Two weeks ago, market darling Nvidia showcased a remarkable 12-fold y-o-y surge in its 3QFY2023 ended October earnings, positioning itself as a prominent player in AI technology.
Revenue for the quarter saw a 206% y-o-y increase, reaching US$18.12 billion. The stellar results significantly contributed to Nvidia’s staggering ytd share price growth of about 320%, solidifying its status as the most sought-after tech stock in the current market.
AI fervour also swept through Singapore last week as Nvidia made the headlines. The city-state reportedly contributed to 15% of the company’s total revenue of US$18.12 billion ($24.3 billion). This positions Singapore as the fourth-largest market for Nvidia’s chips, following the US, China (including Hong Kong) and Taiwan.
But this 15% of Nvidia’s revenue which came from Singapore does not indicate that 15% of worldwide shipments went to Singapore. More accurately, 15% of billings have been made to Singapore, in which some regional and global companies may have multi-layered subsidiary units but are headquartered here, resulting in the financial transactions recorded as occurring in Singapore.
This is according to the Nvidia team, which made the clarification during a media roundtable with Nvidia’s founder and CEO Jensen Huang on Dec 6. This is Huang’s first visit to Singapore in 25 years, following an invitation he had accepted earlier this year from Jacqueline Poh, managing director of the Economic Board of Singapore.
Huang is due to "potentially announce some large investments" and meet Prime Minister Lee Hsien Loong during his time here. The Taiwan-born American businessman’s tour to Asia is against the backdrop of an ongoing tech war centred on semiconductors between the US and China, in which the US announced fresh bans on the A800 and H800 chips in late 2022.
“We’ve been rumoured to be working on two products called L2O and H2O,” says Huang. “What I will say is that whatever we decide to do with those two products, if we offer it, we’ll comply very specifically with the regulation as the last generation of products are also compliant with regulations.”
In its earlier years, Nvidia was better known for making graphics processing chips to run ultra-realistic 3D computer games. With its chips as the mainstay for powering AI capabilities, the company now finds itself in a critical yet delicate position as the US-China trade war wages on.
When asked by The Edge Singapore about the allegations that Singapore had been used as a port for AI chip transhipment to China, Huang doubled down on his stance of sticking to a regulatory-compliant approach. “The regulations are very, very clear, and there’s little room to interpret the regulations, and we will comply with the regulations,” he adds.
See also: Anthropic CEO says mandatory safety tests needed for AI models
According to Reuters, Nvidia commands more than 90% share of China’s US$7 billion AI chip market. Still, analysts said new US curbs on chip exports are likely to create opportunities for Chinese rivals to make inroads.
With China contributing about 20% to Nvidia’s revenue historically, Huang says that it is hard to project exactly what the proportion of its business will be from this market moving forward. He admits that his competitors in China, specifically Huawei Technologies, are “a very formidable competitor”.
But even as other chipmakers — some of whom are Nvidia’s customers themselves — are fast advancing the creation of their own chips, Huang remains confident. “We’re the only AI company in the world that builds chips, systems, networks, switches, all the software, and we can help everyone, anyone become an AI company,” he claims.
Nations and companies of all sizes reach out to Nvidia because of its “incredible sense of scale”, which allows its customers to stay at the cutting edge of technology with the most advanced chips, he adds.
Huang’s address to journalists in Singapore came a day after his visit to Japan. Huang, who founded the company in 1993, said he would do his best to supply artificial intelligence processors to Japan amid extremely high market demand. “We will build a network of AI factories here in Japan so that Japan can process the data of the society and create intelligence for the society and for the industry,” Huang said.
In Singapore, Huang lauded the city-state’s initiative in pushing for the use and development of AI at the national level. Just this week, the Smart Nation Group, a unit under the Prime Minister’s Office, announced an updated version of the National AI strategy (NAIS) or NAIS 2.0, which includes a new $70 million funding package to help develop Singapore’s research and engineering capabilities in multi-modal large language models.
Huang praised the second edition of the strategy, saying it was “one of the best documents he’s read about AI so far”. In solidarity with the government’s NAIS 2.0 plans, Huang sketched out ambitions to build an even larger supercomputer in Singapore, which will be in addition to a supercomputer and an AI technology centre it has here.
To stay ahead of Singapore and the region’s corporate and economic trends, click here for Latest Section
The new computing muscle would enable Singapore to build its own foundational model instead of relying on Western AI models, which may not be suited to Singapore’s vision. Data used or generated by an AI model usually contains information that provides a unique advantage, and is therefore important for those investing in the use and development of AI, Huang explains.
Meanwhile, Nvidia is already working with the Infocomm Media Development Authority (IMDA) on an AI foundation model, based on the existing open-sourced Sea-Lion (Southeast Asian Languages in One Network) model, launched by the national AI Singapore programme.
Earlier this week, the government of Singapore coordinated a three-day conference, titled the Singapore Conference on AI (SCAI). It saw academics, industry experts and government officials discussing critical questions of AI to result in global good.
Launched by Deputy Prime Minister Lawrence Wong on Dec 4, who announced a pledge to triple the AI talent pool to 15,000 in the coming few years, the conference concluded with a series of 12 guiding questions, which would presumably help frame the development of this field here where rules and norms of behaviour are evolving and debated on.
These questions ranged from the reliability of AI to mitigating risks and enhancing safety. According to the Smart Nation Group, they are intended as a comprehensive set of challenges to help drive the global AI agenda.
As Wong acutely pointed out, nearly every country in the world is vying to be the AI hub of the world, this conference is, therefore, Singapore’s attempt at making an example of AI governance. Meanwhile, Huang welcomed the increasing participation of countries in conversations around AI.
He observed that his first wave of customers were primarily US Internet companies. But now he is witnessing the second wave of customers made up of the world’s countries. “Each one of these countries wants to build its own foundation to support its own startups, its own companies and its own industries,” he says. “Nobody can outsource their application programming interface (API), nobody can outsource their intelligence. What makes you, you, is your intelligence. You have to build it yourself.”
With that, Huang expects Nvidia to see a continuous significant demand for the next 10 years. “This is the beginning of a new technology cycle and we’re changing everything,” he says.
