Musk snapped up 10,000 GPUs and entered the AIGC war

News on April 12, according to foreign media reports, Twitter CEO Elon Musk recently purchased about 10,000 GPUs (graphics processing units) for use in one of the company's two data centers. The purchase shows Musk's "commitment" to this work, especially considering that Twitter has little reason to spend so much money on data center-class GPUs if it doesn't intend to use them for AI work.

Currently, GPUs can be very expensive. With an estimated 95% market share, NVIDIA produces GPUs for large AI models for $10,000 (currently about 68,800 yuan). People familiar with the matter said that Musk has repeatedly said that Twitter's financial foundation is not stable enough, so it is likely that he spent tens of millions of dollars to add so many GPUs.

The project reportedly involves creating a generative AI that Twitter will train on based on its own massive amounts of data. Twitter was founded in 2006 and has accumulated a database for almost 20 years, which is a great tool for Twitter to enter the LLM field. It's unclear how Twitter will take advantage of the technology.

One use could be to improve search, and Musk has complained about a poor experience with the feature. He even hired security developer George Hotz to advance a three-month project to "fix" the search function. However, Hotz left after only 1 month on the job.

People familiar with the matter also revealed that another possible use of Twitter AIGC is advertising, a business that is the largest pillar of revenue for social media platforms. Since Musk took over and changed many policies, Twitter has struggled to attract advertisers. After training, AIGC has the ability to create new ad images and text to target specific audiences.

In March, Musk posted a tweet with only one letter: X. Meanwhile, an April court filing shows that Twitter Inc has officially merged with X Corp, meaning that Twitter no longer exists and now only X Company. So far, the tweet is Musk's only public statement about the change. It's not clear what this change ultimately means. Analysts speculate that this is the next step in Musk's plan to launch a "universal app" similar to China's WeChat. Some argue that merging Twitter into X shows Musk is serious, and he has said it was a "good idea" to close all of his companies and merge them into a parent company called X, including Twitter, SpaceX, Tesla, Neuralink and The Boring Company.

In addition to spending thousands of dollars to buy GPUs to build large models, Musk is also vigorously grabbing people.

At present, two DeepMind engineers have been successfully recruited: one is Igor Babuschkin, engaged in AI system research, is a senior scientific research engineer at DeepMind, who has worked in DeepMind for a total of about 5 years, and jumped to OpenAI for a year and a half. The other is Manuel Kroiss, a software engineer at DeepMind for six years and, before that, a solution engineer at Google. Musk hired him for the position of senior director of software engineering at Twitter.

It's worth noting, however, that Musk has been an outspoken critic of OpenAI, an artificial intelligence research institute he co-founded in 2015. Recently, he took the lead in signing an open letter calling for a 6-month moratorium on AI development. "I'm still confused about how my $100 million nonprofit somehow turned into a $30 billion market cap for-profit organization. If this is legal, why not everyone does? In a recent tweet, Musk criticized OpenAI's for-profit subsidiary, OpenAI Limited Partnership.

The inside story of Musk's past with OpenAI

Recent reports by emerging media outlet Semafor revealed that Musk's feud with OpenAI is more for personal reasons. In 2018, Musk told OpenAI co-founder Sam Altman that the lab was too far behind Google. Musk suggested that he should be the one running the company, an offer that Altman and the other founders of OpenAI rejected.

The power struggle led to Musk's departure from OpenAI, despite public statements from both sides that Musk left because of a conflict of interest involving Tesla. At the time, OpenAI said Musk would continue to fund its research. However, according to Semafor, Musk's funding stopped after he left, despite his pledge to provide the company with about $1 billion. The sudden shortage of funds left OpenAI eager to raise cash. In 2019, the organization announced that it would create a for-profit subsidiary to ensure the funding needed for its work. That same year, the company announced a $1 billion investment from Microsoft. When OpenAI opened ChatGPT to the public last November, Musk was reportedly "very angry." A month later, he cut off OpenAI's access to Twitter data.

In fact, AIGC (artificial intelligence generated content) has always existed, but it was not until this year that it was popular with capital, firstly, because of the maturity of the technology, and secondly, the capital that originally focused on the commercialization of visual AI turned around and found that overseas NLP companies like Jasper.ai began to make significant profits.

Due to the advantages of creating digital content, AIGC technology has been highlighted by enthusiasts who have followed the metaverse in the past year as a tool to build the metaverse of the future. But behind the gimmick, more AIGC practitioners believe that AIGC can create the next generation of digital worlds faster than the metaverse, a new track that completely belongs to AIGC.

*Disclaimer: This article is the original author's creation. The content of the article is his personal opinion, we reprint only to share and discuss, does not mean that we agree or agree, if you have any objections, please contact the background.