Twitter CEO Elon Musk buys thousands of GPUs for major generative AI project: Report

Although Twitter CEO Elon Musk has called for a stop to AI training across the industry, it has been reported that he has initiated a significant artificial intelligence endeavour within Twitter. As part of the project, which focuses on a large language model (LLM), the company has acquired roughly 10,000 GPUs (graphics processing unit) and enlisted AI experts from DeepMind, according to a report from Business Insider.

Two anonymous sources in the report have revealed that the CEO of Twitter has acquired almost 10,000 GPUs for utilisation on the platform. As a rule, GPUs are employed in the handling of large-scale AI models due to the enormous computational power that the technology demands. This decision seems to be in line with a tweet from Elon Musk on March 18, in which he expressed the intention to apply AI for the purpose of "detecting and drawing attention to the manipulation of public opinion" on Twitter.

In the months ahead, we will use AI to detect & highlight manipulation of public opinion on this platform.

Let’s see what the psy ops cat drags in …

— Elon Musk (@elonmusk) March 18, 2023

The specifics of the GPUs that Twitter has obtained are currently unknown. Nevertheless, the company is said to have invested tens of millions of dollars in acquiring these computational GPUs, despite facing ongoing financial difficulties, which Elon Musk has described as an "unstable financial situation".

It is anticipated that the GPUs will be deployed in one of Twitter's two remaining data centers, with Atlanta being the most probable location. Curiously, Musk shuttered Twitter's primary data center in Sacramento towards the end of December, a move that would undoubtedly have reduced the company's computational capacity.

Twitter expanding its engineering team: Report

Along with the procurement of GPU hardware for its generative AI project, Twitter is also expanding its engineering team. The company has recently brought on board Igor Babuschkin and Manuel Kroiss, who are both engineers from AI research firm DeepMind, a subsidiary of Alphabet. It appears that Elon Musk has been actively searching for AI talent to compete with OpenAI's ChatGPT since at least February of this year.

These recent developments have occurred just a few weeks after Elon Musk, along with thousands of other researchers in the tech industry, signed an open letter calling for a temporary halt in the development of AI, citing the risk it poses to humanity.

Moreover, Musk had also expressed his concerns regarding AI back in 2017 while addressing the United States National Governors Association. At the event, he warned regulators that AI research needs to be regulated "before it's too late".

Large companies such as Twitter typically purchase hardware in bulk, which allows them to take advantage of special rates. For instance, when Nvidia's H100 boards are purchased individually from retailers such as CDW, they can cost more than $10,000 per unit. This suggests that Twitter may have spent a substantial amount of money on hardware for its AI initiative.