"Everything That Moves Will Become Robots"... Jensen Huang Announces Expansion of Robot Business

Jensen Huang CEO Delivers Keynote at GTC
AI Competitiveness to Expand with Robots
Unveils Blackwell GPU Supporting HBM3E

Jensen Huang, CEO of NVIDIA, emphasized business expansion in his keynote speech at GTC 2024 on the 18th (local time). In his speech, he declared a challenge into new territories, stating, "Everything that moves in the future will be a robot."

On the 18th (local time), at the keynote speech of 'GTC 2024' held in San Jose, USA, Jensen Huang, CEO of NVIDIA, introduced a robot that was self-trained.
[Image source=Getty Images Yonhap News]

Kim Sun-wook, Executive Director of Tech Marketing at NVIDIA Korea, who watched the keynote speech, described the event as "an occasion to broadly review what NVIDIA has achieved so far, starting from AI to the robotics business."

At the end of the keynote, CEO Huang showcased a robot he personally trained, concretizing the blueprint for the robotics business. NVIDIA has previously introduced robot processors and platforms for training robots. At this event, they announced plans to upgrade the technology from simple robotics to AI robotics and to further develop related businesses. They have embarked on full-scale work for humanoid robots.

Recently, various technological developments and business attempts to introduce humanoid robots have been ongoing in the industry. In NVIDIA's case, while growing its AI business, it has expanded by providing comprehensive infrastructure, including chip design and software. Since NVIDIA has strong competitiveness in AI, which will be the brain of robots, it is explained that the company can exert significant influence in the robotics business as well.

NVIDIA plans to materialize the concept of a giant graphics processing unit (GPU) through its new AI semiconductor series called ‘Blackwell’ architecture. Based on the 'GB200,' which combines two Blackwell GPUs, B200, with NVIDIA's own Grace central processing unit (CPU), the concept is to stack dozens of these to present a massive GPU.

Jensen Huang, CEO of NVIDIA, introduced the next-generation artificial intelligence (AI) chip "Blackwell" during the keynote speech at "GTC 2024" held in San Jose, USA, on the 18th (local time). [Image source=AFP Yonhap News]

CEO Huang emphasized, "Blackwell will enable AI implementation across all industries," and added, "I believe it will be the most successful product in the company's history." He also explained that many companies, including Amazon, Google, Meta, Microsoft (MS), and OpenAI, will adopt Blackwell.

In Korea, there was great interest in the event because NVIDIA AI chips include high-bandwidth memory (HBM) from Samsung Electronics and SK Hynix. HBM is a high-capacity, high-performance product implemented by stacking multiple DRAMs. It is called AI memory as it is used together with GPUs. The Blackwell GPU introduced by NVIDIA this time will incorporate the 5th generation HBM product, HBM3E.

Samsung Electronics and SK Hynix have exhibition booths at this year's GTC event, showcasing their latest HBM3E products. Market research firm TrendForce predicted that the share of HBM in global DRAM market sales will rapidly increase from 8.4% last year to 20.1% this year. TrendForce explained, "The annual bit growth of HBM (production increase rate converted to bit units) is expected to reach 260% this year."