Sunday 28 / April / 2024
Taipei: Nvidia Corporation CEO Jansen Huang said that Taiwan was in the middle of the new artificial intelligence revolution. He made the remarks during a recent dinner gathering of Taiwanese people in Silicon Valley, Central News Agency (CNA) Taiwan reported.
Jansen Huang stressed that it must be celebrated that "Taiwan is right in the middle" of a new industrial revolution in which "something new is being made, and made in a new way," according to CNA report.
He recalled the manufacturing process of the RIVA 128, Nvidia's first commercial success, which he called the "most complicated chip at the time." He said that Nvidia did not have the budget and he wrote a letter to Taiwan Semiconductor Manufacturing Co. (TSMC) founder Morris Chang, who called him and the two started their nearly 30-year collaboration.
Jansen Huang said, "Taiwanese people made Nvidia's GPU the fastest in the world." He called himself a "a very good ambassador of Taiwan", as he understood the importance of Taiwan at the "centre of this new computing revolution," CNA reported.
He vowed to talk about this story of Taiwan and asked the attending Taiwanese to join him in sharing the story. Huang has announced his participation in COMPUTEX Taipei 2024 which will be held at Taipei Nangang Exhibition Center on June 4-7.
Last week, Nvidia kicked off its annual GPU technology conference (GTC) in California's San Jose by unveiling its latest AI super chip that uses TSMC's 4-nanometer process technology, according to CNA report.
In his keynote address that was live-streamed, Jensen Huang introduced Nvidia's new flagship AI chip 'Blackwell' that has been named in honour of mathematician and game theorist David Harold Blackwell and said it will speed up AI computing.
According to the news report, the superchip has 208 billion transistors and is manufactured using TSMC's N4P (4nm) process technology. A GPU is an electronic circuit that can do mathematical calculations at high speed and was initially create to accelerate image processing.
He termed Blackwell "a very big GPU" and stressed that it is increasingly necessary as AI models larger than the one used by ChatGPT will be required as "we are going to train [AI] with multimodality data -- not just text on the internet but images and graphs ... and videos."
