SpaceX may turn to failed startups for something its AI badly needs

SpaceX (SPCX, Financials), Elon Musk's aerospace and satellite company, is reportedly considering an unusual way to get more of the raw material needed to build better AI models.

The company has held internal discussions about buying customer, operational and other data from troubled or failed startups, according to Bloomberg.

The talks remain informal and may not result in any transactions.

Still, the idea points to a growing issue across AI: valuable training data is becoming an asset of its own.

SpaceX has already used information from Musk's social network X, along with internal AI tutors, to help train and improve its software. The company reportedly paused hiring for those tutors earlier this year.

Buying data from struggling companies could give SpaceX another source without having to generate everything internally.

The push comes as AI becomes a larger part of SpaceX's spending.

The company invested $15.8 billion in AI-related capital expenditures during the second quarter. Its AI business generated $2.56 billion in revenue during the period but posted a $1.26 billion operating loss.

That makes the next question more important than where SpaceX finds the data.

Investors will be watching whether that spending can eventually turn the company's expanding AI operation into a profitable business.