Chief Executive Kim Yu-won of Naver Cloud said on the 6th that the company will unveil its Humanoid Robot under development at the end of November.
At Naver's annual conference DAN25 held the morning of the same day at COEX in Samseong-dong, Gangnam-gu, Seoul, Kim said, "Naver Labs is preparing a humanoid," adding, "Unfortunately, you won't be able to see the humanoid at today's event, but toward the end of this month, you will be able to see a cute one about 1 meter tall being put to hard work at Naver's headquarters," and made the announcement.
During the keynote session led by CEO Choi Soo-yeon that day, an image of a robot presumed to be Naver Labs' Humanoid Robot was also shown in the presentation materials. This robot operates in conjunction with Naver's robot cloud "ARC" and the web-based robot operating system "ARC Mind." Its name is known as "Mininoid."
Previously, Naver Labs CEO Seok Sang-ok, in a lecture on "Digital Twin, AI and robots for future cities" at the Digital Insight Forum hosted by the Korea Information & Communication Promotion Association (KAIT) at Josun Palace in Gangnam-gu, Seoul, on the 16th, said a Humanoid Robot would be unveiled within the year.
Naver Cloud is internalizing "physical AI" technology so that artificial intelligence (AI) can operate in real industrial sites. Naver began research on robots and Autonomous Driving in 2016, and in 2021 advanced it into physical AI based on a foundation model.
Kim said, "Physical AI is technology in which machines perceive and understand space, and make decisions and act on their own," adding, "Naver is building a constantly evolving physical AI ecosystem by combining the real data accumulated by robots throughout the 1784 headquarters and data centers, the cloud platform that connects them, and on-board AI."
Kim also introduced the "Sovereign AI 2.0" vision, saying, "We will make AI transformation of industry a reality with our data, infrastructure and technology."
He explained, "While the existing sovereign AI focused on technological self-reliance centered on language and culture, Sovereign 2.0 is a concept that expands this across industry and daily life to enhance national competitiveness."
Kim also said the company plans to release an upgraded version of its private cloud "Neurocloud" in June next year for corporations that want to operate data and AI safely.
At a press briefing held in the afternoon, Kim said the 60,000 Blackwell GPUs that Naver recently agreed to procure from Nvidia may be insufficient.
Kim said, "The supply plan does not have an exact timeline yet; we are scheduling it," adding, "Because it will be used across all areas, including HyperCLOVA X as well as on-service AI and physical and vertical AI, I don't think 60,000 units will be enough."
Regarding the recent "AI bubble" argument, he said, "In the end, the value relative to the input expense has to grow," adding, "A lightweight version of an AI model is essential, and collaboration with corporations and government agencies is also important."
On the issue of reliance on Nvidia GPUs, he noted, "Many Neural Processing Unit (NPU) developers are trying to build their own chips, and to do that, they must move beyond the lab to real testing and use," adding, "We are actively testing chips made by companies like FuriosaAI even now, and if the performance is good, we plan to adopt them. The same goes for Intel or AMD chips."