Elon Musk called Sam Altman a scammer on X over the weekend, and Altman struck back by questioning Musk's pitch to investors about space-based data centers. The exchange, reported by TechCrunch, is the latest chapter in a public dispute between the two that has been building since February, when Altman first dismissed Musk's ambitions to run artificial intelligence computing from orbit.

Altman's response read: "homeboy you're the one sellling [sic] public market investors on short-term space datacenters." Musk replied tersely: "we start flying them next year." Neither man elaborated, and neither company issued a formal statement.

The core disagreement involves SpaceX's plan to launch orbital data centers—platforms floating in space that would perform AI inference work (the computational step where trained models generate predictions). According to TechCrunch, this concept is the primary driver behind SpaceX's two-trillion-dollar valuation, meaning the company's market value rests heavily on hardware that has never flown.

SpaceX would use Starship, its fully reusable rocket, to launch this infrastructure. The vehicle was expected to attempt its thirteenth test flight around July 16, 2026. Notably, during its IPO road show, SpaceX disclosed that Starship will not achieve full reusability near-term and will need to be expended on each launch—a significant admission for a company whose valuation partly depends on cheap, repeatable rocket access.

Altman's skepticism predates the weekend exchange by months. In February, he called Musk's orbital data center plans "ridiculous" given current demand for AI computing power.

The engineering obstacles are well-known among infrastructure specialists and are worth understanding in context. Orbital data centers face a heat-dissipation problem that ground-based facilities avoid: there is no air to carry heat away, so radiators must shed 100 percent of the computational heat generated, and radiator mass balloons unfavorably when matched against today's dense AI accelerators (the chips that power inference). Add latency delays sending data to ground-based users, the need to radiation-harden chips not originally designed for space, and the complete absence of in-orbit repair capability, and the challenge becomes clear. These are not speculative objections; they form the basis of why Altman's skepticism aligns with expert consensus, per TechCrunch's reporting.

Separating the legitimate engineering critique from the personal animus matters here. Musk and Altman have feuded publicly for years over OpenAI's founding structure and direction. The "scammer" label and the typo make for social media theater, but neither constitutes a technical argument on either side.

The genuinely interesting dimension is the valuation question. A two-trillion-dollar SpaceX valuation that depends substantially on a product category with no demonstrated commercial track record—and which SpaceX's own IPO materials conceded depends on expendable rocket stages, undermining the reusability story that has anchored SpaceX's cost advantage for a decade—creates a gap between narrative and hardware reality. Altman's jab at Musk "selling public market investors" on the concept, stripped of personality, points directly at that gap.

The scheduled July 16 Starship flight, if it launches, will not by itself resolve the orbital data center debate. Flight thirteen is a vehicle test, not a payload demonstration, and SpaceX has not indicated that inference hardware will be aboard. But investors and competitors will read whatever happens—a successful landing, an anomaly, stage loss—as supporting or undermining the broader thesis Musk is asking markets to stake capital on.

In this author's view, the durable story here is not the insults themselves, but the structural fact that SpaceX's headline valuation now rests so heavily on an unproven orbital compute product rather than on its proven launch business. That is a genuinely new dynamic in how technology companies are valued, independent of whether Musk and Altman continue trading barbs.