SpaceX chip lenders reportedly ask that no chips go to orbital data centers

TechCrunch looks at an obstacle to space data centers that gets less attention than engineering or launch capacity: paying for the chips.
The terrestrial data center business runs on debt. Cloud providers borrow to build data centers and repay the loans with a share of the revenue. Those deals typically use the chips as collateral. If a data center defaults, lenders can take possession of the chips and resell them, and that backstop makes financing cheaper. Even older chips hold value. According to Compute Credit Index Research, some H100s are selling for about half of their original cost basis from 2023, and some H200 configurations are selling for the same price they had when introduced in 2024.
That logic breaks down in orbit. As the article puts it, you are not going to repossess chips installed in a satellite 300 miles above Earth, travelling at 17,500 miles per hour. The Financial Times reports that lenders behind SpaceX's $40 billion chip financing deal asked that none of the chips purchased be used for the company's orbital data center plans. TechCrunch says it is not clear whether the lenders got SpaceX to agree to that provision.
The article suggests this may not matter to SpaceX in the near term. Its ability to deploy compute at terrestrial ground stations has proven lucrative, and on the most recent earnings call CEO Elon Musk described sending engineers from the rocket team to speed up that build-out, including casting turbine blades for generators faster than existing suppliers. Open questions remain on the ground too: how easy it will be to find land and power, with many projects facing delays and local opposition, and what happens to compute demand and to buyers' ability to turn it into revenue.
Musk began saying at the beginning of 2026 that he believes space will be the "cheapest" place to operate data centers in two to three years. TechCrunch calls that unlikely until SpaceX can fly Starship in a fully reusable configuration. In the most recent flight test, the company failed to recover the booster stage but landed the orbital stage back at its launch facility in Texas. It is not clear when that second stage will fly again or how expensive it will be to prepare it for a second trip to space.
Even with Starship, SpaceX and startups such as StarCloud, Cowboy Space Corporation, Sophia Space and Orbital, among others, will need to buy chips. On top of the already daunting economics, the article says it looks like they will spend more on chips than their terrestrial counterparts.
That could favour Google, which is focused on deploying its own TPU chips in space through Project Suncatcher. SpaceX and Tesla are working together on their own chip foundry in the U.S., called Terafab, but the project has yet to break ground and will need years and billions of dollars before it produces semiconductors at scale.
Key facts
- The Financial Times reports that lenders behind SpaceX's $40 billion chip financing deal asked that none of the chips be used for its orbital data center plans; it is not clear whether SpaceX agreed.
- Terrestrial data center loans typically use the chips as collateral that lenders can repossess and resell, which makes financing cheaper; chips in orbit cannot be repossessed.
- Per Compute Credit Index Research, some H100s resell at about half their 2023 cost basis, and some H200 configurations at the price they had when introduced in 2024.
- Space data center players, including SpaceX, StarCloud, Cowboy Space Corporation, Sophia Space and Orbital, will need to buy chips and look likely to spend more than terrestrial rivals.
- Google's own TPUs (Project Suncatcher) may be an advantage; SpaceX and Tesla's Terafab foundry has yet to break ground.
Why it matters
Space data centers are usually discussed as an engineering and launch problem. This piece adds a financing problem. Cheap debt for chips depends on collateral that lenders can seize and resell, and a satellite moving at 17,500 miles per hour 300 miles up offers no such backstop. If lenders will not finance chips headed for orbit on the same terms, the economics of orbital compute get harder, on top of the question of whether Starship becomes fully reusable. The reported request from SpaceX's own lenders is an early sign that this concern is already showing up in real deals.
Who it affects
SpaceX first, given the $40 billion chip financing deal and Musk's claim that space will be the cheapest place to run data centers in two to three years. Also the startups named in the article: StarCloud, Cowboy Space Corporation, Sophia Space and Orbital. Lenders and cloud providers that rely on chip-backed loans are the other side of the question. Google, with its in-house TPUs and Project Suncatcher, may have an advantage, while the Terafab foundry planned by SpaceX and Tesla has yet to break ground.
How to use it
This is analysis, not a product or a tool. The practical takeaway for anyone evaluating an orbital data center plan is to check how the chips are paid for, not only how they are launched and powered. The article's own benchmark is the terrestrial model: debt secured on chips that keep resale value, such as H100s at about half their 2023 cost basis.
How solid is it
The central fact, the lenders' request, comes from the Financial Times and reaches the reader secondhand through TechCrunch. TechCrunch itself says it is not clear whether SpaceX agreed to the provision. The resale figures are attributed to Compute Credit Index Research. The claim that space operators will spend more on chips than terrestrial ones is the article's own reading ("it looks like"), and it gives no figures for the gap. The view that Musk's two-to-three-year timeline is unlikely without a fully reusable Starship is also the author's judgment.
Risks and caveats
The article says the lenders' request may not matter to SpaceX in the near term, since its terrestrial ground-station compute has proven lucrative. Ground-side risks remain: finding land and power, delays and local opposition, and uncertainty over compute demand and buyers' ability to turn it into revenue. On Starship, the booster was not recovered in the latest test and it is unclear when the second stage will fly again or what refurbishment will cost. Terafab has yet to break ground and needs years and billions of dollars to reach scale.