'AI Breakthrough' by China's DeepSeek Rocks US Tech Giants

Startling advances by a little-known Chinese artificial intelligence (AI) startup have shaken some of the America’s top tech conglomerates. This sent a shockwave that knocked down leading tech stocks such as Nvidia.

China’s DeepSeek has rocked the AI sector earlier this month with its latest V3 model, which allegedly spent less than $6 million – a fraction of the billions splashed by companies such as OpenAI, Meta and Microsoft – and used less than optimal Nvidia chips, to produce an AI assistant that rates up with the best of its rivals.

While some tech chiefs have criticized DeepSeek’s “breakthrough” as simply copying OpenAI’s recent developments, other AI experts in the US say the Chinese company has also shown impressive innovation to get around cost limitations, plus chip curbs put in place by the Biden administration to try to maintain America’s lead in the superpowers’ tech race.

ALSO SEE: Bytedance ‘Plans $20 Billion Outlay on AI Infrastructure This Year’

Analysts interviewed by top US media outlets seem to agree with the creators of DeepSeek’s V3 model, who say it “tops the leaderboard among open-source models and rivals the most advanced closed-source models globally.”

And AI enthusiasts appear to have discovered that, as DeepSeek’s V3 model today (January 27) – 17 days after its release – overtook rival ChatGPT to become the top-rated free application available on Apple’s App Store in the US.

Data research firm Sensor Tower is one that has noted the surge in popularity among US users for the cheap Chinese model, Reuters said. The app’s key feature is it gives reasons first before answering a query, which is different to chatbots like OpenAI’s ChatGPT.

The milestone highlights how DeepSeek has left a deep impression on Silicon Valley, upending widely held views about US primacy in AI and the effectiveness of Washington’s export controls, which sought to limit China’s access to Nvidia’s most advanced chips.

AI models from ChatGPT to DeepSeek require advanced chips to power their training. The Biden administration has since 2021 widened the scope of bans designed to stop these chips from being exported to China and used to train Chinese firms’ AI models.

However, DeepSeek researchers wrote in a paper last month that said its V3 used Nvidia’s H800 chips for training, while spending $5.5 million.

Although this detail has since been disputed, the claim that the chips used were less powerful than the most advanced Nvidia products Washington has sought to keep out of China, as well as the relatively cheap training costs, has prompted US tech executives to question the effectiveness of tech export controls.

Nvidia, ASML shares plunge

The fact that companies may be able to run far cheaper AI models has undermined the rich valuation for Nvidia, whose shares boomed last year. Shares of the California chip designer were down 11% in premarket trading on Monday.

News about China’s low-cost, low-power AI model sparked worries about the need for costly chips. That sent the European tech index tumbling by 5.8% and caused futures tracking the tech-heavy US Nasdaq Composite to drop by 3.1%, according to Reuters.

Chip equipment maker ASML slid 11.5% to a near nine-week low, while ASM International slumped over 15%.

Siemens Energy, which provides electric hardware for AI infrastructure, sank 17.4% to the bottom of STOXX 600, while other AI-exposed firms such as Schneider Electric dropped 8.1%.

“This idea of a low cost Chinese version hasn’t necessarily been forefront, so it’s taken the market a little bit by surprise,” Fiona Cincotta, a senior market analyst at City Index, told Reuters.

“So if you suddenly get this low-cost AI model, then that’s going to raise concerns over the profits of rivals, particularly given the amount that they’ve already invested in more expensive AI infrastructure.”

Meanwhile, DeepSeek is a small Hangzhou-based startup founded in 2023 by Liang Wenfeng, the head of High Flyer, a hedge fund that used AI, after search-engine giant Baidu released the first Chinese AI large-language model.

DeepSeek develops open-source AI models, which allow developers to inspect and improve the software.

Since then, dozens of Chinese tech companies large and small have released their own AI models, but DeepSeek is the first to be praised by the US tech industry as matching or even surpassing the performance of cutting-edge US models.

Jim Pollard with Reuters

NOTE: Further details were added to this report on January 27, 2025 about stock price falls caused by DeepSeek’s revelations and DeepSeek itself. Further minor edits and corrections were done on January 28, 2025.