The Pair Program

The episode contains several genuinely instructive technical insights about 3D vs 2D imaging, multi-static radar, and the cost/quality economics of disaggregated apertures, but a meaningful chunk of the 72 minutes is consumed by coffee icebreakers and a long rapid-fire personal segment.

“with a disaggregated aperture, you get exponential quality the more you add to it, but the cost is linear because you're just adding more small satellites”

“there's a missing component which Dave has really been pushing on quite a bit here, which is the temporal resolution component”

The temporal-resolution-as-the-forgotten-dimension framing and the back-projection-algorithm-finally-feasible-with-GPUs story are fresh and non-obvious, though some of the broader 'small sats replacing exquisite systems' narrative is now well-circulated in defense tech.

“this algorithm called back projection... they immediately said, this is ridiculous. Nobody will ever do back projection”

“multi-static radar systems for like 50 years and nobody's built one. And the reason we haven't built one is because it's too expensive”

Both guests are highly relevant practitioners: a venture-backed founder/CEO who is an aerospace/radar engineer that built ~$4B of radar hardware at General Atomics, and a former NGA Chief Strategy Officer with ~30 years in national security who founded the agency's commercial operations.

“co-founder and CEO of Array Labs”

“Former director of strategy and commercial at NGA, National Geospatial Intelligence Agency”

Strong concrete detail in places - named companies, satellite dimensions, dollar figures, funding, headcount, and specific use cases - though the commercial value claims often stay at the 'this would be amazing' level without hard customer metrics.

“We built about 4,000 of those radars. It was... $1 million per radar. So we built about $4 billion worth of radar hardware”

“now have 35 employees. We've raised $35 million”

The hosts ask reasonable framing questions and one good data-handling follow-up, and notably let the guests interview each other, but they rarely push back or challenge claims, leaving optimistic statements unexamined; the back half is a soft rapid-fire personal segment.

“How does your team handle all the downstream storage processing and handling of that data? And then maybe as a follow-up there, how's AI changing that?”

“is this like a bottleneck more about like the - looks like the physics of the sensors themselves or how we kind of process the data?”