AIToday
ITmedia AI+Published: Oct 20, 2026, 10:01 JST

Turing taps Google, Nvidia chips for 3x faster self-driving AI

Turing taps Google, Nvidia chips for 3x faster self-driving AI

3 Key Points

  1. What happened

    Turing CTO Yu Yamaguchi said adding data and GPUs alone did not speed development, so the Tokyo startup rebuilt its compute base around Nvidia and Google chips, claiming roughly 3× speed at the same price.

  2. Why it matters

    Turing has already completed test runs in five Japanese cities including Tokyo and Osaka, and faster training may shorten the path toward the level 5 full autonomy it is targeting, which no one has achieved yet.

WHO IT HITSAutonomous-driving engineering teams building their own training infrastructure are the clearest audience: the claim is that data volume and GPU count alone hit a wall, and chip choice alongside the compute setup mattered more.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

Turing's approach diverges from the usual sensor-heavy recipe. While the industry generally stacks LiDAR, radar and multiple sensors to ensure safety, Turing argues that pushing AI performance hard enough lets a vehicle drive smoothly in city streets with fewer, less capable sensors. Its modified Toyota Alphard carries just seven cameras plus motion sensors, a satellite positioning receiver and an onboard computer — a deliberate bet that software, not hardware, is the bottleneck.

The Osaka run is presented as evidence for that bet. Yamaguchi notes the AI had learned almost entirely from Tokyo data and was driving an unfamiliar road for the first time, yet performed at that level. That matters because generalizing to unseen roads is where camera-only approaches are usually challenged.

Behind the AI is a loop of collecting driving data and retraining on it, and Turing says scaling data and GPU count alone did not accelerate that loop. The rebuild of its compute foundation, and the comparison of Nvidia and Google chips at the same price, is where the 3× claim comes from. The details were shared in a session at Google Cloud Next Tokyo 26, an event hosted by Google Cloud Japan.

FAQ
What kind of cars does Turing use?
It modifies a commercially sold Toyota Alphard minivan, fitting seven cameras — two in front, four on the sides and one at the rear — plus motion sensors, a satellite positioning receiver and an onboard computer.
What did Turing's AI do in Osaka?
It drove an area of central Osaka with no human touching the steering wheel, accelerator or brake. According to Yamaguchi, it was the AI's first time on Osaka roads and it had learned mostly from Tokyo data.
What is Turing's stated mission?
The company, founded in 2021 and based in Ota, Tokyo, has the mission 'We Overtake Tesla' and aims to achieve level 5 full autonomous driving, which it says no one in the world has accomplished yet.

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Next articleHPE launches ProLiant Gen13 servers on 6th Gen AMD EPYC