SpaceX closes $40 billion debt deal for AI chips, Elon Musk says

Elon Musk announced that SpaceX has secured $40 billion in debt financing aimed at acquiring graphics processing units (GPUs) essential for artificial intelligence (AI) systems. In a message shared on X this past Sunday, Musk stated that the financing deal has been finalized and emphasized that SpaceX is now classified as investment grade, generating substantial revenue from its Starlink satellite internet services and rocket launches.

In a subsequent update, Musk explained that the financing, which primarily consists of callable debt intended for the procurement of GPUs, is projected to be delivered in the first and second quarters of 2027. Bloomberg previously reported, citing sources familiar with the negotiations, that SpaceX had been in discussions with various banks and investors regarding raising $40 billion specifically to buy Nvidia chips.

The financing plan involves the use of a special purpose vehicle, which would enable SpaceX to keep this debt off its balance sheet while maintaining its investment-grade credit ratings. Goldman Sachs is reportedly leading the discussions with investors, including Apollo Global Management, to establish the financing structure.

This financing arrangement highlights the significant capital demands stemming from the rapid growth of the AI sector, as technology firms invest heavily in advanced chips and data centers necessary to support increasingly sophisticated AI models. In recent months, SpaceX has made substantial strides into the AI domain, branding this effort as "Super Intelligence" or SI.

The AI segment has rapidly evolved into SpaceX's largest business unit, with third-quarter revenues skyrocketing ninefold to $7.56 billion, a notable increase from $813 million during the same period the previous year, as indicated in the latest earnings report. Musk also remarked on the current state of debt markets, noting that access to financing has largely diminished for companies that lack significant revenue streams outside of AI data centers.

He forecasted that over 200 so-called neocloud companies, which are designed to provide cloud computing infrastructure tailored for AI and machine learning workloads, may face widespread failures and eventual consolidation, with only a few likely to survive in the competitive landscape.