Elon Musk reportedly launched AI startup to compete with OpenAI, creating new large-scale compute infrastructure demand.
Elon Musk is reportedly seeking to raise funding for a new artificial intelligence startup to compete with OpenAI LP, according to reporting from the Financial Times and Wall Street Journal. The Wall Street Journal reported that Musk has formed a company headquartered in Nevada called X.AI Corp, with Musk as the only listed director. Jared Birchall, director of Musk's family office, serves as its secretary, according to a filing from last month. X.AI has authorized the sale of 100 million shares for the privately held company. Business Insider earlier suggested that Musk intends to build a large language model, similar to systems like OpenAI's GPT-4, which are among the most advanced AI systems on the market.
The Tesla chief executive reportedly began planning the startup launch earlier this year and has since recruited a half-dozen AI engineers to the venture, including Igor Babuschkin, a former employee of Alphabet Inc.'s DeepMind machine learning unit. Beyond recruiting engineering talent, Musk is also searching for investors, reportedly holding fundraising discussions with investors who have previously backed SpaceX Corp. and Tesla. While the specific amount of capital being sought remains unclear, the sum is likely to be significant, as developing large language models such as ChatGPT is an expensive endeavor requiring large amounts of hardware. Amazon CEO Andy Jassy stated this week that "really good" AI models can take billions of dollars to develop.
It is believed that Musk's AI venture may use content from Twitter to train its neural networks, utilizing "thousands" of Nvidia Corp. graphics processing units. A recent report indicated that Musk has purchased 10,000 GPUs. Nvidia's newest data center graphics card, the H100, can train large AI models up to nine times faster than the previous-generation A100 chip. Rather than using Nvidia's prepackaged appliances that combine GPUs with supporting infrastructure, Musk's approach of deploying standalone GPUs with custom components may enable greater customization of AI infrastructure—a strategy Tesla already employs.
As of last year, Tesla operated two AI supercomputers optimized for AI workloads. The first featured 7,360 of Nvidia's A100 graphics cards as of last August. Tesla's other AI supercomputer is based on a custom chip called the Dojo D1, which combines elements of a CPU with circuits optimized for machine learning, and includes a proprietary networking protocol dubbed the Tesla Transport Protocol for sharing data between Dojo D1 chips. Musk's new AI venture may implement similar technologies developed by Tesla.
The startup's expected launch would increase competition in an already crowded market. OpenAI, the most prominent player in the nascent generative AI segment, recently received a $10 billion investment from Microsoft Corp. Google LLC and Meta Platforms Inc. are also developing generative AI systems alongside multiple well-funded startups. This expansion comes weeks after Musk co-signed a public letter urging a pause on the development of neural networks more advanced than GPT-4, asking the tech industry to halt development for at least six months.