Do data centers in space make any sense? Time for a rare summer-time Geek Out! Richard chats with Carl about all the hype surrounding building orbital data centers for AI workloads. Richard points out that enthusiasm for this idea surged in the fall of 2025, when the backlash against ground data centers peaked. But could you actually make the orbital data centers work? The conversation works through a reasonable satellite design, covering off the details of power, cooling, communications, and satellite management. But how many satellites would be enough? This leads to an exploration of Kessler Syndrome, where orbital debris gets out of control - and what we can reasonably do about it. Which leads to another idea - how do we make ground-based data centers not suck?