The rocket company just became a data center landlord, and Wall Street thinks it's worth more than the rockets.
The Summary
- SpaceX has been in talks with Microsoft to lease AI computing capacity, marking a potential pivot into infrastructure-as-a-service that could rival its satellite internet business
- TD Cowen initiated coverage with a Buy rating and $200 price target, citing AI compute leasing as a revenue stream that could redefine the company's financial profile
- xAI is already executing this playbook, turning surplus GPU capacity into a leasing business while it scales its own infrastructure, proving the model works
- The shift signals that compute capacity, not just connectivity or launch services, is becoming the strategic asset in the race to build Web4
The Signal
SpaceX isn't just launching satellites anymore. The company that redefined space access is now positioning itself as a supplier of the raw computational power needed to train and run AI models. Discussions with Microsoft about leasing AI computing capacity suggest SpaceX sees its infrastructure investments as something more than a cost center for internal use.
The timing makes sense. Building AI training clusters requires massive capital outlays for GPUs, power infrastructure, and cooling systems. These assets sit idle between training runs. SpaceX has already built distributed compute infrastructure to support Starlink's network operations and has access to power and real estate at scale. Leasing that capacity to hyperscalers like Microsoft turns fixed costs into revenue generators.
"SpaceX's AI compute leasing could significantly diversify its revenue streams, potentially rivaling its core businesses like Starlink."
Wall Street is paying attention. TD Cowen's $200 price target and Buy rating specifically call out AI compute leasing as a major driver of future valuation. That's not a side bet. That's analysts saying the infrastructure play could be as valuable as the rockets and satellites combined. The firm also flagged execution risks and high capital demands, acknowledging this isn't a low-stakes move.
SpaceX isn't alone in seeing this opportunity. xAI, Elon Musk's AI company, is already leasing surplus GPU capacity to other AI labs while it builds out its own training infrastructure. This dual strategy of scaling internal capacity while monetizing excess inventory creates a financial cushion and proves market demand. The xAI model relies on flexible contracts, which introduces revenue volatility but also allows rapid scaling when demand spikes.
Here's the deeper pattern: companies with access to power, space, and capital are becoming infrastructure banks for the AI economy. Amazon did this with AWS, turning internal server capacity into the backbone of the internet. SpaceX could do the same for compute-intensive AI workloads. The difference is speed. AWS took years to mature. AI compute demand is hitting exponential curves right now.
Key leasing economics:
- GPU clusters cost hundreds of millions to build but generate revenue immediately when leased
- Hyperscalers like Microsoft need more capacity than they can build fast enough
- Short-term leases create revenue volatility but allow infrastructure owners to capture premium pricing during demand spikes
The Implication
If SpaceX closes deals with Microsoft and other hyperscalers, expect more aerospace and satellite companies to look at their infrastructure portfolios differently. Compute capacity becomes a tradable asset, not just an operational necessity. For AI companies scrambling to secure GPUs, this opens new supply channels outside the traditional cloud oligopoly. For SpaceX, it's a hedge. If launch demand softens or Starlink growth plateaus, compute leasing becomes a standalone business line with its own growth trajectory. Watch who else enters this market. Companies with power infrastructure, distributed facilities, and existing hardware deployments have the raw materials to compete. The question is whether they see the opportunity before capacity tightens again.