Elon Musk Reportedly Seeking Customer Data From Dead Startups to Train AI

AI needs data to train, and data costs a good penny. Seems like all the user data on X and all the data SpaceX has ever produced hasn’t been enough for Elon Musk’s Grok, because he has a new solution to the conundrum: buying customer data from defunct companies for cheap.

Left scrabbling for data, Elon Musk’s SpaceX is allegedly thinking about buying the customer and operational data left behind by struggling or bankrupt startups, according to sources who spoke with Bloomberg .

SpaceX needs cheaper data options, fast. In its first earnings report last month, SpaceX announced a whopping $15.8 billion commitment to its AI efforts, even though the company’s AI division xAI ended the quarter with a $1.3 billion net operating loss. Musk has lofty plans for the company’s Grok chatbot to catch up to rival models from other AI labs like OpenAI and Anthropic. The company is set to unveil Grok 5 before the end of the year, a model that Musk has claimed will reach AGI-level. Now, I’m no computer scientist, but I know that requires a lot of training data to reach.

The talks are apparently informal, and might not even result in a deal. If they do, it’s bound to have serious privacy ramifications. Any company that you may have given your data to, and is now struggling financially, might end up selling it to Musk.

SpaceX isn’t the only tech giant looking to get its hands on some data, ethical grey areas and privacy concerns be damned. As agentic AI makes its way into the workplace, AI labs have been increasingly valuing industry-specific knowledge across sectors. Facilitating failed companies’ sale of employee data to AI labs has become a niche market of its own.

In a bankruptcy auction last month, Google made a $10 million bid to win the data of the now dead-and-buried Spirit Airlines, ultimately to be used to train its Gemini models. Even though Spirit did not include passenger information, the data sold included 100 million company emails and 500 million Microsoft Teams chats belonging to former employees. It also includes aircraft operations and employee productivity data, 1 million time-card records, 17 million individually owned Microsoft OneDrive items, and hundreds of thousands of employee records, including tax forms, employment contracts, and litigation files.

Google assured that the data would be scrubbed of personally identifiable information, but that wasn’t enough for some parties included in the data. A union representing former Spirit employees, called the Association of Flight Attendants, filed an objection with the U.S. Bankruptcy Court in the Southern District of New York.