Zuckerberg Plans Huge Meta Investment into AI
Meta Platforms CEO Mark Zuckerberg unveiled that the company plans CAPEX of $60 billion to $65 billion in 2025.
Chinese AI start-up DeepSeek’s latest AI model beat ChatGPT in downloads on Apple’s US chart. The reports show that the latest model release indicates it is possible to develop powerful models at a much lower cost.
The company introduced its DeepSeek-R1 AI model last week putting it into direct competition with OpenAI’s ChatGPT. The new mobile AI application rose to the top of the free app download listing in Apple’s App Store for the US region and topped the same rankings in China.
The release sent shockwaves through the US tech industry because it is open source and was developed at an exceptionally low cost, according to a China Daily report. The model is powered by less advanced chips than are typically used to run AI workloads. Bloomberg reported that it sparked debate about being more cost-effective and raised doubt on the validity of the high valuations for companies such as Nvidia and ASML.
UBS estimated the DeepSeek-R1 model costs $0.55 per million input tokens and $2.19 per million output tokens, compared with $15 and $60, respectively, for OpenAI’s o1 model. The use of lower-cost chips also questions the need for the many US export controls put in place to restrict China’s access to advanced chips and slow the country’s development of AI tools.
The news sent industry shares tumbling down yesterday. Nvidia lost around $593 billion in value, Reuters reported. Nvidia dropped 17%, as the biggest loser was Broadcom which finished down 17.4%. Microsoft fell 2.1%, while Alphabet ended the trading day with a 4.2% loss in value.