Jensen Huang “AI chip upgrades yearly… “The following wave of generative AI is robots”

by times news cr

2024-06-04 12:12:44

Subsequent-generation chips unveiled at Computex
Announcement of ‘Blackwell-Rubin’ one-year cycle launch
“AI Large Bang 60 years after the arrival of computer systems
“We are going to create a pc that walks or rolls.”

“We can even lead the event of AI robots.” Within the keynote speech at ‘Computex 2024’ held in Taipei, Taiwan on the 2nd, NVIDIA CEO Jensen Huang is introducing an industrial robotic utilizing synthetic intelligence (AI) know-how. CEO Hwang emphasised, “Robots are the subsequent wave of synthetic intelligence (AI),” and “NVIDIA will lead the event of AI robots.” Taipei = AP Newsis

“We are going to shorten the factitious intelligence (AI) chip launch cycle to 1 yr.”

Nvidia CEO Jensen Huang declared on the 2nd that he would shorten the discharge interval for next-generation AI chips from two years to 1 yr, saying, “We transfer on a yearly foundation.” Even earlier than ‘Blackwell’, the next-generation graphics processing unit (GPU) unveiled in March of this yr, has even been shipped, plans to launch ‘Blackwell Extremely’ in 2025 and ‘Rubin’ in 2026 had been introduced. CEO Hwang as soon as once more revealed his ambitions for humanoid robots, saying, “60 years after the arrival of computer systems, a ‘Generative AI Large Bang’ has occurred. Generative AI robots with bodily properties are the subsequent runner.”

● AI chip geared up with next-generation HBM unveiled

In his keynote speech at ‘Computex 2024’ held on the Nationwide Taiwan College Gymnasium in Taipei, Taiwan on this present day, CEO Hwang stated, “NVIDIA is taking the lead in international AI. “Every thing is accelerating,” he stated, revealing the next-generation AI chip roadmap. Particularly, it attracted consideration by asserting that it could cut back the event of the next-generation ‘AI accelerator’ from a two-year cycle to 1 yr.

The so-called AI accelerator is an AI-specialized semiconductor that’s made by combining GPU and high-bandwidth reminiscence (HBM). SK Hynix’s sixth technology HBM ‘HBM4’ is deliberate to be put in in Rubin in 2026. CEO Hwang stated, “Rubin will likely be geared up with 8 HBM4s, and Rubin Extremely will likely be geared up with 12 HBM4s.”

Competitors in AI know-how is predicted to grow to be extra intense as NVIDIA publicly introduced its plan to speed up the discharge cycle of next-generation AI chips. Final month, SK Hynix additionally introduced that it could advance the HBM4 mass manufacturing interval by one yr from 2026 to subsequent yr, and Samsung Electronics additionally plans to mass produce HBM4 subsequent yr.

At this occasion, CEO Hwang stated, “NVIDIA has opened the door to the AI ​​period,” and stated, “Present computing strategies can’t deal with the exponentially elevated quantity of knowledge, and that is solely potential by means of ‘NVIDIA-style accelerated computing.’” Hong Kong’s South China Morning Publish (SCMP) commented, “Jensen Huang has declared an period by which every part from semiconductors to software program runs on the NVIDIA platform.”

● “The following wave of AI is robots”

CEO Hwang didn’t conceal his need for ‘AI robots’ that day. He stated, “NVIDIA has been making computer systems for desks (desktops), pockets (smartphones), and knowledge facilities,” and added, “Sooner or later, we’ll make computer systems (robots) that stroll or roll on wheels.” He beforehand revealed his aspirations for a robotic that may be known as ‘bodily AI’ in his keynote speech at ‘GTC 2024’, an NVIDIA developer occasion in March.

CEO Hwang stated, “The economic revolution pushed by AI has begun,” and added, “Software program runs in accordance with enter instructions, however generative AI creates the required abilities for customers.” He additionally launched an instance of partnering with a producing plant. At present, Foxconn has launched distant management and AI evaluation features utilizing ‘Omnibus’ know-how, a digital digital house know-how developed by NVIDIA. Samsung Electronics additionally plans to cooperate with NVIDIA to advance semiconductor processes by 2030.

Because of the AI ​​craze, the environment of this yr’s Computex has utterly modified to ‘the AI ​​expo that pulls the world’s consideration.’ Computex, which began in 1981, is Asia’s largest info know-how (IT) exhibition, however has just lately been lagging behind CES within the US and IFA in Europe. Nonetheless, this yr, with the theme of ‘Connecting AI’, the standing has modified considerably with the participation of many CEOs of main semiconductor corporations similar to NVIDIA, Intel, AMD, and Qualcomm.


Reporter Lee Ji-yoon [email protected]
New York = Correspondent Kim Hyun-soo [email protected]

Sizzling information now

2024-06-04 12:12:44

You may also like

Leave a Comment