Twitter has procured 100,000 GPU units, potentially to train its generative-AI project

According to a report by Business Insider, Twitter has discreetly procured 100,000 GPU units, potentially to train its auto-generative artificial intelligence (AI) system.

The details of the GPUs acquired by Twitter have not been disclosed, but it is speculated that they may be NVIDIA’s newly launched G100 GPU accelerators, to be installed in Twitter’s data center located in Atlanta.

In addition to purchasing GPUs for AI training, Twitter has recently been headhunting AI specialists from companies like Google and DeepMind. This demonstrates Twitter’s plans to expand its AI technology applications, which may include content recommendations, search features, interactive chatbots similar to those offered by competitors, or even automatically generate engaging content for users.

Prior to this, Elon Musk, who owns Twitter, had endorsed a call for companies to temporarily suspend the deployment of large-scale natural language models for at least six months. He also suggested establishing proper AI technology guidelines, but this has not halted his own endeavors in AI development.