Why DeepSeek’s breakthrough could blow up an AI market bubble
DeepSeek’s new AI model made history on Monday by triggering US markets and wiping billions off leading tech stocks. Yet, its breakthrough may have a positive impact on the AI landscape.
The Chinese startup’s new chat-bot, a rival to ChatGPT’s OpenAI, gained traction due to its ability to deliver high performance, at a far lower price.
Its recent model was recently developed for only $6m (£4.8m), just a fraction of the $100m (£80m) usually spent by its competitors.
This drove Nvidia, a key driver of the AI stock boom, to make history by experiencing the biggest one day loss of any company in history, suffering a $589bn (£493bn) drop.
Meanwhile, Nasdaq 100 futures fell 3.2 per cent, with the S&P 500 also dropping 1.9 per cent early Monday.
As well as triggering the market, the platform also poses major security and data risks.
By leveraging the app, China is on its way to revolutionising cyber warfare, posing significant threats to the US and the West.
This week, CityAM revealed how the app’s guardrails prevent answers on sensitive topics to the Chinese Community Party, including the 1989 Tiananmen Square massacre, the treatment of Uyghur Muslims, Taiwan and free speech in the country.
Despite causing initial market jitters, analysts have viewed DeepSeek’s rise as a net positive for the general sector.
By significantly reducing the resources required for AI training and use, the company has lowered costs by over 90 per cent.
Why DeepSeek could be a good thing after all
Increasing AI efficiency could make it more pervasive, as reduced costs typically lead to an increase in demand. In turn, overall consumption may rise too.
Analysts are drawing parallels between this development and previous infrastructure capital expenditure cycles.
For example, Peel Hunt has likened it to past tech booms, saying that the commoditisation of AI could benefit emerging tech companies.
This “should accrue value to companies yet to be built, as in the past”, wrote the city broking group.
It urges UK investors not too overlook companies’ whose value propositions are currently unaffected by AI advancements, such as Raspberry pi, who is increasing tech spend as a percentage of its overall expenditure.
The enhanced efficiency in AI training and use is anticipated to accelerate adoption across multiple industries, expanding the market as a result.
Tom Lee at fundstrat global advisors also called DeepSeek’s effect on the stock market an overreaction.
“This pullback in Nvidia will prove to be a buying opportunity”, he told CNBC.
Competitors such as OpenAI reacted, saying it found evidence that the startup used its models to train its own competitor.
It said it has seen evidence of ‘distillation’ which it told the Financial Times it suspects to be from the new chat-bot owner.
This surge in AI could drive demand for hardware providers like Nvidia, as businesses look to upgrade their tech infrastructure to support advanced AI.