# Skydio made irreversible bets before the technology was proven Markdown copy of https://realalpha.app/s/skydio-made-irreversible-bets-before-the-technology-was/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy. - Speaker: [Adam Bry](https://realalpha.app/builders/adam-bry/) · Adam Bry is cofounder and CEO of Skydio. · Co-founded Skydio, building autonomous drones from research pioneered at MIT and Google’s Project Wing. - Host: [Ti Morse](https://realalpha.app/people/ti-morse/) · Host, Relentless · Interviews hard-tech founders on Relentless; guests include Sam Altman, Brian Armstrong and Hadrian's Chris Power. - Show: Relentless - Episode: [The Drone Startup That's Outcompeting China | Adam Bry, Skydio](https://realalpha.app/v/nb7EinVtdok/), Nov 8, 2026 - Moment: 42:34 to 45:54 of the episode (3 min), playing on https://realalpha.app/s/skydio-made-irreversible-bets-before-the-technology-was/; the episode from there: https://realalpha.app/v/nb7EinVtdok/?t=2554 ## Quote > “And we built the rolling shutter cameras into the drone **before we had solved the algorithms**.” > — [Adam Bry](https://realalpha.app/builders/adam-bry/) at [45:00](https://realalpha.app/v/nb7EinVtdok/?t=2700) · Relentless ## Takeaways 1. Skydio bet on autonomous flight when experts were skeptical. 2. It built rolling-shutter cameras before solving the algorithms. 3. Public safety became a beachhead for dock-based drones. ## About this segment Adam Bry recalls bets that could have ended the company, from computer-vision autonomy to rolling-shutter cameras and drone docks. Their logic: commit when the product's future depends on solving a hard problem. From “The Drone Startup That's Outcompeting China | Adam Bry, Skydio”, Nov 8, 2026. [Watch the full episode](https://realalpha.app/v/nb7EinVtdok/) ## Transcript Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there. [42:38](https://realalpha.app/v/nb7EinVtdok/?t=2558) I had a conversation with Chris Power probably three weeks ago from Hadrian. And I was talking about like one-way and two-way doors, where a two-way door, you know, you can make a decision and walk forward and then walk back if it's not the right one. And one-way doors, you walk in and you can't walk backwards. And he had this interesting idea where he said, yes, there are one-way and two-way doors, but in a lot of cases, especially with hardware companies, you have to effectively just walk in the one-way door and like roll a hard six where you have to bet the company and then just be right um what have been situations you know like the doc example where there was a lot of you know there's a lot of quote-unquote mixed feelings about making those bets but you knew that they were the right ones and just bet the company on it uh yeah i i generally agree with [43:28](https://realalpha.app/v/nb7EinVtdok/?t=2608) that and i think that may be true for non-hardware companies as well i just think for hardware companies, it's a little bit more visceral and clear. I mean, there's been a whole series of key decisions where if it didn't work, we really had no, like the company just wasn't going to work. You can go all the way back to the beginning. I mean, the big, big bet that we made when we started the company was that you could build a reliable, capable, autonomous drone based on computer vision which uh in 2014 was very unclear like even experts in computer vision research labs and robotics were skeptical of that claim because the stuff barely worked in the lab. And that was, you know, if that turned out to be wrong, there was going to be no Skydio. A subtle one, but a critical one, when we developed Skydio 2, [44:16](https://realalpha.app/v/nb7EinVtdok/?t=2656) which was our second generation drone, we knew we needed to simultaneously get like better three-dimensional sensing of the world around us in a lower cost way in order for our economics and products to make sense. And so we shifted from using what are called global shutter cameras, where basically the whole frame is exposed at the same instant, which is what historically basically all had been used for computer vision, to rolling shutter cameras, which is what go into our phones and basically everything, where every line is exposed one row at a time. And so if the camera's moving, like if you take a video with your cell phone where it's moving, you'll see this skew across the image because it's a rolling shutter sensor. So at the time when we started working on this in 2018, there was almost no computer vision done on [45:00](https://realalpha.app/v/nb7EinVtdok/?t=2700) rolling shutter cameras because the math just kind of gets explosive. And we built the rolling shutter cameras into the drone before we had solved the algorithms. And there was no turning back from that, right? Like we had to make that decision to bake it in and just say like, you know, we think we're going to be able to solve this. And we did. And we got, not only did we get to like parity from the old capability, we took a huge step up. So that was a really big one. I think the focus on dock was a really big one that's worked out for us. And then I think that we correctly recognized that public safety and emergency response was going to be the first kind of dock infrastructure use case to reach substantial scale and that it would be a great beachhead and kind of lighthouse to other applications. And we went all in on that a couple of years [45:49](https://realalpha.app/v/nb7EinVtdok/?t=2749) ago, which has also worked out super well for us.