A Chinese AI chatbot called DeepSeek has rocketed to the top of the charts on Apple's App Store — beating the ever-popular ChatGPT, among other competitors.

But what had Silicon Valley leaders breaking out in a cold sweat over the weekend were claims that the open-source large language model it's based on, DeepSeek V3, was trained with a fraction of the computing power they've been relying on, developed for what was reportedly less than $6 million.

Western AI models, on the other hand, have sucked up billions of funding, with companies including OpenAI looking to spend hundreds of billions of dollars in the coming years to build out the infrastructure required to train these behemoth models and keep them running.

The timing couldn't be worse. Just last week, US president Donald Trump announced a $500 billion AI infrastructure deal, dubbed Stargate, that involved the ChatGPT maker — alongside investment company SoftBank, tech giant Oracle, and others — spending a planned $500 billion on AI datacenters over the next few years. Not to be outdone, Meta CEO Mark Zuckerberg announced on Friday that the company would be spending a record $60 billion on AI this year alone.

DeepSeek bringing something impressive to market for far less raises an unnerving question for all those check writers: what if AI can actually be done for cheap, without all that epic infrastructure?

AI chipmaker Nivida, in particular, felt the ground shake when trading began on Monday, with shares dropping around eleven percent in early trading. Microsoft, which has invested billions of dollars in OpenAI, slid almost four percent. SoftBank and Oracle plummeted around nine and seven percent, respectively.

DeepSeek claims its latest R1 model has a performance that's on par with OpenAI's o1 model, which was released last fall.

If true — to be clear, there are plenty of experts out there who claim the Chinese outfit may be fudging the numbers — the latest AI chatbot could be a point of reckoning for the likes of OpenAI, who are riding an enormous wave of AI hype and lavish, multibillion-dollar deals.

"DeepSeek’s power implications for AI training punctures some of the [capital expenditure] euphoria which followed major commitments from Stargate and Meta last week," investment bank Jefferies analysts wrote in a note to investors, noting increasing "pressure on AI players to justify ever-increasing capex plans."

Other analyst groups claimed the latest hype surrounding DeepSeek was overblown.

"In short, we believe that 1) DeepSeek DID NOT 'build OpenAI for $5M'; 2) the models look fantastic but we don’t think they are miracles; and 3) the resulting Twitterverse panic over the weekend seems overblown," Bernstein analysts wrote.

Nonetheless, DeepSeek has sent ripples across the AI chatbot landscape and its leaders are broadcasting a clear message.

"OpenAI is not a god, they won't necessarily always be at the forefront," the company's founder Lian Wenfeng told JP Morgan analysts.

OpenAI CEO Sam Altman has so far remained silent on the matter.

It's certainly a story to watch, as ungodly sums of money are being poured into building AI infrastructure in the US. Could the datacenters currently being built already be out of date as soon as they come online? Could OpenAI match DeepSeek's extremely lean operating model? How long would that take?

For now, the Chinese company is certainly an outlier.

"The number of companies who have $6 million to spend is vastly greater than the number of companies who have $100 million or $1 billion to spend," Page One Ventures investor Chris Nicholson told the New York Times.

In short, DeepSeek has thrown down the gauntlet, and investors of US-based AI companies will be asking plenty of questions going forward.

More on Stargate: A Key Trump Goal Emerges: Replacing Human Jobs With AI


Share This Article