Chinese ChatGPT rival storms to No1 on appstore sparking $1trillion market chaos

The launch of a cheap Chinese rival to chatbots like ChatGPT has shocked the tech world and sent stock prices haywire.

DeepSeek was released earlier this month, and has now rocketed to be the top free app on Apple’s store in the UK and the US.

While famous AI models like Gemini and ChatGPT are proprietary, DeepSeek is ‘open source’ and will allow anyone to build on its code.

The other big deal about it is it requires much less computing power to run.

This could all be bad news for the profit margins of big AI companies, who have banked on assuming only those with the deepest pockets will be able to invest in this revolution.

Stocks today were already hit hard by the unexpected development.

Nvidia, the most valuable public company in the world, was on course to see over $300bn of its market value wiped out today, according to the Financial Times.

This would be the biggest recorded drop for any company in history.

Other American AI-related firms such as Microsoft and Meta were also down this morning, with over a trillion dollars set to be wiped off the Nasdaq stock exchange.

Investors are worried that their bets on the company, which makes computer chips used to power AI, may not pay out if similar performance can be achieved with much less power.

Venture capitalist and Trump advisor Marc Andreessen wrote on X: ‘DeepSeek R1 is AI’s Sputnik moment,’ referring to the first satellite put into orbit, a huge milestone in the space race.

He called it ‘one of the most amazing and impressive breakthroughs I’ve ever seen’.

The company claims it managed to train DeepSeek-V3 with less than $6 million, a fraction compared to the money pouring into the industry heavyweights.

China had been thought to be well behind in the AI contest because of blocks on export of the chips used.

In 2021, Joe Biden restricted the sale of the most powerful chips to preserve American dominance in the field.

This led to more cooperation between Chinese AI firms, who shared work with one another to try and boost performance.

But it now seems that necessity bred invention, as DeepSeek may have found a way to run a comparable model with less processing power.

Little is known about the company behind the model, a small Hangzhou-based startup founded in 2023.

It claims to be a side project of hedge fund manager Liang Wenfeng who challenged himself to disrupt the AI landscape by buying up the Nvidia chips avialable and working with what he had.

While the company says it is open source, some doubted this said it is still essentially a ‘black box’ when it comes to transparency, as data on how it as trained is not available.

Dr Sukant Khurana, who has worked extensively with AI and medicine, wrote that the company’s approach, ‘while a step in the right direction for accessibility, falls short of the transparency some people may wrongly want you to believe it offers.

‘Let’s not mistake open weights for open-source; they are not interchangeable terms.’

Get in touch with our news team by emailing us at webnews@metro.co.uk.

For more stories like this, check our news page.

MORE: CIA reveal what they think started Covid-19 pandemic

MORE: The little-known dessert chain quietly planning to take over UK high street

MORE: Is Stephen Hawking’s eerie warning on the future already coming true?