USA: Meta has reportedly set up four “war rooms” of engineers to figure out how DeepSeek, an AI startup backed by Chinese hedge fund High-Flyer Capital Management, recently launched its open-source AI chatbot R1 at a fraction of the cost, Fortune reported.
Meanwhile, Meta CEO Mark Zuckerberg announced plans on Friday to invest up to US$65 billion (S$87.9 billion) to advance the company’s AI goals.
This announcement came just days after OpenAI revealed its US$500 billion (S$676.2 billion) AI project, Stargate. The White House is backing this project, which will build new data centres across the US in partnership with SoftBank, Oracle, and others.
According to The Information, Mathew Oldham, Meta’s AI infrastructure director, has reportedly told colleagues that DeepSeek’s latest model could outperform the upcoming version of Meta’s Llama AI, which Mr Zuckerberg has said might launch in early 2025.
The report, citing two sources familiar with Meta’s efforts, also revealed that Meta has set up four “war rooms” to address DeepSeek’s potential breakthrough.
According to one source, two of these teams will focus on understanding how High-Flyer reduced the cost of training and running DeepSeek and apply these strategies to Llama.
As per the report, the other two teams will focus on determining which data DeepSeek used to train its model and exploring how Llama’s models can be adjusted based on DeepSeek’s attributes.
A Meta spokesperson told The Information: “We regularly evaluate all competitive models in our development process and have done so since [the company’s] Gen Al [group] was formed. Llama has been foundational in establishing the ecosystem for open-source AI models and we couldn’t be more excited to extend this leadership with the upcoming release of Llama 4.” /TISG
Featured image by Depositphotos (for illustration purposes only)