Huang Renxun turns Japan into an NVIDIA “physical AI” hub: a life-saving favor from 30 years ago, fully bound together end-to-end 30 years later

Author: Wall Street Insights

NVIDIA is turning Japan into a core pivot for its global physical AI roadmap.

This week, NVIDIA CEO Jensen Huang appeared in Tokyo, kicking off a flurry of cooperation signings and business meetings. Reports say that on Thursday, NVIDIA announced partnerships with Japanese robotics giants FANUC and Yaskawa Electric to jointly advance the development of robotics and AI technologies.

At the same time, NVIDIA also announced an expansion of its partnership with Toyota, covering multiple areas including autonomous driving, factory simulation, and city intelligence. At a Tokyo media event, Huang said, “With AI, robots will become intelligent, easy to adapt to, and within reach.”

The strategic intent behind this Japan trip is clear: NVIDIA is systematically binding Japan’s manufacturing base, semiconductor supply chain, and its end-to-end AI full-stack technology.

In an interview, Huang pushed back on public concerns about an AI investment bubble, saying, “We are still a long way from an AI bubble—demand is extremely strong,” and emphasizing, “We need to build infrastructure for at least ten years.” He also said NVIDIA will announce later this week cooperation plans related to Japan’s sovereign AI.

Also worth noting is that this visit came with a highly historic reunion—Huang shared the stage again with Shoichiro Irimajiri, former president of Sega, who saved him about 30 years ago when NVIDIA was on the verge of bankruptcy. The emotional depth added another layer to this business trip.

(Video screenshot: Huang and former Sega president hugging)

Reunion in Akihabara: the “life-saving favor” from 30 years ago

Outside the business meetings, the most emotionally charged moment of Huang’s trip took place at the former Sega game center site in Tokyo’s Akihabara.

According to Sega’s disclosure, on July 15 in local time, Huang attended an event hosted by Sega (SEGA) and shared the stage again with former Sega president Shoichiro Irimajiri after many years. Huang said at the event, “If it weren’t for everything Sega did—if it weren’t for everything Irimajiri did, NVIDIA wouldn’t be alive today.”

The connection dates back to around 1996. At that time, NVIDIA, newly established, was developing graphics chips for Sega’s next-generation console. But due to a mistaken bet on the technical route, the project completely failed, and the company was close to bankruptcy.

Huang took the initiative to admit the failure to Irimajiri, then Sega’s vice president. Irimajiri did not choose to pursue accountability—instead, he pushed Sega to invest about $5 million into this “startup that had run out of food,” according to Huang. Huang recalled:

“I told Irimajiri: if they put this money into us, they would likely get nothing back; but if they didn’t, we would go under… After thinking for a few days, he told me: we’ll invest.”

With this funding, after laying off 60% of its workforce, NVIDIA rebuilt and in 1997 launched RIVA 128. Then, with products like RIVA TNT and GeForce 256, it established its market position in the GPU field.

The twist is that after NVIDIA went public in 1999 at an approximate valuation of about $300 million, Sega quickly chose to cash out, selling its shares for roughly $15 million. Today, NVIDIA’s market cap has exceeded $50k.

For this reunion, both parties simultaneously announced they would extend their collaboration. Sega’s future works will support NVIDIA’s newly launched RTX Spark platform, including the upcoming VIRTUA FIGHTER CROSSROADS. NVIDIA and Sega’s partnership began 30 years ago, when NVIDIA’s NV1 chip provided graphics support for the PC version of the first Virtua Fighter—one of the earliest 3D fighting games in the world.

(Image source: NVIDIA official website)

Izakaya banquet: currying favor with Japan’s core semiconductor supply chain

On the night before the official cooperation announcement, Huang completed a collective “PR” effort targeting key nodes in Japan’s AI supply chain with a low-key yet meaningful dinner.

According to Singapore’s The Straits Times, on the evening of July 15 local time, Huang appeared at an izakaya in Tokyo’s Kanda district and had dinner for about two hours with a group of top executives from key Japanese supply-chain companies.

(Image source: The Straits Times)

Attendees included:

Kioxia’s CEO, a leading advanced flash memory chip maker; Shin-Etsu Chemical’s leadership, one of the world’s leading silicon wafer suppliers; Tokyo Electron’s chief; leadership from Ajinomoto, a unique supplier of advanced chip packaging film; and executives from Sumitomo Electric Industries, a fiber optic cable manufacturer, and Taiyo Yuden, an advanced capacitor producer. Yuki Kusumi from Panasonic Holdings was also there.

It is said Huang’s group ate yakitori and goryōke (beef offal hot pot) and drank Japanese whisky.

The lineup of this gathering nearly sketches the full picture of Japan’s hardware supply chain that NVIDIA’s next-generation AI system depends on. According to those present, conversations included remarks like “Let’s work together to promote the prosperity of industries like semiconductors, so that stock prices keep rising.”

Outside the izakaya, a group of people holding smartphones gathered, hoping to catch a glimpse of the AI-era idol known on social media as “kawajan-san” (the leather jacket man). Reportedly, a 57-year-old Taiwanese Chinese tourist, Chang Hui-Yu, said outside the Sega event site, “I think he’s the most influential person on Earth.”

Betting on physical AI: Japan’s manufacturing industry is a “natural ally”

The core strategic narrative of Huang’s trip is to position Japan as a key battleground for the global development of physical AI.

According to Kyodo News, in an interview in Tokyo, Huang said, “This is a historic moment for Japan, because Japan has long been outstanding in precision manufacturing and large-scale manufacturing.”

He believes AI can help Japan address severe labor shortages. “Through automation, AI, and robotics technology, we can enhance the capabilities of the existing workforce and improve the country’s overall productivity.”

On specific cooperation, NVIDIA’s partnership with FANUC and Yaskawa Electric directly targets intelligent upgrades of industrial robotics. The Toyota cooperation is even broader:

Toyota is developing next-generation vehicles with L2++ functionality based on NVIDIA’s DRIVE AGX platform and DriveOS operating system. At the same time, it uses NVIDIA’s Megatron-LM to train code-assistant AI models that comply with MISRA standards, accelerating vehicle software engineering. It also uses NVIDIA’s Omniverse and Isaac Sim frameworks to advance factory digital twins and robot simulation.

Toyota subsidiary Woven by Toyota is also developing a multimodal visual-language model for urban transportation intelligence based on NVIDIA H100 GPUs.

NVIDIA Vice President Rishi Dhall said, “Physical AI will bring intelligence to every moving machine—from cars, to robots, to trucks, and to the cities and factories they operate in.”

Full-stack layout: from healthcare and finance to quantum computing

Beyond robots and cars, NVIDIA’s cooperation footprint in Japan extends into multiple key industries, showing a systematic pattern of full-stack penetration.

In healthcare and life sciences, multiple Japanese pharmaceutical giants are using NVIDIA’s BioNeMo platform to accelerate AI drug discovery, including Eisai, Astellas, Daiichi Sankyo, and Ono Pharmaceuticals. Canon has rolled out Japan’s first NVIDIA-accelerated photon-counting CT system, while Fujifilm has commercialized Japan’s first whole-body CT system equipped with NVIDIA Blackwell. Kawasaki Heavy Industries plans to use NVIDIA Holoscan IGX, Isaac GR00T, and the Cosmos platform to develop surgical-assistance robots and hospital transport robots.

In the financial sector, Mizuho Bank plans to build the largest local AI factory in Japan’s financial industry, starting with NVIDIA DGX B200 systems. The Japanese Research Institute (JRI) under Sumitomo Mitsui Financial Group has already deployed an AI factory, converting financial data into intelligence based on NVIDIA Nemotron open models. Rakuten Bank will use NVIDIA Agent Toolkit to develop transaction foundation models.

In quantum computing, two supercomputers driven by NVIDIA GB200 at Japan’s RIKEN have begun operation: RIKYU has deployed 1,600 NVIDIA Blackwell GPUs to support development of open foundation models. The quantum-HPC system ROQUO integrates 540 Blackwell GPUs, tightly linking with the quantum computers on the RIKEN campus. Mitsubishi Chemical, Mizuho Bank, Keio University, AIST, and other institutions are working with NVIDIA to achieve 13.4x acceleration compared with pure CPU nodes in molecular spectroscopy analysis workflows.

Additionally, according to reports, market speculation suggests NVIDIA may announce cooperation with Japan’s “physical AI model national team,” Noetra. Noetra was set up with SoftBank at the lead, bringing together 44 Japanese companies including Honda and NEC. The Japanese government also provides $1 trillion yen in fiscal subsidies for it.