Nvidia Steps Up AI Push

Nvidia has invested in “more than 20” companies over the past 12 months, a sixfold increase over last year, with targets ranging from key rivals of OpenAI to startups in areas from healthcare to energy. The company denies using Nvidia products as an investment condition.

Jensen Huang criss-crossed Asia this year, visiting Chinese Taipei, India, Japan, Singapore, Malaysia and Vietnam and meeting with politicians, business leaders and scientists to broadly lay out collaborations around AI. During a visit to Vietnam, the Nvidia CEO was seen eating noodles on the street.

2023 has been a standout year for Nvidia, which has also picked up the pace. Thanks to the explosion in generative AI, chipmaker Nvidia is on track to make more profit this year than its total profits over the previous 25 years since going public. It is using some of that new money to invest in AI startups, potentially positioning itself for even greater windfalls. It has invested in “more than 20” companies over the past 12 months, a sixfold increase over last year.

In addition to stepping up investments to cement Nvidia’s position as the market leader in AI processors, CEO Jensen Huang has also been exceptionally active this year, travelling extensively across Asia to meet with politicians, scientists and business leaders to broadly lay out collaborations around AI. Most recently, he visited four countries – Japan, Singapore, Malaysia and Vietnam – within 8 days, all with a focus on AI.

Investing in companies that use Nvidia products

In a December 11 blog post, Nvidia detailed its multifaceted investment approach. It said over the past 12 months, it has invested in “more than 20” companies and provided free technology to thousands of startups, with partners ranging from innovative early-stage projects in healthcare and energy to major AI businesses.

Consultancy Dealroom estimates Nvidia was involved in 35 deals in 2023, almost six times as many as last year. This represents Nvidia’s thoughtful yet bold investment strategy, as it has overtaken Silicon Valley venture capitalists like Andreessen Horowitz and Sequoia.

According to Nvidia’s latest quarterly report, it made investments worth $872 million in “non-affiliated companies” in the nine months to October, more than 10 times what it invested in the same period last year.

Nvidia explains it invests in the ecosystem in three ways to drive the change that accelerated computing enables. First, through Nvidia corporate investments; second, through its venture investment arm Nventures; and third, through Nvidia Inception, its program to support startups and connect them with venture capital.

“The AI startup ecosystem is very important to us,” Nventures lead Mohamed Siddeek told MarketWatch in an interview. “We invest in companies in that ecosystem to really help drive the innovation that accelerated computing and AI is unlocking.”

“The primary criteria for Nvidia is relevance, overall,” Siddeek told the Financial Times about the strong confidence underpinning its investment activity. “Companies that use our technology, depend on our technology, build businesses on our technology – I can’t imagine us investing in a company that does not use Nvidia products.”

Nvidia’s portfolio includes two of OpenAI’s biggest rivals in AI development – Inflection AI and Cohere. Another investment went to Paris-based AI startup Mistral, which was valued at €2bn earlier this month. Hugging Face and CoreWeave are two other targets, both of which use Nvidia GPUs or software. Runway, a platform for artistic creation via generative AI, has also gained Nvidia investment.

With clear dominance in “distilling” large language models, Nvidia GPUs have become the most coveted products among AI firms this year. Asked if recipients of funds get preferential access to purchase GPUs, Siddeek responded: “We don’t help anyone jump the queue.” He denied using Nvidia products as part of investment conditions but added: “We try to be friendly with the investors.”

Other investment recipients include Ayar Labs, which specialises in optical chip-to-chip connection; Databricks, which provides a machine learning data platform; Ready Robotics, developer of an industrial robot operating system; Recursion, Kore.ai and Utilidata, which offer solutions in drug discovery, conversational AI and smart grids. More than a third of the 19 companies invested by Nventures focus on using AI to enhance new drug discovery. Other startups in the portfolio also concentrate on healthcare, developing new approaches for laparoscopic surgery and medical imaging.

Jensen Huang’s whirlwind Asia tour

According to Vietnamese media, when giving a speech at a Hanoi semiconductor and AI conference hosted by Vietnamese government agencies on December 11, Jensen Huang announced Nvidia would set up a design centre in Vietnam but did not reveal technical details of the facility.

On his first visit to Vietnam, Huang said Nvidia was willing to cooperate with Vietnam to improve the country’s AI infrastructure and workforce. Vietnam has many computer scientists and is a global leader in software exports. He said Vietnam could create 1 million AI engineers.

In a meeting with Vietnamese Prime Minister Pham Minh Chinh, Huang said Nvidia “considers Vietnam its home” and hopes to establish a semiconductor base there. In a statement after the meeting, the Vietnamese government said “the base will attract talent from around the world and contribute to Vietnam’s semiconductor ecosystem and digitalisation.”

On December 10, Huang met with Vietnamese Prime Minister Pham Minh Chinh. Vietnam houses large chip assembly plants including those of Intel. The Vietnamese government’s news website reported on December 10 that Nvidia has invested about $250 million in Vietnam.

Huang’s Hanoi visit came as the Vietnamese government was seeking closer diplomatic and economic ties with the US and help developing Vietnam’s semiconductor ecosystem. In September, the US State Department announced a partnership with Vietnam under the CHIPS Act. John Neuffer, president of the Semiconductor Industry Association, and representatives of US companies visited Prime Minister Chinh earlier this month.

Jensen Huang eating on the street in Vietnam. Vietnam was the fourth stop on Huang’s Asia tour this month.

On December 5, meeting with Japan’s Economy Minister Yasutoshi Nishimura, Huang said Nvidia would collaborate with Japanese research institutes and companies to build an AI technology ecosystem in Japan, including an AI factory network. Meeting with Prime Minister Fumio Kishida, he also said he would try his best to prioritise Japan’s demands for GPUs.

In November, the Japanese government passed a budget bill deciding to subsidise around 20 trillion Japanese yen for the chip sector to strengthen Japan’s global position in semiconductors.

On a December 6 visit to Singapore, Huang said Nvidia was working with the Singapore government to create a large language model trained on 11 languages. He held talks with Prime Minister Lee Hsien Loong, thoroughly discussing how Nvidia could help, support and participate in the country’s just-launched National AI Strategy 2.0, which sets out to double Singapore’s AI workforce to 15,000.

Huang then visited Malaysia. Nvidia announced it would invest 20 billion Malaysian ringgit (around $4.29 billion) in Malaysia and collaborate with YTL Corporation Berhad (YTL), a Malaysian conglomerate, to build AI infrastructure.

Analysts said Huang’s Asia tour this time could be to look for potential partners. Currently, TSMC and Samsung are important wafer foundries for Nvidia, but Nvidia is also considering adding other potential foundry partners. Nvidia cooperating with local manufacturers in Asia could also be in response to the US government’s new chip export restrictions.

In November, Nvidia warned in its earnings report that sales in China are expected to decline significantly in Q4 after the new US regulations take effect.

Looking back at 2023, Huang has frequently appeared in Asia. He made two visits to Chinese Taipei in May and October.

At the end of May during COMPUTEX Taipei, Huang delivered a keynote speech and announced collaborations with Taiwanese manufacturers on “digital twins” and an AI R&D centre in Taiwan, as well as partnerships with Taiwanese universities to nurture AI talent.

In mid-October, Huang and Foxconn chairman Young Liu appeared together at Foxconn’s annual tech showcase, announcing they would jointly build “AI factories”, using Nvidia chips and software to construct new types of data centres for development including self-driving cars, autonomous robots and industrial robots.

In early September, Huang visited India and met with Prime Minister Modi, “detailing the tremendous potential India offers in AI”. Huang also hosted a dinner with dozens of researchers from institutes like the Indian Institutes of Science and Technology, including some top talents in areas from large language models to astrophysics, medicine, quantum computing and natural language processing.