Japanese AI Unicorn Sets Sights on Real-World Impact

In a bold move to mend practical challenges using deep learning, Japan’s Preferred Networks is reshaping industries and reaching for the global stage. Founded on a mission to deploy AI in tackling real-world issues, the company has carved out innovative solutions in sectors as varied as trucking, healthcare, and robotics.

A Decade of Innovation

For over ten years, Preferred Networks has been quietly advancing AI technologies that address the everyday problems of industries facing rapid change. As chief research executive Daisuke Okanohara explained in a conversation with CNBC’s Managing Asia, the company’s journey began with a simple yet ambitious goal: to solve tangible issues through technology.

Strategic Partnerships and Investments

The company’s promising approach has not gone unnoticed. Notably, automotive giant Toyota has significantly backed Preferred Networks, first investing 1 billion yen in 2015 and later adding 10.5 billion yen ($95.4 million) in 2017. This financial muscle has catapulted the startup into the limelight as one of Japan’s rare AI unicorns, standing out amidst a startup ecosystem where companies valued over $1 billion constitute a mere 0.5% of the global total according to an International Monetary Fund report.

Tackling Industry Challenges

Preferred Networks isn’t stopping at automotive innovations. In collaboration with trading powerhouse Mitsui & Co., the company has embarked on a joint venture in the autonomous driving arena. This partnership has funded a venture—T2—designed to revolutionize the trucking industry. By addressing issues such as exhaustive working hours and a shrinking labor pool, the venture aims to create safer and more efficient operations.

Despite the fierce competition in the autonomous vehicle market, CEO Toru Nishikawa acknowledges the intricacies of applying AI to more traditional transport sectors. "If we were to tackle the problem of regular cars," Nishikawa remarked, "it would be extremely challenging given the crowded field of competitors."

The Road to Commercialization

Real-world applications of deep learning demand patience and precision. Unlike digital solutions that can quickly generate user results, technologies aimed at industries like materials discovery may take between three to five years to transition from research to market-ready products. Nishikawa detailed the process: initial joint research, evaluation of commercial viability, and finally, product development before launching services for customers.

Advancing AI Hardware

In an era when Nvidia’s generative AI strategy stands strong yet reveals limitations with its GPU architecture, Preferred Networks is betting on the future by developing advanced processors. This initiative is fueled by a vision to capture a competitive edge in the burgeoning chip industry, a sector that is expected to see new entrants and dynamic transformations soon.

Looking to the Future and IPO Prospects

The company’s ambition transcends borders. With an eye on global markets and growing inquiries from sectors like entertainment, semiconductors, and computing, Preferred Networks is actively courting foreign investors. Nishikawa emphasized that broadening their market footprint is essential, noting, "There are limits to what we can achieve in Japan alone."

Plans for an initial public offering are also on the horizon. Okanohara warned that substantial funds will be necessary once the company’s new hardware products, including semiconductors, are ready for worldwide distribution. With expectations set for an IPO in the next three to five years, Preferred Networks appears steadfast in its pursuit of innovation and global influence.

In summary, Preferred Networks is not just another tech startup; it is a visionary force set to address everyday challenges through AI, poised to leave a mark on industries and global markets alike.