Turing raises ¥15.3B for autonomous driving

What the business is

A Japanese startup building a fully autonomous driving system: uses generative AI to turn the entire process from camera perception to vehicle control into one end-to-end model, first enters the existing market with enhanced ADAS, and ultimately targets L5 fully autonomous driving.

Starting capital：Cumulative funding since establishment is about ¥20.8 billion (208 hundred million yen): Pre-A round total ¥5.558 billion (55.58 hundred million yen) in 3 tranches in 2024; Series A 1st close is ¥15.27 billion (152.7 hundred million yen) in 2025-11, comprising ¥9.77 billion (97.7 hundred million yen) equity plus ¥5.5 billion (55 hundred million yen) syndicated loan.

How it started

In 2017, the shogi AI Ponanza developed by Kazunari Yamamoto (山本一成) defeated the sitting Meijin for the first time, convincing him that 'once AI starts self-learning, it will surpass humans exponentially.' In 2021-08, he founded Turing Inc. (チューリング株式会社), setting the mission 'We Overtake Tesla' because rule-based autonomous driving will eventually hit a wall—humans cannot write complex driving behaviors as rules, while E2E (end-to-end) lets AI learn to drive directly from camera data, which is the path to L5.

What happened

From 2024-04, the Pre-A raised ¥5.558 billion (55.58 hundred million yen) in 3 tranches, led by ANRI, to invest in the multimodal generative AI 'Heron' and the large-scale GPU cluster 'Gaggle Cluster', and launched the 'Tokyo30' project to drive 30 minutes in Tokyo without human intervention in 2025; in 2025-11, it completed a Series A 1st close totaling ¥15.27 billion (152.7 hundred million yen), co-led by JIC Venture Growth Investments (JIC VGI) and Global Brain, with participation from Denso (電装), GMO Internet, and others; valuation was about ¥60 billion (600 hundred million yen), about 4 times the previous round; it is cooperating with Denso and negotiating with multiple automakers, targeting commercialization in 2030.

How it ended up

Still ongoing: as of 2026-09-01, not yet commercialized; it has completed L2-level testing on public roads in Tokyo and is expanding to other regions; it was selected for Japan's national domestic AI project; the CEO said 'the hard part is the technology, not the business model,' and he also acknowledged that overseas competitors remain ahead.

What can be applied

Bet on transferable beliefs like 'AI can self-learn', not specific products. Self-learning burns cash; fundraising is critical. Confirm capital efficiency before choosing high-ceiling problems.