AWS Summit NYC 2026 | Vibhor Sood, Burro.ai & Alla Simoneau, AWS

Clips
More from AWS Summit NYC 2026

Vibhor Sood

Co-Founder & VP of Engineering

Burro.ai

Alla Simoneau

Emerging Technology Physical AI Lead

AWS

play_circle_outline AWS-Backed Grande 44 Launches; Burro AI Expands Autonomy From Agriculture Into Industrial Sectors

play_circle_outline Market drivers: global labor shortages, aging workforce, and economic pressures.

play_circle_outline Edge-Cloud Hybrid for Agriculture Autonomy: AWS Greengrass and SageMaker, Vision-Language-Action World Models, Iterative Learning for Anomalies

Info

Vibhor Sood, Burro.ai & Alla Simoneau, AWS

Vibhor Sood

Co-Founder & VP of Engineering Burro.ai

Alla Simoneau

Emerging Technology Physical AI Lead AWS

search

Gemma Allen

>> Welcome to theCUBE Coverage here at AWS Summit 2026. We are at the Java Center. We are bringing all the action on the ground, talking to the builders, breakers, and innovators, forming this next wave of the AWS story. Joining me now with two folks key to that. I'm joined by Vibhor Sood, who is the co-founder and VP of technology at Burro AI, and Alla Simoneau, who is the emerging technology physical AI lead at AWS.

Vibhor Sood

>> So we are an autonomy company. We started out in agriculture because that's where we saw the largest labor shortage in the world. But slowly, over the last two years, we've expanded into industrial settings like CSX, Carvana, Koch Industries, and all of these other companies, which they're just running through this issue where they can't find labor to do tasks. We have over 750 robots across 18 countries. And what we are doing is really augmenting labor. And we have this new product coming out called Grande 44 that we are launching with AWS in collaboration. It's this new frontier of physical AI. It has more power, it has more compute on board. Yeah.

Gemma Allen

>> Wow. Exciting stuff. I mean, 750 bots. That sounds like a lot of ever done, a lot of CapEx too, right? But let's talk a little bit, Alla, about your relationship. I know you run a very interesting division at AWS. We just spoke off-camera shortly about how it's really great to see something so physical because we can all understand and contextualize. Talk about how you guys work together and what your practice does.

Alla Simoneau

>> Yeah. So thank you for having us. Look, the AWS team has been quite committed to advancing AI beyond the digital spectrum. And what we're seeing is a real transformation in the physical world. We're talking about physical products working and engaging amongst humans and not just physical robots like the way that we've historically perceived them. Think about advanced automation and manufacturing that's been there since the 1950s. We're talking about a true autonomy, where our systems, and I don't want to call them robots, but systems are engaging, learning, adapting, interacting with the physical world. We have a practice in the GenAI Innovation Center at AWS focusing around physical AI. About September of last year, we launched what was called the Physical AI Fellowship. It's a premier accelerator around the globe in partnership with NVIDIA and MassRobotics. And the focus of this was to extend the capabilities of these truly breakthrough startups, building at the intersection of robotics and physical AI to advance their solutions to market. Burro Robotics is one of those fellows we've had the pleasure of working with this year. But if we take a big step back, the reason this is so monumentally changing is, and I think you kind of heard this just a moment ago, is that the technology, the conditions, the market, the economics have all come together at this really profoundly interesting focal point in time. So you have cheaper hardware. It's cheaper to buy LiDAR, sensors, all this technology. There's more access to models and data. And then you have the physical, I would say, transformation of society. You have aging populations. You have pretty significant labor shortages all across the globe. And we're talking about like the World Health Organization announced an 11 million shortage in healthcare by 2030 in manufacturing. Just in the United States, there's a shortage of 2.9 million people by 2033. Today, in construction, there's half a million labor shortage, and about 40% of that population who's currently in those jobs is set to retire in the next 10 years. In agricultural installation, we're talking about solar panel install. There's 53,000 jobs shortage today just in the US. You start to see these little compounding challenges. It's hard labor. It's physically demanding. It's difficult to hire, to staff, to keep. All of these factors have come together at a perfectly ripe time for true autonomous solutions, solutions that are learning, engaging with people in its natural environment. This is the transformation we're seeing with Burro, and we're delighted to be working with them and game changing what I think agriculture possibilities could be, but re-imagining what is possible fundamentally across all industries around the globe.

Gemma Allen

>> And I think that's the wonder of this moment is we can reimagine suddenly what is possible and where you have a less humanly constrained opportunities in areas like agriculture. You're so right. And congratulations on becoming an AWS fellow. I'm sure that's no easy feat.

Vibhor Sood

>> Yeah. It's definitely been very helpful. We moved a lot of our infrastructure to AWS in the last month or so. And this is the only way how we see we can scale, because we are looking at growing up to like 1,100, 1,200 robots by the end of this year and doubling that fleet in 2027. And the only way we have been able to do in the past is moving stuff to the cloud and AWS, and we are continuing to do it. You can imagine with 750 robots, we collect... The robots see very random things every single day. These are things that you don't see in a lab. You just see it once you deploy it, once you're running every single day, eight hours a day, six days a week. So they don't work five days a week. It's six days a week and it's in every continent. You're up all the time. So every time the robot sees something, we then figure out if this is new, it gets stacked, uploaded to AWS, gets labeled, a new model is trained, and then pushed out to the robot at the edge so the robot continues to improve every single day.

Gemma Allen

>> So let's talk a little bit about this whole space, right? Physical AI, AI on the edge, it's sometimes referred to edge deployment, because I think what a lot of folks really don't fully understand is robots can't always call home, right? They have to actually sometimes record offline, upload. And the technology and the developments in that space, especially by companies like AWS, are quite innovative in a very short space of time. Edge AI has become a huge deal in what seems like a blink. As someone who reports on tech, I listen to it all the time. Talk about the specific areas of the AWS infrastructure stack that we're kind of innovating and developing specifically towards that goal. Let's do the product side of it a little bit, if you can.

Vibhor Sood

>> Yes. So I think for us, we have a hybrid model. We have some edge compute. We do all of the autonomy on the edge to make sure that even if we lose network, we can still function safely. But when we do have connectivity, we use LTE. So every month we push a new update to the robots, and we do it using IoT Greengrass. We know it's pushed fleet wide and it's pushed across, again, 18 countries at the same time. It takes maybe two days to update the entire fleet, thanks to AWS and how distributed the cloud is. Once new data is labeled, we use AWS SageMaker to then train new models, validate them before they get pushed out to the robot, because we have to make sure that they're better than what the last model was. We use ECS or EC... Every time when the robot is turned down, we push everything that the robot sees to the S3 bucket so that at any point we can go back and look where the robots are. If it misbehaved, how did it do it? Why did it happen? So we can fix as quickly as possible.

Gemma Allen

>> And I think Alla, what's fascinating about a company like AWS, it's just like Burro AI is in this recursive learning mode all the time. So are you, guys, right? Because you're also learning across so many customers who are sending data to your cloud at places we can't even possibly imagine. Talk a little bit about what you are seeing and learning in the field broadly right now, especially as we enter this very interesting moment where it feels like there is somewhat of a split between edge AI and the kind of data center model for AI we're also familiar with.

Alla Simoneau

>> This is a transformational point in time. And I kind of alluded, robotics have been around since the 1950s, and we talk about systematic manufacturing automation traditional area and agentic automation in that environment. What we're now doing is we're pushing the boundaries beyond just a controlled, highly connective environment, into real world scenarios. Talk about a tractor in the middle of the field, an autonomous cargo ship in the middle of the ocean, all of these use cases. A vehicle like a Waymos driving in the middle of the road, they're now interacting with a 360 set of an environment with dataset that doesn't exist in a traditional large language model. And the model doesn't need, like, tell me the 17th President of the United States. What it needs to know is like the temperature conditions, the tactile nature, the sensor, the pressure, the conditions of climate, force factors, all of these data points are not memorized in an encyclopedia anywhere that a model can learn from. And so as you're talking about the transition from a highly controlled digital system to a physical AI, autonomous system operating in the real world, it requires vast amounts of data and information. It's like learning or teaching. It's even requiring a different type of model, like a non-large language model, but a vision-language-action model, a world model, because the kind of data that you need to simulate to understand, to train a model is nothing that we currently have anywhere stored. And so that's kind of what we're seeing. And Burro AI is essentially deploying this 750 robotic fleet to be able to perfect the brain of these systems. So while it's operating and interacting at the edge, the processing, real-time operations happens at the edge, call it like a different time horizon that data gets uploaded back into a model into the brain. It gets optimized, fine-tuned, retrained, pre-trained back to this edge device so it can actually operate better. And the more data that you have, the richer data, the more you can account for key anomalies because we're talking about real world scenarios, like a pigeon could sit down in a device. What do you do? Or there's a weed that's blocking the device from moving. All of those different scenarios have to be learned. And that's what's really exciting is all of the data and training and the real type scenarios that have to be essentially absorbed by these companies to really build safe, resilient solutions in the field. That's what's really magnificently apparent today and where we are-

Vibhor Sood

>> Yeah. By the way, the weed stopping the robot used to happen to us a couple of years ago because the data sets that we are creating, no one has it. So the only way for us to learn is to ship the robots, collect as much data as we can, and then improve as quickly as we can.

Gemma Allen

>> In Ireland, I'm Irish, we see cattle roam into the middle of the street and stop farmers and tractors. There's all sorts of problems out there to be fixed, I can tell you.

Vibhor Sood

>> Exactly.

Gemma Allen

>> So, we're here. It's day one, the main day at AWS Summit in New York. I know it's going to be a lot of announcements. Talk to me a little bit about what you guys are excited about, what you're looking forward to, what's ahead for you for the next, I guess, 12 or so hours, maybe longer if it's a big party afterwards, right?

Vibhor Sood

>> Yeah. I think for us, we are launching Grande 44. It's our next-generation robots. We've partnered with the largest sensor providers in the world. We are bringing 9X improvement in compute, power, reliability across the boat, and we are expanding into Europe in 2027 with it. This is physical AI in the world, doing real work, learning from it, and improving and working alongside people. We are creating a new category that hasn't existed.

Gemma Allen

>> Wow. So exciting.

Alla Simoneau

>> I'm a plus 1,000 to that. I know there's a humanoid somewhere here on the floor. I'm excited about the industry education knowledge and acknowledgement of what we're seeing today, which is a transformation from digital to physical. And these conversations happening now, but really across this entire event, are showcasing a reinvention of possibilities and across all industries and how we live, how we work, how we interact with technology, and what this means for society. It's a really monumental, I think, experience that's happening in the background here with different customer conversations like this one. But I'm excited for what the future holds, and I'm really thrilled that AWS and the Gen AI Innovation Center and Burro are here to be able to showcase a little bit of the art of possible in the space of physical AI.

Gemma Allen

>> I love that, the art of possible in the space of physical AI. Well, Alla and Vibhor, thank you so much for joining us on theCUBE.

Vibhor Sood

>> Thank you so much.

Alla Simoneau

>> Thank you so much for having us.

Vibhor Sood

>> Thank you for having us.

Gemma Allen

>> I'm Gemma Allen coming to you from the Java Center. It's the AWS Summit 2026. We're here with theCUBE. We are covering all the action on the ground, talking to some of the key players in the AWS ecosystem. Stay tuned.

Vibhor Sood, Burro.ai & Alla Simoneau, AWS

search

Gemma Allen

>> Welcome to theCUBE Coverage here at AWS Summit 2026. We are at the Java Center. We are bringing all the action on the ground, talking to the builders, breakers, and innovators, forming this next wave of the AWS story. Joining me now with two folks key to that. I'm joined by Vibhor Sood, who is the co-founder and VP of technology at Burro AI, and Alla Simoneau, who is the emerging technology physical AI lead at AWS.

Vibhor Sood

>> So we are an autonomy company. We started out in agriculture because that's where we saw the largest labor shortage in the world. But slowly, over the last two years, we've expanded into industrial settings like CSX, Carvana, Koch Industries, and all of these other companies, which they're just running through this issue where they can't find labor to do tasks. We have over 750 robots across 18 countries. And what we are doing is really augmenting labor. And we have this new product coming out called Grande 44 that we are launching with AWS in collaboration. It's this new frontier of physical AI. It has more power, it has more compute on board. Yeah.

Gemma Allen

>> Wow. Exciting stuff. I mean, 750 bots. That sounds like a lot of ever done, a lot of CapEx too, right? But let's talk a little bit, Alla, about your relationship. I know you run a very interesting division at AWS. We just spoke off-camera shortly about how it's really great to see something so physical because we can all understand and contextualize. Talk about how you guys work together and what your practice does.

Alla Simoneau

>> Yeah. So thank you for having us. Look, the AWS team has been quite committed to advancing AI beyond the digital spectrum. And what we're seeing is a real transformation in the physical world. We're talking about physical products working and engaging amongst humans and not just physical robots like the way that we've historically perceived them. Think about advanced automation and manufacturing that's been there since the 1950s. We're talking about a true autonomy, where our systems, and I don't want to call them robots, but systems are engaging, learning, adapting, interacting with the physical world. We have a practice in the GenAI Innovation Center at AWS focusing around physical AI. About September of last year, we launched what was called the Physical AI Fellowship. It's a premier accelerator around the globe in partnership with NVIDIA and MassRobotics. And the focus of this was to extend the capabilities of these truly breakthrough startups, building at the intersection of robotics and physical AI to advance their solutions to market. Burro Robotics is one of those fellows we've had the pleasure of working with this year. But if we take a big step back, the reason this is so monumentally changing is, and I think you kind of heard this just a moment ago, is that the technology, the conditions, the market, the economics have all come together at this really profoundly interesting focal point in time. So you have cheaper hardware. It's cheaper to buy LiDAR, sensors, all this technology. There's more access to models and data. And then you have the physical, I would say, transformation of society. You have aging populations. You have pretty significant labor shortages all across the globe. And we're talking about like the World Health Organization announced an 11 million shortage in healthcare by 2030 in manufacturing. Just in the United States, there's a shortage of 2.9 million people by 2033. Today, in construction, there's half a million labor shortage, and about 40% of that population who's currently in those jobs is set to retire in the next 10 years. In agricultural installation, we're talking about solar panel install. There's 53,000 jobs shortage today just in the US. You start to see these little compounding challenges. It's hard labor. It's physically demanding. It's difficult to hire, to staff, to keep. All of these factors have come together at a perfectly ripe time for true autonomous solutions, solutions that are learning, engaging with people in its natural environment. This is the transformation we're seeing with Burro, and we're delighted to be working with them and game changing what I think agriculture possibilities could be, but re-imagining what is possible fundamentally across all industries around the globe.

Gemma Allen

>> And I think that's the wonder of this moment is we can reimagine suddenly what is possible and where you have a less humanly constrained opportunities in areas like agriculture. You're so right. And congratulations on becoming an AWS fellow. I'm sure that's no easy feat.

Vibhor Sood

>> Yeah. It's definitely been very helpful. We moved a lot of our infrastructure to AWS in the last month or so. And this is the only way how we see we can scale, because we are looking at growing up to like 1,100, 1,200 robots by the end of this year and doubling that fleet in 2027. And the only way we have been able to do in the past is moving stuff to the cloud and AWS, and we are continuing to do it. You can imagine with 750 robots, we collect... The robots see very random things every single day. These are things that you don't see in a lab. You just see it once you deploy it, once you're running every single day, eight hours a day, six days a week. So they don't work five days a week. It's six days a week and it's in every continent. You're up all the time. So every time the robot sees something, we then figure out if this is new, it gets stacked, uploaded to AWS, gets labeled, a new model is trained, and then pushed out to the robot at the edge so the robot continues to improve every single day.

Gemma Allen

>> So let's talk a little bit about this whole space, right? Physical AI, AI on the edge, it's sometimes referred to edge deployment, because I think what a lot of folks really don't fully understand is robots can't always call home, right? They have to actually sometimes record offline, upload. And the technology and the developments in that space, especially by companies like AWS, are quite innovative in a very short space of time. Edge AI has become a huge deal in what seems like a blink. As someone who reports on tech, I listen to it all the time. Talk about the specific areas of the AWS infrastructure stack that we're kind of innovating and developing specifically towards that goal. Let's do the product side of it a little bit, if you can.

Vibhor Sood

>> Yes. So I think for us, we have a hybrid model. We have some edge compute. We do all of the autonomy on the edge to make sure that even if we lose network, we can still function safely. But when we do have connectivity, we use LTE. So every month we push a new update to the robots, and we do it using IoT Greengrass. We know it's pushed fleet wide and it's pushed across, again, 18 countries at the same time. It takes maybe two days to update the entire fleet, thanks to AWS and how distributed the cloud is. Once new data is labeled, we use AWS SageMaker to then train new models, validate them before they get pushed out to the robot, because we have to make sure that they're better than what the last model was. We use ECS or EC... Every time when the robot is turned down, we push everything that the robot sees to the S3 bucket so that at any point we can go back and look where the robots are. If it misbehaved, how did it do it? Why did it happen? So we can fix as quickly as possible.

Gemma Allen

>> And I think Alla, what's fascinating about a company like AWS, it's just like Burro AI is in this recursive learning mode all the time. So are you, guys, right? Because you're also learning across so many customers who are sending data to your cloud at places we can't even possibly imagine. Talk a little bit about what you are seeing and learning in the field broadly right now, especially as we enter this very interesting moment where it feels like there is somewhat of a split between edge AI and the kind of data center model for AI we're also familiar with.

Alla Simoneau

>> This is a transformational point in time. And I kind of alluded, robotics have been around since the 1950s, and we talk about systematic manufacturing automation traditional area and agentic automation in that environment. What we're now doing is we're pushing the boundaries beyond just a controlled, highly connective environment, into real world scenarios. Talk about a tractor in the middle of the field, an autonomous cargo ship in the middle of the ocean, all of these use cases. A vehicle like a Waymos driving in the middle of the road, they're now interacting with a 360 set of an environment with dataset that doesn't exist in a traditional large language model. And the model doesn't need, like, tell me the 17th President of the United States. What it needs to know is like the temperature conditions, the tactile nature, the sensor, the pressure, the conditions of climate, force factors, all of these data points are not memorized in an encyclopedia anywhere that a model can learn from. And so as you're talking about the transition from a highly controlled digital system to a physical AI, autonomous system operating in the real world, it requires vast amounts of data and information. It's like learning or teaching. It's even requiring a different type of model, like a non-large language model, but a vision-language-action model, a world model, because the kind of data that you need to simulate to understand, to train a model is nothing that we currently have anywhere stored. And so that's kind of what we're seeing. And Burro AI is essentially deploying this 750 robotic fleet to be able to perfect the brain of these systems. So while it's operating and interacting at the edge, the processing, real-time operations happens at the edge, call it like a different time horizon that data gets uploaded back into a model into the brain. It gets optimized, fine-tuned, retrained, pre-trained back to this edge device so it can actually operate better. And the more data that you have, the richer data, the more you can account for key anomalies because we're talking about real world scenarios, like a pigeon could sit down in a device. What do you do? Or there's a weed that's blocking the device from moving. All of those different scenarios have to be learned. And that's what's really exciting is all of the data and training and the real type scenarios that have to be essentially absorbed by these companies to really build safe, resilient solutions in the field. That's what's really magnificently apparent today and where we are-

Vibhor Sood

>> Yeah. By the way, the weed stopping the robot used to happen to us a couple of years ago because the data sets that we are creating, no one has it. So the only way for us to learn is to ship the robots, collect as much data as we can, and then improve as quickly as we can.

Gemma Allen

>> In Ireland, I'm Irish, we see cattle roam into the middle of the street and stop farmers and tractors. There's all sorts of problems out there to be fixed, I can tell you.

Vibhor Sood

>> Exactly.

Gemma Allen

>> So, we're here. It's day one, the main day at AWS Summit in New York. I know it's going to be a lot of announcements. Talk to me a little bit about what you guys are excited about, what you're looking forward to, what's ahead for you for the next, I guess, 12 or so hours, maybe longer if it's a big party afterwards, right?

Vibhor Sood

>> Yeah. I think for us, we are launching Grande 44. It's our next-generation robots. We've partnered with the largest sensor providers in the world. We are bringing 9X improvement in compute, power, reliability across the boat, and we are expanding into Europe in 2027 with it. This is physical AI in the world, doing real work, learning from it, and improving and working alongside people. We are creating a new category that hasn't existed.

Gemma Allen

>> Wow. So exciting.

Alla Simoneau

>> I'm a plus 1,000 to that. I know there's a humanoid somewhere here on the floor. I'm excited about the industry education knowledge and acknowledgement of what we're seeing today, which is a transformation from digital to physical. And these conversations happening now, but really across this entire event, are showcasing a reinvention of possibilities and across all industries and how we live, how we work, how we interact with technology, and what this means for society. It's a really monumental, I think, experience that's happening in the background here with different customer conversations like this one. But I'm excited for what the future holds, and I'm really thrilled that AWS and the Gen AI Innovation Center and Burro are here to be able to showcase a little bit of the art of possible in the space of physical AI.

Gemma Allen

>> I love that, the art of possible in the space of physical AI. Well, Alla and Vibhor, thank you so much for joining us on theCUBE.

Vibhor Sood

>> Thank you so much.

Alla Simoneau

>> Thank you so much for having us.

Vibhor Sood

>> Thank you for having us.

Gemma Allen

>> I'm Gemma Allen coming to you from the Java Center. It's the AWS Summit 2026. We're here with theCUBE. We are covering all the action on the ground, talking to some of the key players in the AWS ecosystem. Stay tuned.