A quick and dirty timeline of the market’s DeepSeekFreak

Mega tech stocks plunged in their worst sell-off of 2025 early Monday, with AI darlings like Nvidia, Palantir, and Broadcom notching some of their biggest stumbles in recent memory.

Associated AI plays, like the energy companies that have soared on expectations of endless power demand for massive data centers, think Vistra and Constellation Energy are getting absolutely smoked.

The catalyst for the cataclysm arose over the weekend, as the US tech cognoscenti began to grow convinced that a new model called R1 released by DeepSeek, the Chinese artificial intelligence company, seemed to pose a major strategic threat to AI arms-race strategy pursued by US tech giants, and by extension, the massive market rally for companies like Nvidia that have been at the heart of this boom.

Here’s a quick refresher on what happened.

Jan. 22 — “ByteDance Unveils Upgraded Model Behind Its AI Chatbot” (Dow Jones)

“DeepSeek, a startup funded by quantitative hedge-fund manager High-Flyer, also officially unveiled its own large language model, R1. DeepSeek claims its performance is on par with OpenAI’s reasoning model, o1.”

Jan. 23 — “China’s cheap, open AI model DeepSeek thrills scientists” (Nature)

“Part of the buzz around DeepSeek is that it has succeeded in making R1 despite US export controls that limit Chinese firms’ access to the best computer chips designed for AI processing. ‘The fact that it comes out of China shows that being efficient with your resources matters more than compute scale alone,’ says François Chollet, an AI researcher in Seattle, Washington.”

Jan. 25 — “Silicon Valley Is Raving About a Made-in-China AI Model” (WSJ)

“DeepSeek said training one of its latest models cost $5.6 million, compared with the $100 million to $1 billion range cited last year by Dario Amodei, chief executive of the AI developer Anthropic, as the cost of building a model.”

Jan. 26

Jan. 27 — “Chinese AI disrupter DeepSeek claims top spot in US App Store, dethroning ChatGPT” (South China Morning Post)

“DeepSeek has integrated the reasoning model into the web and app versions of its chatbots for unlimited use at no cost.

In comparison, OpenAI charges US$200 per month for unlimited access to its o1 models, or a minimum of a US$20 monthly fee for a standard plan that includes limited access.”

Jan. 27 — “A shocking Chinese AI advancement called DeepSeek is sending US stocks plunging” (CNN)

“‘The bottom line is the US outperformance has been driven by tech and the lead that US companies have in AI,’ Lerner said in a note to investors Monday morning. ‘The DeepSeek model rollout is leading investors to question the lead that US companies have and how much is being spent and whether that spending will lead to profits (or overspending).’”

Jan. 27 — “What is DeepSeek? Everything to Know About China’s ChatGPT Rival and Why It Might Mean the End of the AI Trade.” (Barron’s)

“If a top-end model costs millions of dollars, not hundreds of millions or billions, and an API can be offered at 27 times less than what is being sold by OpenAI, the massive expense of the past two years may have been wasted... If what DeepSeek says is true and can be replicated, the catalyst driving the AI bull market would quickly reverse, and could even lead to a market crash.”