ADVERTISEMENT

The Chinese artificial intelligence startup DeepSeek sent jitters to global markets, sparking a tech-led selloff in the US and Europe on Monday. The Chinese AI is reportedly to be more cost-effective and performs better than America’s leading models, raising concerns about the valuation hype of the soaring tech stocks. 

Market responses

Nvidia, the world’s largest AI chipmaker, saw its shares plummet by 17%, erasing $600bn (€573bn) in market capitalisation in what was the largest intraday decline in history. Europe’s AI chip equipment manufacturer, ASML, also experienced a 7% drop in its share price. The US tech-heavy Nasdaq Composite shed over 3%, with the technology sector as a whole down 5.6%. European stock markets followed suit, weighed down by losses in the tech sector.

On Monday, the US dollar softened against the other currencies in the G-10 group as the US government bond yields sharply declined on risk-off sentiment. The Euro strengthened against the dollar to a six-week high. Commodity prices, including gold and crude oil, were also sharply down due to the broad selloffs as investors liquidated positions for safety.

However, the trend reversed on Tuesday’s Asian session, with the dollar rebounding, causing pull backs in other currencies, with the euro-dollar pair down by 0.52% to 1.0437 at 4 am ECT. 

“As far as the markets are concerned, I think there is a shoot first and ask questions later quality to the sell-off,” Kyle Rodd, a senior market analyst at Compital.com, wrote in a note to clients. “There’s a lot of information to parse and discount”, he said, referring to the credibility of DeepSeek’s computing figures. 

How does DeepSeek threaten the US AI dominance?

DeepSeek, founded in 2023 by hedge fund High Flyer, unveiled its latest free, open-source large language model (LLM), R1, in late December. According to the company’s release, the model was developed for under $6m (€5.7m) in just two months. By contrast, US hyperscalers have invested billions of dollars in AI infrastructure.

Remarkably, the cost-efficient Chinese LLM outperformed some of the most prominent training software in the market, including OpenAI’s GPT-4o, Google’s Gemini 2.0 Flash, Anthropic’s Claude 3.5 Sonnet, and Meta’s Llama 3.1, particularly in solving complex mathematical and coding problems.

On Monday, DeepSeek AI Assistance overtook ChatGPT o1 as the most downloaded free app on the Apple App Store. However, the company temporarily limited registrations later that day, citing a large-scale cyberattack on its services.

DeepSeek uses Nvidia’s less advanced AI chips, H800s for its LLM training. The US has been tightening AI chip exports to China, with only lower-quality products allowed. Bloomberg reported that Nvidia called DeepSeek an “excellent AI advancement” that complies with US laws on export controls. The US tech giant also emphasised that the work of running AI models nonetheless requires lots of its products. 

Risk for a renewed US-China trade war

US President Donald Trump commented that DeepSeek’s breakthrough as “a wake-up call” for American tech companies and the need for the US “to be laser-focused on competing”. Meanwhile, he expressed confidence that the US tech companies would maintain dominance in the AI industry. 

Trump’s return reignites concerns about a US-China trade war No 2. “For now, it can confidently be said that today has sparked a re-think of the ‘AI trade’ and risks of a US-China trade war,” Rodd noted.

So far, President Trump has not imposed the 60% tariff on Chinese imports he pledged during his campaign. Instead, his administration has initiated an assessment, with findings expected by 1 April.