The pitch for orbital compute is straightforward: put data centers in space, eliminate terrestrial land and power constraints, serve a global footprint from above. The reality is considerably more complicated.
The physics are not the problem. Low Earth orbit is a viable location for processing hardware. Thermal management is harder than on the ground, but solvable. Latency to surface users is competitive with transoceanic fiber for certain use cases. The orbital environment is not fundamentally hostile to the concept.
The economics are. Launch costs, even at post-Falcon 9 prices, make orbital infrastructure orders of magnitude more expensive per unit of compute than ground-based alternatives. Terrestrial data centers are also improving rapidly in density and efficiency. The gap between what you can deploy on the ground and what you can afford to put in orbit is not closing quickly enough to make the business case work in most scenarios.
The use cases that do make sense are narrow: military ISR processing, communications relay for contested or remote environments, and potentially certain edge-compute scenarios where geography or denial-of-access requirements favor space. These are real markets. They are not mass markets.
What is being sold to investors is often the broader pitch — orbital compute as a general alternative to terrestrial cloud infrastructure — which is much harder to justify. The hype cycle here has outrun the engineering reality by several years, which is not unusual in the space industry but does not make it less misleading.
Watch what the defense contractors are buying, not what the venture-backed startups are announcing. The former tracks genuine operational need. The latter tracks narrative momentum.