The $3B Company Trying to Beat Amazon Prime

Doing great. It all blends together to me as well. But it's incredible to be here. What a beautiful venue. And thanks for having me on the show.

Yeah, so Stored really exists to level the playing field with Amazon Prime for the other two-thirds of commerce on the internet. When we look out there, we think that Amazon's product really is delivery. And that's what all other merchants struggle to achieve. And it's what keeps all of us as consumers going to their website to check out instead of independent merchants. So we really see ourselves as the physical intelligence layer combining all the fulfillment and delivery, all of the commerce software and all of the applied robotics and AI to achieve Amazon-level delivery costs, speeds and consumer experiences.

Absolutely. So earlier this year, we announced a $250 million Series F. That was really to not only continue scaling the business rapidly where this year we'll be crossing a billion in annual revenue. And over the last four years alone we've 10x'd the business in terms of scale and revenue and both volume. And so it's a lot of fueling that future growth where there's so much of an advantage of scale. Scale drives economics. It also drives density and that's really what builds the moat in delivery enabling Amazon-level speeds and costs. So continuing to press on the gas hard with scale. We also announced Stored Labs along with it which is very much an application of all of the latest robotics, AI, drones, but also a data foundry actually training the robotics in reverse. And so we're really excited about how really all the new technology is enabling a vertically integrated industrial business like us to just keep separating our moat from the existing industry we live in.

Yeah, if you really look back over history, a lot of the problem with robotics is kind of the dumb robots where if anything goes wrong, if any variable is changed, they can't succeed in that environment anymore. And so if you have an operationally complex environment where every minute or hour could be hundreds of thousands of dollars or more of productivity, that traditional robotic operation can be quite risky. And so now you're getting to this point of agentic robotics where you're having robotics that are building real world models that can see videos for very brief times and learn how to self-heal and get back on track. So, let's say you get a slightly different size or form factor in a unit in a fulfillment center. Does all of a sudden that go down until it can show up and code or do something different, or does the robot learn on the fly with one worker walking up and holding it for 3 seconds and showing them how to put it in the box? And so, that's really where robotics is going and why now is such a moment for it. Yet at the same time, if you look back through history and if you study kind of the Teslas and SpaceXes of the world, the whole process and equation is how do you go from refining your processes physically and operationally to then software and kind of repeat that loop over and over again. And only then can you really get the advancement you want in hardware because now you're applying it and really hard-coding whatever that process was. So if it was under-optimized, you're going to fall behind now in a hard-coded way. That's really the issue with robotics in today's world is a lot of the physical world out there that needs these robots doesn't have the basis of that best-in-class process or vertically integrated software to go apply the robot. And so that's kind of the chasm that robotics is going to have to cross in the period ahead now that they are getting more agentic and self-healing as well.

Yeah, for us we process about 8 billion operational data points a year. And that could be anything from how to place a piece of inventory in the network to how to plan for certain consumer demand to how a unit is picked in the facility with labeling on the hands so that they can learn all of these different picking and packing patterns. There's also post-deployment training which is videos of the robots after they've actually gone live to show those errors and to help them keep learning more and more. And so there's really a ton of data that goes into a business like Stored. And when we're processing about a hundred million packages a year to about a third of US households now, almost 20 billion dollars of GMV, that's really one of the most data-rich environments that all of these labs and robotics companies are trying to either simulate and learn. And that's why we have this not only head start in the e-commerce and logistics industry that we sit in, but also have such an advancement in the robotics in the early phase with these partnerships with leading companies and labs deploying across the Stored network because of that data-rich environment where let's say you deploy a camera to do the post-deployment training on a robot. Well, because of our software, we also know in that video what's the exact dimensions and weight of the box. Where's the weight if it's placed? So, it's going to have balance difference because of the packout process. And when you can enrich these videos with not only labeling of the hands, but that level of vertically integrated data, the learning is just so much faster.

Over time, I think that you'll see 2x plus efficiency output in manufacturing and robotic environments due to really all these different form factors of robots. And I think humanoids are the most eye-catching when you look at a video of it as a consumer or just a human and you see this thing. That's really what gets the attention. But there's a lot of different applications of them because the history was that you could kind of either have like a mounted arm that would pick things or something on the ground like a Kiva that would move things. Not two in one. Well, you don't actually have to have legs for a robot to move on the ground. It doesn't literally have to look like a human to have the same level of... it actually should not look like a human if you want to maximize dexterity. And so I think humanoids are interesting but very much stuck in the pilot phase. And you're seeing a lot of advancement in different levels of robotics as well.

That might be the quote of the day.