Media Week NYC theCUBE + NYSE Wired | Abe Tarapani, Atlas AI

Clips
More from Media Week NYC theCUBE + NYSE Wired

Abe Tarapani

CEO

Atlas AI

play_circle_outline Advancing Sustainability: Atlas AI's Tech for Climate Change and Real-Time Global Trends

play_circle_outline Evolution of AI in geographic information and forecasting

play_circle_outline Impact on physical supply chains and dynamic operations

play_circle_outline Unlocking Opportunities: Digital Twins, Data Leverage, and Potential Hiring in Company's Commercial Progress

play_circle_outline Maximizing Market Potential Through Ecosystem Collaboration, Weather Data Integration, and Hyperscaler Utilization

Info
Transcript

Abe Tarapani, Atlas AI

Abe Tarapani

CEO Atlas AI

Atlas AI, founded by climate scientists at Stanford in 2018, focuses on understanding global change in real-time. Their goal is to create a digital twin of the global economy to help companies adapt to supply and demand patterns. The team, mainly from Stanford, raised $10 million and aims to expand their customer base. They see potential in collaborations with other data companies and increasing compute power. Following Climate Week, their focus is on partnerships with weather modeling and biodiversity monitoring companies to offer holistic solutions. Climate... Read more

explore Keep Exploring

What is the background behind the founding of Atlas AI? add

What type of data is Atlas AI specializing in building large AI models around? add

What challenges are customers facing in their physical supply chain operations and how is Atlas AI helping them address these challenges? add

What are some potential uses for the technology developed by Atlas AI in the future? add

What are some emerging collaborations in the ecosystem that are bringing more holistic solutions to the industry? add

bolt Powered by CUBE AI

Abe Tarapani, Atlas AI

search

>> Hey, welcome back to theCUBE's coverage here in New York City at the NYSE CUBE Studios here in our East Coast location. We're kicking off a lot more content to come. This is Media Week, it's part of Climate Week, the UN's in town. The conversations have been all about governance, compliance and with climate, it's about sustainability. And our next guest, Abe Tarapani CEO of Atlas AI is here to drill down on the tech for sustainability, but also the bigger picture. Aba, thanks for coming up.

Abe Tarapani

>> Great to be with you, John.

>> So it's been a busy week. We're going to have three days of coverage here at our new location, and it's great New York scene. A lot of themes go to New York. It's Climate Week. Next week will be something week. It's always a week here in New York, the Big Apple, a lot of action. But the sustainability conversation, your company is doing a lot in the area. And with the shift that we're seeing, you're seeing a lot more new categorical things emerge. Jensen Wong at NVIDIA says all the time, "Gen AI is a category because it's generating, it's not static and it's evolving." And he speaks to the bigger picture of data centers becoming larger and bigger, which is going to put a lot more supercomputing capabilities in the hands of everybody. So democratization of supercomputing is going to enable a lot more capabilities. And one of the areas that's coming clearly on the table this week is sustainability, climate change, carbon footprint. Also on the backdrop of the CapEx being spent by all the hyperscalers and the data center build outs for edge and supercomputing, you're seeing a massive investment in capabilities and physical infrastructure. There's a sustainable angle there. So this is what you guys do. You're in that world, also partnering with Google. Appreciate you coming in.

Abe Tarapani

>> That's right.

>> Take us through that. What's going on in the sustainability market right now?

Abe Tarapani

>> Well, first, great to be with you. And as a bit of background, Atlas AI was founded in 2018 by a team of climate scientists at Stanford. And they were observing the evolution of the AI space and recognizing that there was a category of AI that was not being well addressed by research, which was better understanding change that was happening around the planet. And so how were communities evolving and affected by shifts in climate and also broader factors, conflict happening in multiple parts of the world, global development and economic trends. And they developed a set of capabilities that ultimately turned into Atlas AI, which leads us to a point of having a capacity to monitor change that's happening in near real-time around the world. But to tie that back to what's happening this week, that is the topic that's on everyone's mind. The future is more uncertain. We just don't have as good a sense of what's going to happen next week, and much less a year or two, three years from now. And you mentioned the topic of supply chain. Most global organizations are used to planning in decades, not years or months or days. And they're now starting to face a reality where they have to be more agile. That's Atlas AI's business.

>> It's interesting, the physical world and digital and AI comes together. It's the first time in my career I've seen such innovation going on at the back end, the big heavy lifting capabilities, whether it's design, compute, GPUs, and then the front end, which is user experience and societal change.

Abe Tarapani

>> That's right.

>> Both are happening at the same time. So you have tech disruption enabling new things and process change happening at the exact same time. As Atlas evolves, how do you guys look at that market? That's difficult. It's a new environment. This is a historic moment.

Abe Tarapani

>> It is, and it's funny in a way that a decade ago we would've been talking about digital transformation. That was the big trend that was happening where companies were just trying to get a hold of all of the data that were disaggregated, often in paper documents across their organization. Many of them have now completed that journey and they're realizing that there's a whole next leg of the relay race, which is what do we do with all of this information? So it's timely in a way that the AI field is maturing to a point where companies can now ask sophisticated questions of that information, but we're really just scratching the surface in terms of how they harness-

>> All right. Let's talk about Atlas AI, the origination. You mentioned Stanford, you mentioned some new problems you guys are solving. Take us through the journey of Atlas, how you were formed, your mission, your north star and what you guys are doing today. And then how is tech helping you guys be successful to solve ... These are hard problems. We now have scale there. Take us through that.

Abe Tarapani

>> Yeah, absolutely. So if we think about the broader sphere of the AI space right now, we're very familiar with some of the new technologies that are being released that are finding complex patterns in text data or imagery or videos. And there's this whole other field of geographic information that has been around forever, but that AI has been late to the game to be applied to it. So think of this as information that's coming from climate sensors around the planet, information that is economic or demographic in nature, like the census that the US government captures, information that's geographic about where infrastructure is located around the planet. And Atlas AI was formed with a specialization around building very large AI models that are primarily trained on geographic data. The reason that that's important is that as we see in so many spheres of life, locations matter relative to each other. The fact that we can think about a supply chain running from the Port of New Orleans up to Chattanooga is a different thing than if it has to run to Ottawa. And so our models understand fundamentally the importance of location and increasingly are starting to forecast forward what the world is likely to look like tomorrow.

>> When you say supply chain, what do you mean specifically about supply chain? Physical goods, digital supply chain, data. Take us through that progression there.

Abe Tarapani

>> So our customers are generally thinking about physical supply chain. So they're operating often very complex, global networks that are transporting many raw materials or disaggregated suppliers ultimately to a customer at the end of the day. The raw materials are not going to be tomorrow where they were yesterday. That whole picture is changing. There's a range of regulations that they now have to address that were not present in the past and their customers are also moving. The patterns of migration that are happening both within the US and globally mean that they have to be more dynamic in how they reach that end customer with their products. And Atlas AI is helping them to better anticipate where that customer is going to be, where their sources of inputs are going to be and build a more dynamic capacity to navigate each step of that process.

>> So what I like about what you're saying is those workflows are physical workflows, but you guys looking at a holistic perspective around the globe, the environment impact. It's a data tsunami. And you have sensors from whether it's a drone in the ocean giving some information or other telemetry coming in, you got a lot of data coming in.

Abe Tarapani

>> That's right.

>> Talk about the technology that you're dealing with now, because again, the old school world was high-performance computing or HPC, which was we all know is like, okay, go get time on the system, spend a bunch of money, wait 10 years later, whatever the lag between old school HPC. Now cloud technology's there, you've got tons of compute coming in. So now you have got the supercomputing enablement dimension. This next wave is democratizing supercomputing. NVIDIA, Broadcom, these companies making semiconductors, the stack's changing. So it's a perfect storm for you guys, pun intended.

Abe Tarapani

>> Well it is, and it's actually interesting in our world, and this actually goes back to your introductory point about the role that Google plays as a key partner for us in that we don't have to reinvent that part of the wheel thankfully. They've solved some of the hard problems around making the compute available and flexible for companies like us that might train a very large model in February, but not have to revisit that for six or eight months. And so we don't have to buy the data center in order to get that done. But we are blessed with a absolute wealth of information to train our models. Satellites are now capturing imagery of the planet every square foot, every single day. We have that imagery, but that imagery is unstructured, it's huge and it requires massive amounts of processing to make sense of, and that's really the role that Atlas AI plays in the equation.

>> And prior to say Google having the capabilities just go back a decade, was it possible?

Abe Tarapani

>> So the answer it is kind of it possible and now it is available at scale and economical and we can move at the pace that the global economy is-

>> Yeah, Dave Vellante and my co-founder and co-CEO and CUBE pod host with me. We were riffing last Friday on our CUBE podcast, we were talking about what's next after cloud. Because cloud technologies created a lot of great value, SaaS applications, the different development process, basically a back-end development and innovation. And now SaaS apps are getting agents, agentic systems will probably upgrade the SaaS category as next-gen applications with gen AI, so it's the application security and all that data engineering goes with it. That's exciting. But a new category is emerging. I call it scalable apps. I don't have a word yet for it, but it's brand new because these are hard problems that engineers and entrepreneurs are solving that weren't solvable before. This is a brand new category. Do you agree?

Abe Tarapani

>> No, completely. And actually we look at that in two different ways. One is at the infrastructure level, the industry is moving to what we would call a serverless architecture, meaning even with the benefit of Google Cloud years ago, you would have to reserve capacity or make long-term commits to ensure that you had the capacity reserved to run your models. Now as demand increases, the architecture allows you to scale accordingly. On the user-facing side, one of the things that we're seeing is that generative AI is allowing you to push more of the end capability to the customer themselves. So no longer do we have to fine-tune the model and deploy that to our customer. They can take the tools to integrate their internal first-party data, refine their model, have it be deployed in a unique way for their environment. And it allows us to be really part of the tooling ecosystem as opposed to how we might traditionally think of SaaS as giving them everything nicely wrapped up and ready for the end user.

>> It's interesting. You got ... I won't say throwback because engineering never went out of style, but classic engineering is coming into systems thinking.

Abe Tarapani

>> It is.

>> And your company has to look at the supply chain as a system using data that you're now collecting with big data techniques as an old term. But I'll modernize it with the cloud and say, "Hey, I got compute." Throw it at compute, fast turnaround time, so you get feedback.

Abe Tarapani

>> That's right.

>> And then you develop on it. You got to do the engineering work.

Abe Tarapani

>> Well, and I think what I love about our company is that we're really bringing two disciplines together. One is that scalable engineering around geospatial. The other is the economic subject matter expertise that really is at the heart of how we think about these global supply chains. They have to really respond to the way that the economy is shifting. And that's not just a matter of finding patterns in large data. So think of it almost as building a large climate model. You can't just throw the climate data in and expect to get a good output out. You have to bring weather expertise in as well.

>> It's interesting, we're having a hurricane come up the East Coast. It'd be interesting to know that information. There's disruption opportunities that could take place. You actually maybe adjust the supply chain obviously around us. An obvious example. What are some other examples that you see that are enabled by this disruption? Because what I love about this market is it's what I call a disruptive enablement market. It's disrupting, but it's enabling. It's not disrupting for bad, it's disrupting for opportunity. What are some use cases that you see are new opportunities?

Abe Tarapani

>> Sure. Let me give you two on very different sides of our business. So one part of our business is helping public sector agencies, governments and global development organizations serve vulnerable populations. So Nutrition International is a partner who is working with local government partners to distribute vaccines and vitamins and other aid to malnourished communities in lower and middle income countries. They need to be able to adapt to the operating conditions on the ground, which are very volatile and prioritize limited resources to reach the communities that will benefit the most from their support. On the other end, we recently announced a partnership with the commercial aviation group of Airbus. Can you imagine a more complex supply chain that needs to get all of these pieces from every part of the world into make a finished aircraft? And they're thinking both about how do they better anticipate the demand for travel that is going to be non-linear and very fragmented, especially in markets like Asia, but also how do they better anticipate, as you mentioned, weather and other patterns that restrict that just-in-time element of making sure that the components are in the right place at the right moment so that they have an efficient supply chain?

>> So efficiency is critical, better product development. What are some of the model thinkings you guys have? I like what you said earlier. You do the training and then you go about your business. It's funny, I was on a CUBE interview and I had a great metaphor from a guest, I forget who said it. He said, "It's like going to school. You get trained and then you infer and reason from it and then you might go back to school for part-time, reinforce that learning." So again, I'll be seeing the progression here. It's what we're talking about. You train the models, you infer from them, you reason with them, and then you reinforce learning and then build apps.

Abe Tarapani

>> That's exactly right.

>> This is the model. What's your view of that? Because this seems to be a lot of emphasis on training right now, but once you're done training, I don't want to go back to fourth grade. I don't need to. I might want to get my skills going, but ...

Abe Tarapani

>> That's exactly it. And it is actually interesting in that you're right, the training is a huge component of this because part of what we're trying to do is create knowledge about the world that doesn't otherwise exist. And that's a key part of why companies engage with Atlas AI. But the problem that they're actually facing is that they're sitting on a mountain of data themselves that they're not making use of in a way that helps provide specific feedback loops into the business. And so to your exact point, the training gets you to a set of patterns, but it's really that refinement and the inference that can occur once a company integrates its own internal systems. That's where the magic happens because now they can learn based on what was successful or not yesterday, how do they adapt tomorrow to better adjust their organizational and operational plan?

>> Well Abe, I really love your mission and I think it's awesome, because I think it's a unique time in history. And again, this is a first generation attempt at solving a lot of hard things at the same time. So congratulations on that.

Abe Tarapani

>> Thank you.

>> Not putting more pressure on you. But we're at Climate Week, so imagine go forward in your mind at steady state. Atlas AI is growing, you're successful. In that steady state of your business, what does that look like for you? Is it agents, agentic systems working? What's happening in your mind's eye? So it's all working.

Abe Tarapani

>> That's right.

>> Assume it's working.

Abe Tarapani

>> We've solved everything.

>> The string gets pulled, everything kind of happens. It's all good.

Abe Tarapani

>> So what really motivates our team is building models that can identify increasingly complex patterns about the global economy. We might call that a digital twin of the global economy. So every company that has a physical presence in the world needs to be more responsive to increasingly volatile supply and demand patterns. Our goal is that there are likely to be others, including the Googles of the world, who build that agentic capacity to tap into data and to work with it in a more accessible way. But that system of intelligence underlying those systems, we want to come from Atlas AI.

>> Yeah, I've always loved Google Earth. And so to me, I think of it like you're instrumenting the earth with data. And so more data the better. So you need more data, or do you need more data or is there plenty of data? There's incumbent data out there, there's pre-existing data, but we're going to get more telemetry. Take me through that vision because I think as you start seeing the Google Earth, "I can see the earth." Imagine I got a digital twin of the earth. What am I doing with it? Am I planning my employees to work schedules based upon weather disruptions, giving hyper personalization, maybe helping on where I want to have my office space, build a data center? What are some of the things?

Abe Tarapani

>> I think that the reality is it could be all of the above, and that's why it's so important to create the capacity to fine tune applications on top of our platform. So you can easily imagine that part of the housing expansion plan in the US relies on a better understanding of the socioeconomic patterns that are occurring across the US in very granular ways. Where can new housing have the greatest benefit on equalizing the playing field at the same time as you might have very high frequency operations that are dependent on weather and other patterns that need to better anticipate what preparation is needed in order to operate. So Airbus again is a very good example of that.

>> You got real customers. Again, great mission. Talk about the company. Give a commercial for the company, put a plug in. Where you guys at in the journey? What are you looking to do? You looking to hire? Take it away.

Abe Tarapani

>> Yeah, absolutely. So for those familiar with the startup life cycle, we're still relatively early stage. We've raised about $10 million today. We have a team of about 20 people, and actually almost all scientists and engineers still within the company-

>> From mostly Stanford?

Abe Tarapani

>> Yes, yep. And the founding team are professors at the Doerr School of Sustainability and in the Department of Computer Science at Stanford. And so the next leg of our journey is, as you mentioned, we have a set of customers that we believe represent a problem that's quite large. We're now building the infrastructure to deliver that to more customers out there.

>> If you're watching, it's a great mission. Again, it's technical, a lot of hard problems to solve. So it's the classic recruitment dream scenario. A lot of people want to work on hard problems and now's a great time to do it.

Abe Tarapani

>> Please reach out to us.

>> Final question, are you pleased with Climate Week this week? Did it meet your expectations? Is it progressing along in the right way? Is sustainability a bolt-on, or it moving to be designed in from day one?

Abe Tarapani

>> I can only be in awe of the scale of this week. Compared to even last year, it seems like we have 10 times the amount of people and companies that are taking part of the dialogue. And it is an opportunity for people from all over the world to come together in a common place. And so I'd say that a lot of the value has been in the side meetings that have emerged from some of the big events. But companies are now starting to reframe climate as resilience. They realize that in a changing climate, while there are scary scenarios, 10, 20 years out, their operations are affected today. This is a very real business challenge that they're facing. And so they are here to identify solutions because this is a financial reality that they need to deal with as much as a social responsibility aspect of what they hope to-

>> I love how the side meetings drive a lot of engagement. What did you learn from some of those side meetings and what are you going to take back to the ranch, so to speak, at Stanford, or the farm I should say? What are you are going to take back to the farm at Stanford with the team? What's your walk-away mandate? What are you going to do?

Abe Tarapani

>> So the first is that there are some collaborations that are emerging from historically very siloed parts of the ecosystem. So for example, we don't internally do a lot with climate and weather modeling, but a lot of those companies that are building very advanced weather models are here. And there's a whole other ecosystem around monitoring biodiversity and forestry loss and other hit impacts on natural resources, the reality is that most solutions rely on the intersection of all of those. And so there's a lot of interesting collaborations among partners that are starting to form to bring more holistic solutions to the industry. And I'll say that the presence of the large hyperscaler companies here and their attempt to bring their customers to the table who are struggling with these questions in real-time, offers us an opportunity to market our solution and to raise awareness.

>> Not only do you have the brains from Stanford and the team, but you guys are a data company. You guys are building massive datasets and maybe merging or integrating with others.

Abe Tarapani

>> Right.

>> That will be a future state for you guys, integrating and talking to other data in real-time.

Abe Tarapani

>> No, that's exactly it. And the more of those connections we can make, and as we see the acceleration of the compute power continuing to reach new heights, it is really within reach to think about the type of patterns we're able to-

>> It's the perfect innovation storm.

Abe Tarapani

>> Well said..

>> Hey, thank you folks so much for coming on. Really appreciate you. Great insight.

Abe Tarapani

>> Yeah, John, thank you.

>> Again, high frequency insights coming here from the New York Stock Exchange. I'm John Furrier, host of theCUBE. A lot of action. Day two of our media week, we got climate, we've got the UN. Of course, generative AI's in every conversation, data and governance. Making the world a better place, and ultimately making sure productivity continues to be a great feature of our tech business. Thanks for watching.