Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Toru Nishikawa (right), co -founder and executive director of Prefered Networks, and Daisuke Okanohara, co -founder and executive researcher, with a robotic arm Fanuc Corp. at the headquarters of his company in Tokyo, Japan, March 16, 2018.
Bloomberg | Bloomberg | Getty images
Prefereted Networks is a little known name outside Japan, but the artificial intelligence unicorn backed by Toyota has great plans to use deep learning to solve problems of “real world”, with the ultimate goal of globalizing.
“During the last 10 years, we have been working on the use of AI to solve real world problems”, executive researcher, Daisuke Okanohara, he told CNBC “Asia management“ The company has ventured in industries as varied as transport, medical care and robots.
Prefered Networks has caught the attention of big names such as the automotive giant Toyotathat in 2017 invested 10,500 million yen ($ 95.4 million) in the company in addition to 1 billion initial yen in 2015.
Prefereted Networks is one of the few unicorns in Japan. Although Japan’s starting ecosystem has grown, Japanese companies valued at more than $ 1 billion represent only 0.5% of the world’s total, According to a working document of the International Monetary Fund Posted in December.
The deep learning startup is also in a Joint ventura with Trading House Mitsui & Co. In the field of autonomous driving. The two have invested in a company, T2, to address the challenges associated with the truck industry, including “extremely long” work that extend at night and a group of drivers in Declive, said CEO Toru Nishikawa.
But competition in the automotive industry was also a factor. Recognizing the current limitations of the company, Okanohara added: “If we had to address the problem of regular cars, it would be extremely difficult. There are many companies that work in autonomous driving, and it is a very competitive field.”
Since then, the ambitions of the favorite networks have gone much further. But compared to other AI solutions in digital space, in which it is “relatively easy to achieve results with users”, dealing with real world problems requires more time, Okanohara said.
When asked how long it takes for the company’s technologies to start earning money, Nishikawa said he could take up to five years in a field such as the discovery of AI -based materials.
“First, we begin the joint research with the companies and then evaluate whether it can be marketed or not. Once we determine that it can be marketed, we work to create services and products that can be provided to customers,” he said, and added that “in total, this process takes between three and five years before commercialization and practical launch.”
As for the last developments of AI, Okanoharo said that the favorite networks had recognized the “very good technology” of Depseek and “paid close attention” before she was catapulted to fame, and added that “several players” will soon enter the chip industry.
Prefereted Networks hopes to be among them. Recognizing that Nvidia The generative strategy of AI has been very good so far, but it has a graphics processing unit design that is not yet “optimal”, Nishikawa said that preferred networks are in the midst of “creating more advanced processors” to obtain a competitive advantage.
It remains to be seen if the company can achieve the scale of its ambitions. For now, he is looking at foreign investors, and is receiving “many investigations in entertainment, semiconductor and computers,” according to Nishikawa, who said that “there are limits for what we can do only in Japan, and the global market is much larger than the Japanese market.”
Obtaining an initial public offer is one of the company’s objectives, he added. When a period of time was requested, Okanohara said: “I think that the need for huge funds will come at the time we present our hardware products, such as semiconductors, the world. So I think we will aim to make public at that time,” and added that it could take between three and five years.