Meta threatened by DeepSeek? Mark Zuckerberg scrambles war rooms of engineers to figure out how the Chinese startup outsmarted AI titans at a fraction of the price

Meta is scrambling to respond to a major threat from a small Chinese AI startup DeepSeek, which could rival Meta's technology at a fraction of the cost, as per a report.

Meta sets war rooms

According to The Information report, the company has set up four "war rooms" of engineers to figure out how DeepSeek managed to create an AI chatbot, R1. The new chatbot not only matches the performance of models like ChatGPT but is also much cheaper, as per the Fortune report.

Meta’s Llama AI team is on high alert

Meta AI infrastructure director Mathew Oldham suggested that R1 could even surpass the upcoming version of Meta’s Llama AI, as per The Information. The breakthrough has put Meta's AI team on high alert, with engineers now tasked with deciphering how DeepSeek managed to lower the cost of training and running its model. Meta wants to find out if those strategies could be applied to Llama AI, according to The Information.


Also Read : Donald Trump’s ‘border tsar’ Tom Homan slams Selena Gomez over immigration crackdown video

Two of the four teams in Meta’s war rooms are focused on understanding DeepSeek’s cost-saving methods, as per the report. The other two teams are working on analyzing the data DeepSeek used and exploring ways to adjust Llama’s architecture based on DeepSeek’s model, added the report. Llama AI is expected to be released in early 2025.

Meta regularly evaluates all competitive models during the development process and has done so since Meta's Gen Al group was formed, as per The Information report.

Last week, Meta also announced plans to invest up to $65 billion in AI projects over the next year. The investment includes constructing a massive new data centre and expanding its AI workforce.

Also Read : Blake Lively and Justin Baldoni’s legal tussle gets a March 2026 trial date

FAQs:

Q1. How many war rooms did Meta set up to crack DeepSeek’s model?
A1. Meta has set up four war rooms for engineers to find DeepSeek's model. Meta wants to find out how DeepSeek managed to create a cheaper AI chatbot.

Q2. Will DeepSeek’s R1 surpass Meta’s Llama AI?
A2. Meta AI infrastructure director Mathew Oldham said DeepSeek’s R1 may surpass Meta’s Llama AI. It is expected to be released by early 2025.