Advances by China’s DeepSeek sow doubts about AI spending

Stay informed with free updates

Tech stocks tumbled on Monday as advances by Chinese artificial intelligence start-up DeepSeek cast doubt on whether the US could sustain its leadership in AI by spending billions of dollars on chips.

DeepSeek last week released its latest large language AI model, which achieved a comparable performance to that of US rival OpenAI, even though the company claims to have far fewer Nvidia chips.

The results sent a shockwave through markets on Monday, with Nvidia on course to lose more than $300bn of market value, the biggest recorded drop for any company, as investors reassessed the likely future investment in AI hardware.

Venture capital investor Marc Andreessen called the new Chinese model “AI’s Sputnik moment”, drawing a comparison with the Soviet Union stunning the US by putting the first satellite into orbit.

Shares in Nvidia, one of the biggest winners from the AI revolution, fell 13 per cent at the open in New York. The tech-heavy Nasdaq Composite index opened down 3.5 per cent, while the S&P 500 index declined 2.2 per cent. Microsoft fell 4.3 per cent.

In Europe, chip equipment maker ASML was down 9 per cent, while Dutch semiconductor company ASM International extended earlier losses to trade 12.5 per cent lower.

The rout extended well beyond traditional tech names. Siemens Energy, which supplies electrical hardware for AI infrastructure, plunged 20 per cent. Schneider Electric, a French maker of electrical power products that has invested heavily in services for data centres, fell 9.3 per cent.

To some, the sell-off in the companies making the “picks and shovels” of the AI revolution echoed Cisco’s share-price crash when the dotcom bubble burst.

Nvidia, Broadcom and other chipmakers have benefited from Silicon Valley’s race to build ever-larger clusters of chips, which the likes of xAI boss Elon Musk and OpenAI’s Sam Altman have argued are needed to keep advancing AI’s capabilities.

Nvidia’s chief executive Jensen Huang and Broadcom’s Hock Tan have argued in recent weeks that they expected the data centre building frenzy to continue until the end of the decade.

Some content could not load. Check your internet connection or browser settings.

“It shows how vulnerable the AI trade still is, like every trade that is consensus and based on the assumption of an unassailable lead,” said Luca Paolini, chief strategist at Pictet Asset Management.

But some Wall Street analysts and AI researchers have questioned the hype surrounding DeepSeek’s achievement. “It seems categorically false that ‘China duplicated OpenAI for $5M’ and we don’t think it really bears further discussion,” wrote analysts at Bernstein in a note to clients.

Some researchers have even speculated that DeepSeek was able to take shortcuts in its own training costs by leveraging the latest models from OpenAI, suggesting that while it has been able to replicate the latest US developments very quickly, it will be harder for the Chinese company to pull ahead.

AI investment by large-cap US tech companies hit $224bn last year, according to UBS, which expects the total to reach $280bn in 2025. OpenAI and SoftBank announced last week a plan to invest $500bn over the next four years in AI infrastructure.

Even following DeepSeek’s latest release, Meta chief Mark Zuckerberg said in a Facebook post on Friday that he planned to spend as much as $65bn on AI infrastructure this year.

Founded by hedge fund manager Liang Wenfeng, DeepSeek last week released a detailed paper explaining how to build a large language model that could automatically learn and improve itself.

“It seems as if there is a bit of reality dawning that China has not been sitting idle, even as these tariffs and investment restrictions on tech companies have been put in place,” said Mitul Kotecha, Asia head of emerging markets macro and foreign exchange strategy at Barclays.

The US imposed stringent restrictions on chip exports to China under former President Joe Biden, banning the sale of Nvidia’s most advanced models to the country.

Recommended

Some analysts argued that DeepSeek’s advances would ultimately prove positive for AI chipmakers such as Nvidia.

Dylan Patel, chief analyst at chip consultancy SemiAnalysis, said cutting the costs of training and running AI models would over the longer term make it easier and cheaper for businesses and consumers to adopt AI applications.

“Advancements in training and inference efficiency enable further scaling and proliferation of AI,” said Patel. “This phenomenon has occurred in the semiconductor industry for decades, where Moore’s Law drove a halving of cost every two years while the industry kept growing and adding more capabilities to chips.”

Some Chinese tech stocks advanced amid the excitement over DeepSeek, although the wider CSI 300 index closed down 0.4 per cent. In Hong Kong Baidu closed 4 per cent up and Alibaba was up 3 per cent.

Video: AI is transforming the world of work, are we ready for it? | FT Working It

Source link