Nvidia could invest in 'iconic' AI site in Singapore: CEO Jensen Huang
The Taiwan-born American also wants to build a new, larger supercomputer here.
This audio is generated by an AI tool.
SINGAPORE: Artificial intelligence chip giant Nvidia could back an "iconic site" in Singapore for the booming technology, and also construct a new and bigger supercomputer here, said co-founder and CEO Jensen Huang on Wednesday (Dec 6).
The Taiwanese-American tech executive, whose 30-year-old company recently hit US$1 trillion in valuation, told reporters he was slated to meet Singapore's Prime Minister Lee Hsien Loong and executives in the Economic Development Board (EDB) to discuss and "potentially announce some large investments" Nvidia is making in the Southeast Asian nation.
Asked for more details, Mr Huang said he could only reveal more after going over the proposals with Singapore officials. CNA has contacted EDB for comment.
Clad in his trademark black leather jacket, Mr Huang spoke effusively about Singapore's updated national artificial intelligence strategy (NAIS), which was unveiled by Deputy Prime Minister Lawrence Wong on Monday.
At a Singapore Conference on AI on Monday, Mr Wong said Singapore wanted to triple its AI talent pool and have more companies establish their AI "Centres of Excellence" here. An "NAIS 2.0" report also proposes establishing an "iconic AI site" bringing together a community of creators and practitioners.
"I love that the ambition of Strategy 2.0 is not just to be participating in artificial intelligence but to be excellent in artificial intelligence, to be a world leader in artificial intelligence - and there's every reason to believe that you can," said Mr Huang, who called Singapore's report "incredibly thoughtful".
Nvidia already has a supercomputer and an AI technology centre in Singapore, and it has collaborated with hundreds of researchers and start-ups here over the past seven years, the 60-year-old added.
According to the Singapore's National Supercomputing Centre's website, its AI platform - hosted on a supercomputer - includes Nvidia chips and software.
California-headquartered Nvidia has also been working with the Infocomm Media Development Authority (IMDA) on an AI foundation model.
IMDA said on Monday it was developing a localised multi-modal large language model (LLM) that can understand and manage context-switching between Singaporean and regional languages. It will be based on the existing, open-sourced Sea-Lion (Southeast Asian Languages in One Network) model launched by the national AI Singapore programme.
LLMs are machine learning tools trained to understand and generate human language. And AI foundation models, in contrast to narrow AIs, can perform many tasks, enabling different applications to be built upon them.
"It is essential that Singapore create your own AI ... it will encode, capture, it will learn the knowledge, the culture, the intelligence of the society ... once we have that Singapore foundation, then the rest of the industries, the rest of society, the rest of the companies, researchers can then build upon that," said Mr Huang.
Some potential uses for the AI foundation model would be in financial services, fraud detection and medical research, said Mr Huang.
And Nvidia is looking to do more in Singapore, including building an even larger supercomputer and "potentially investing in a significant iconic site for AI".
Related:
Mr Huang's Singapore stop is part of an Asian tour and comes after a Japan visit, where he said he would prioritise Tokyo's demand for GPUs - graphics processing units now used extensively for AI development - amid extremely high market demand, Reuters reported.
On Wednesday, Mr Huang was asked about rumours of delayed shipments of Nvidia's chips to China. He again pointed to a surge in demand for Nvidia's GPUs, noting there was now a "second wave" globally from countries and companies looking to replicate generative AI tech.
Generative AI uses deep-learning models that can create new content based on past data that they were trained on, with chatbot ChatGPT perhaps the most famous example thus far.
"Our sales this year was a big surprise for the world and it was a big surprise for our supply chain .... while we're catching up to demand of our customers, new demand is coming in," said Mr Huang.
Nvidia last month reported US$18.12 billion in revenue in the third quarter, up 206 per cent from a year ago and an increase of 34 per cent from the previous quarter.
Reporters also quizzed Mr Huang on a recent report on Singapore accounting for about 15 per cent or US$2.7 billion of Nvidia’s revenue for the third quarter, jumping 404.1 per cent from the same period a year ago. The country was fourth behind the United States, Taiwan and China in Nvidia's sales revenue ranking for that quarter.
Nvidia clarified that such country revenue figures may not reflect actual shipments, due to many companies being headquartered in Singapore and thus billed here. The chips may have been shipped elsewhere instead.
Nonetheless, demand for GPUs in Singapore is large and growing, and a "significant amount" was shipped to Singapore, said Mr Huang, adding that much of the demand here is fuelled by cloud service providers supporting AI start-ups.
He predicted that GPUs would pave the way for a whole new industry and new sector of the economy.
"Data comes into this computer, this GPU and what comes out of it (is) intelligence," said Mr Huang. "And every country is very excited about this and wants to get involved, because you want to use your own data (and) manufacture your own intelligence."
