theCUBE + NYSE Wired: Physical AI & Robotics Leaders | Sam Khosroshahi, Lambda

Clips
More from theCUBE + NYSE Wired: Physical AI & Robotics Leaders

Sam Khosroshahi

VP Strategy

Lambda

play_circle_outline Rising Demand for AI Solutions: Lambda's Transition from Startup to Scale-Up in a Growing Market

play_circle_outline Importance of infrastructure innovations driving business transformation in AI applications.

Info
Transcript

Sam Khosroshahi, Lambda

Sam Khosroshahi

VP Strategy Lambda

Sam Khosroshahi, vice president of strategy at Lambda Inc., joins theCUBE’s John Furrier during theCUBE + NYSE Wired: Robotics & AI Infrastructure Leaders 2025 event to explore how high-performance infrastructure is powering the future of AI and robotics. The conversation centers on Lambda’s strategy for enabling scalable compute, deep learning and next-gen AI deployment.

Khosroshahi breaks down Lambda’s approach to supporting enterprise growth through adaptable infrastructure and partner-driven innovation. From distributed training to robotics applica... Read more

explore Keep Exploring

What is Lambda's journey in the machine learning and AI industry, and how does it perceive current trends within this space? add

What are the key components and goals that organizations are seeking when developing AI infrastructure and applications? add

bolt Powered by CUBE AI

Sam Khosroshahi, Lambda

search

John Furrier

>> Welcome back everyone to theCUBE here in our Palo Alto Studios. I'm John Furrier, host of theCUBE. We have Dave Vellante flying in. We've got three days of wall-to-wall cover robotics and AI leaders here in Palo Alto. We've got a big event happening tonight at the Rosewood with the NYSE Wired community, and we're covering all tech, all the infrastructure innovations that are enabling this next business transformation, this next wave. As agents come in, we've got Sam here as vice president of business development, strategic pursuits for Lambda, fast-growing companies serving a lot of the high demand for the scale that's required to run AI. Sam, thanks for coming in. Appreciate you taking the time.

>> Thank you very much. Appreciate it.

John Furrier

>> First of all, Lambda is doing great. You guys are doing good. Give us a quick update on the business. What are you guys working on right now? Give us the scope of the magnitude of the scale we're talking about because we're hearing in the news more and more, 20 billion here, 10 billion there. The CapEx explosion, certainly in the hyperscale is great and demand for large scale infrastructure, but the enterprises are starting to get involved and they want to have different approaches. Talk about what you guys have for momentum.

>> Yeah. First of all, thank you so much for having me. It's great to be in Palo Alto. Last time I had a chance to sit down with Dave was in New York, and certainly the trend continues. We have been seeing early investments in AI and the research that we've been doing. Certainly Lambda has been around since 2012. People are catching up and people are starting to see that there is something to this AI after all. Deep learning distributed training, that is the core essence of where Lambda started built by machine learning engineers right here in California, San Francisco, our headquarters is now in San Jose, Zancar Road not too far from here. So the question is how is business and how are things trending? And to your point, there is a almost insatiable demand for these types of solutions, infrastructure, all the way up to models and how they're being applied across all sorts of use cases and applications. And so yeah, the trend is continuing and we're seeing a lot of demand.

John Furrier

>> The interesting factor in the demand side is the price per tokens are dropping significantly. The models are getting bigger, smarter, faster. Small language models are being distilled off. The big ones, people have their own data and with the token demand going up because you now have reasoning and inference driving, more token demand there, which requires more large scale. So that's forcing the enterprises in any business that wants to use AI to kind of move out of the discovery phase of, hey, we've got this great retrieval augmentation generation thing to let's actually re-engineer our business.

>> Yes.

John Furrier

>> So there are tons of reports out there. I just read a Wall Street Journal article talking about how people are rethinking the labor in their companies. So this is a business model transformation now driven by the demand on the tech side. You guys are the center of it. What is the big factor in this? Because if you look at all the success, the research side of it is instrumental in feeding input into formulation. This isn't the classic call the management consultants in. And give me your watch. I'll tell you what time it is. It's really engineering the business.

>> Yes. I think it's such an exciting time to be in the space. And like I mentioned earlier, we really started out as an application company. Lambda, at its core, machine learning engineers, AI developers, and we were doing our own research. To this day, we still have this research as a core part of our DNA. So when we think about all of these trends, whether it's in robotics, whether it's in reasoning models, whether it's in like agentic workloads that are augmenting the way that we think about work, it's all very, very exciting and it's driving up this demand. And what we're finding is that Lambda has grown from a startup. We are now a scale up, I think as some folks would term it, and that investment into that infrastructure is a key part of it. But what I think is even more interesting for enterprises is that they can partner with us in order to develop those applications that they can adopt. And that's where the rubber meets the road, and to your point is it's starting to reduce their cost and it's starting to potentially increase their top line in certain ways. Nick, one of the things that we read out of Stanford recently was this study around 45% of individuals that were, I'm sure you've read it as well, where we're asked, hey, what are your thoughts on AI as a replacement for workforce or as an enablement, an augmentation for workforce. And it was 45%, but we definitely want to make sure that we were working collaboratively with AI to improve the efficiency and the effectiveness of work. So the nature of that is quickly changing. And also with regards to inference, I think that is starting to be, again, the other point of where the rubber meets the road is all of this research has been happening for a long periods of time, and we're now starting to see it disperse across different facets of-

John Furrier

>> You mentioned the Stanford report. I also saw a video online. Someone was giving a talk at Stanford, forget who it was, it was a leader. And they asked the class, who in here is using ChatGPT to augment their studies? Every hand was in the room was up, and you're starting to see things like perplexity labs, open AI. You can hire a researcher for 30,000 and just get a PhD resource. So the agent assist is definitely here, no doubt about it. So I think that's clear. I'm also on the side of more jobs will be created.

>> Absolutely.

John Furrier

>> Than reduced. And certainly the mundane tasks are going to be gone. That's going to go away. We don't yet know what's coming. That's to my view. So with that in mind, how are your customers looking at working with you? Because the application layer is where the action is, the business logic, the data. It's not an IT purchase, although it's kind of an IT purchase. You're buying essentially resource, but the resource also is going into a different valuation, why they're doing it. What are some of the use cases? What are some of the enterprises using you guys for? What are some of the demands? Can you share any anecdotal data? Could be specific customer name. What are some of the low-hanging fruit and where's the roadmap on this?

>> So it's all super exciting and it's happening in real time, and we're very fortunate to be a part of that journey. What we're finding is the applications are wide-ranging. And for Lambda specifically, this is our opportunity to continue to scale not only the demand, I believe you mentioned the infrastructure, but how we engage with customers and help them develop those applications and actually scale them out for themselves. So what we're finding is things around robotics, right?

John Furrier

>> Yeah.

>> And robotics, the humanoids is one thing that folks have looked at. Warehouse automation is another area. There's other areas where you think about generative AI. So the videos and images that are being generated, and certainly the agentic workloads as well that we've talked about. So when we work with organizations across all of these areas, they're really looking for best-in-class infrastructure. Price per token is another area where you mentioned, and that these chatbots is probably not the right term for it, but we're starting to see-

John Furrier

>> Assistance.

>> Assistance, exactly. And that's really where we want to continue to evolve is being a place where we can collaborate with a wide range of use cases and organizations and help them leverage the best performance on these, which is why Lambda is a pure play AI, because we are not just an infrastructure provider, although we have massive amounts of megawatts, we've actually grown our megawatts by four to five x, and we'll continue to do that over the next few years.

John Furrier

>> You have to because that's where the power is. People want to stand in line for that compute. I mean, you got to get the data to the compute. That's minimum table stakes. I'm interested by the application formulation because a lot of startups and a lot of companies are coming to the enterprise and we're seeing a lot of POCs for all, and the enterprises don't have the mechanisms to kind of work through the POCs. So a lot of these startups have ARR. Some of them are $10 million AR, but there's no R in there. There's no recurring, they don't get into production. There's no recurring. So how do you see that thread? Because this is where I think the enablement will unleash more production workloads. Is it the enterprise have to kind of get their act together? Is it more of a systems play? Is it more of a business logic? What's the bottleneck you see in that? I mean, I'm sure you'd agree that there's a lot of POCs, but you're starting to see a lot more focus on in production.

>> Yes.

John Furrier

>> What's the gate? It's like the airport. TSA pre-clear now exists now with you guys.

>> Yeah. Look, we give customers choice and flexibility. I think data's at the cornerstone of a lot of this, wanting to retain your own IP, wanting to leverage it in your own way. That will continue to evolve and be closer to what your requirements are with your own data. That's one of the ways that we're doing it. We are also seeing multi-modal sort of large-scale models that are being used for generic type of applications. And that's certainly another aspect, but I think this combination, that you said the gate is the strategy level of the desire to do it. These POCs that we've been hearing about and seeing over the last few years, and then now you're starting to see production as consumers that are also in the workforce have adopted AI in their own everyday lives, whether it's shopping or...

John Furrier

>> Sam, I'm curious about your role at Lambda because business development, we all kind of know what that position is. You'll develop the business. Strategic pursuits. What does that mean? Because the strategic endeavors, you're always doing this dev. Like a corp dev function, are you pursuing new verticals? I mean, because we're now in an era where there's so much opportunity, life sciences, healthcare, traditionally slow eye on the IT front, booming with large scale, what are some of these pursuits you're doing? Can you share a little about what you do on a daily basis? Is it a new kind of role?

>> Yeah, I think Dave gave me a little bit of a hard time about that last time we had a chat. But certainly it's a very exciting time for us in this particular role. Lambda's moving extremely, extremely quickly to keep up and be ahead of where the market is. So we do have a corporate development function that sits close auxiliary team to this organization. And we work really closely with our revenue leaders. What we do is we work across the bookends of the market, which is the early stage foundational research type organizations that are pioneering at the frontiers of AI as well as the trillion dollar market cap companies. So our team is really focused on finding new ways to create mutual value, to engage with one another and essentially partner. Of course, Nvidia is one of our largest partners. We're very grateful for the collaboration that's there. And hopefully that gives you a little bit more of an insight. So it's-

John Furrier

>> Essentially it's business development in the modern era because you are knocking down new territory. That might've been a far reach in the classic sense, like robotics, take AI factories, for instance, huge growth area. Physical AI is one of the hottest trends. We're hitting that up here in our program. That's a whole nother thing. You're talking about digital twins, you're talking about use of synthetic data. How much real time data do I get? I mean, these are new problems.

>> So I think our team is constantly working across all sorts of industries and all sizes of organizations. So this traditional way to segment them out is not how we think about it. We are working very closely with organizations, like I said, at the frontier of whether it's robotics, healthcare, and life sciences, drug discovery, things of that sort to understand their use cases. One of things that we've done, for example, is we have an applied AI team that's working with customers to optimize their models sort of beyond this infrastructure, which is the differentiator. When I say Lambda is an AI pure play company, I mean that we go from the infrastructure layer all the way up to model optimization, kernel level optimization with our partners.

John Furrier

>> Yeah, it's a full AI stack. I love that approach. I always love this market because you've got horizontal scale and vertical specialization happening at the same time.

>> Exactly.

John Furrier

>> And that's a great endeavor to be on. Great to have you on, Sam. Congratulations on the success of Lambda. Put a plug in for the company. Where are you guys at? What are you looking to do? You guys looking hire, what kind of customers are you looking for? Give a quick plug.

>> Yeah, thank you so much. We are growing at an exceptional scale. We're very excited by how the market is responding to other organizations that have gone public. I think one of the things that we are looking to do is continue to hire great talent, continue to work with great organizations like we have been, and really, again, engage with delivering outcomes, real world outcomes through the infrastructure, the AI themselves, and of course the collaboration with our customers and partners.

John Furrier

>> All right, well, great to have you on. Thanks for coming in. Lambda, doing extremely well, obviously serving the needs at the infrastructure level. Also shaping the applications and the business value horizontal scale with vertical specialization, whether it's robotics, AI factories, digital twins, or just building that next generation gen AI app that's all happening in the enterprise. I'm John Furrier with theCUBE. Thanks for watching.