to pilot an L4 fleet in the UK.

Now, the autonomous driving field is surging with opportunities. In this market that could potentially generate trillions of dollars, even NVIDIA can't stay on the sidelines.

One More Thing

According to 36Kr, NVIDIA announced at an all - hands meeting that it has internally initiated the incubation of a Robotaxi project.

Its technical route is also similar to Tesla's FSD and Wayve's, using an "end - to - end" neural network and strengthening the neural network through a world model.

If NVIDIA really enters the game, the Robotaxi battlefield may welcome its most significant player ever.

This article is from the WeChat official account "Intelligent Vehicle Reference" (ID: AI4Auto), written by Jessica and published by 36Kr with authorization.