Nvidia and Microsoft shares steady after DeepSeek AI app shock

U.S. tech stocks were steady on Tuesday, a day after a sharp sell-off triggered by the sudden rise of DeepSeek, a Chinese artificial intelligence (AI) application. Chip giant Nvidia saw its shares rise 8.8%, after a significant drop the previous day. Experts have suggested that the AI sell-off may have been an overreaction.

The market was shaken as investors rapidly adjusted their bets on AI after DeepSeek claimed its model cost significantly less to produce than its rivals. Analysts said the development raised questions about the future dominance of US AI, as well as the scale of investment planned by US companies.

US President Donald Trump described the moment as a "wake-up call" for the US tech sector, but also suggested it could ultimately be "good for" the US. "If you can get the same end result with less cost, less input, I think that's a good thing for us," he told reporters on Air Force One. He also said he was not worried about the breakthrough, adding that the US would remain a leader in the field.

Optimism about AI investments has driven much of the boom in US stock markets over the past two years, fueling concerns about a potential bubble. DeepSeek became the most downloaded free app in the US just a week after its launch. Its emergence comes as the US warns of a technological competition with China and takes steps to restrict the sale of advanced chip technology needed for AI to China.

To continue working without a steady supply of advanced chip imports, Chinese AI developers are sharing their work with each other and experimenting with new technological approaches. This has resulted in AI models requiring far less computing power than previously. It also means they are far cheaper than previously thought, potentially disrupting the entire industry. Nvidia—the advanced chip company that underpins many AI investments, whose stock has soared over the past two years due to increased demand—was hit hardest on Monday, with its shares falling by around 17%, wiping nearly $600bn off its market value.

Janet Mui, head of market analysis at RBC Brewin Dolphin, said that investors' first reaction to something that appears to be a breakthrough is to sell off due to uncertainty. But Ms. Mui said she expects many companies, such as Apple, to benefit if AI models become cheaper. It could also be a boon for other tech giants facing scrutiny for spending huge sums on AI. After the US market was shaken on Monday, major indexes were steady. The Dow Jones Industrial Average in New York closed up 0.3%, the S&P 500 was up almost 1%, and the tech-heavy Nasdaq rose 2%.

The UK's biggest listed companies, the FTSE 100, were also steady on Tuesday, closing up 0.35%. Earlier, shares in Japanese AI-related companies, including Advantest, Softbank and Tokyo Electron, fell sharply, causing the Nikkei 225 to drop by 1.4%. Several other markets in Asia were closed for the Lunar New Year holiday. Financial markets in mainland China will be closed from Tuesday and will reopen on 5 February.

The company was founded in 2023 in the south-eastern Chinese city of Hangzhou by Wenfeng Liang. The 40-year-old information and electronic engineering graduate also founded the hedge fund that backs DeepSeek. He recently attended a meeting of industry experts with Chinese Premier Li Qiang. In an interview with the Chinese Academy of Sciences in July 2024, Mr. Liang said he was surprised by the reaction to previous versions of his AI model. "We didn't expect the pricing to be so sensitive," he said, "we just went at our own pace, calculated the cost and priced it accordingly."

After launching DeepSeek-R1 earlier this month, the company claimed its "performance is comparable to" one of OpenAI's latest models in tasks such as maths, coding and natural language reasoning. DeepSeek's technology has been praised by high-profile figures including OpenAI's CEO Sam Altman, who called it "an impressive model, especially in terms of what they're able to deliver at this price point," although he added that OpenAI "will obviously deliver better models." Marina Zhang, an associate professor at the University of Technology Sydney who focuses on China's high-tech industries, said: "DeepSeek being able to match US models despite having limited access to advanced hardware shows that software ingenuity and data efficiency can compensate for hardware limitations."

Ian Stoica, co-founder and executive chairman of AI software company Databricks, told the BBC that DeepSeek's lower costs could spur more companies to adopt AI in their businesses. "If that happens, the reduction in cost can accelerate the progress of AI," he said. "So, overall, the market will expand faster and the value of the market will grow faster." The Chinese company claims its model can be trained on 2,000 specialized chips, whereas leading models are estimated to require 16,000. But not everyone is convinced. Some, including tech magnate Elon Musk, have expressed skepticism about some of DeepSeek's claims. He responded to a post claiming DeepSeek actually had around 50,000 Nvidia chips by saying: "Obviously."

The sudden popularity has raised some concerns about cybersecurity. In Australia, science minister Ed Husic is among the experts urging caution, telling the Australian national broadcaster ABC: "There are a lot of questions that need to be answered in good time about quality, about consumer preference, about data and privacy management."