RAISE Summit 2025 | Paul Bloch, DDN

Clips
More from RAISE Summit 2025

Paul Bloch

President & Co-Founder

DDN

play_circle_outline DDN: Pioneering Self-Funded Storage Solutions and Accelerating AI Innovation Through Strategic Partnership with NVIDIA

play_circle_outline Transition from traditional storage to data intelligence as a mission-critical aspect of AI.

play_circle_outline Regulatory aspects like GDPR still relevant in the context of digital sovereignty.

play_circle_outline Importance of competitive advantage through access to power, infrastructure, and skilled personnel.

play_circle_outline Recent partnership with Blackstone to enhance stakeholder base and leverage connections.

Info
Transcript

Paul Bloch, DDN

Paul Bloch

President & Co-Founder DDN

In this RAISE Summit interview, Paul Bloch, co-founder and president at DDN, sits down with theCUBE host, John Furrier, to unpack how data intelligence platforms are transforming AI at every scale. Bloch explains DDN’s 25-year evolution from high-performance computing leader to billion-dollar AI startup and details how early partnership with NVIDIA positioned Exoscaler as the backbone for training foundation models and high-throughput inference clusters.

The discussion covers multi-cloud deployment, sovereign cloud demands and the shift from data stora... Read more

explore Keep Exploring

What was the history and evolution of DDN as a company, and its relationship with NVIDIA in the context of high-performance computing and AI clusters? add

What advancements have been made in data management and intelligence, particularly in the context of new technologies and enterprise solutions? add

What are the key components and considerations involved in implementing country-wide sovereign AI systems? add

What is the current status of large-scale energy projects and the support from governments in terms of permits and infrastructure development? add

What was the purpose of the $300 million investment from Blackstone in DDN? add

bolt Powered by CUBE AI

Paul Bloch, DDN

search

>> Welcome back, everyone. I'm John Furrier with theCUBE. We're live in Paris, France, theCUBE coverage, SiliconANGLE and theCUBE. And, of course, the NYSE Wired, part of the CUBE effort and collaboration, of course, getting all the action. And we're here on the ground as we go international and bring the AI conversation at scale globally where the AI infrastructure is going under significant acceleration of change, creating massive amounts of value. And all the startups, all the companies are here. Paul Bloch is the DDN co-founder and president, a storied company. Paul, welcome to theCUBE. Thank you for coming on.

Paul Bloch

>> Thank you, John. Appreciate it. Thanks for having me again.

>> Bonjour. - Bonjour.

>> . - .

>> Very good. I'll have one of those. I love DDN because as a company we've been covering it. It's been one of the companies, one, it's self-funded, well- documented, but it's been in the storage business for a long, long time. A series of waves of innovation. We've seen the movie, but right now, as storage becomes a data platform, you have now large- scale storage fabrics powering these supercomputers. It's really a combination of infrastructure innovation, which physically is storage and drives and now software, and then you have the enablement of the data layers on top of it. This is going to open up the aperture of value of how data is stored, how data is processed. The game is still the same, but it's changed.

Paul Bloch

>> Sure. - So what is the perspective as you navigate through the company and look forward as the key drivers for DDN today?

Paul Bloch

>> Sure. So I mean, historically, if you look at it, DDN is a 25-year-old billion dollar start-up. That's the way we look at it. So the first 15 years of the company were really focusing and delivering and becoming the leader in high-performance computing. Meaning what? For 15 years, we did sovereign HPC, we did sovereign research, we did sovereign defense, we did sovereign and so forth. Literally it was an easy transition for us. 10 years ago, we decided to partner with NVIDIA. We had a meeting with Jensen. They were looking for the right data storage and data intelligence partner. They looked at the universe and they zeroed in on DDN. So we're very proud of that. And over the past 10 years, we really leveraged on this relationship, opened new ones as well, but really learned and fine-tuned our technology, our roadmap, our delivery, our customer situation to deliver two type of solutions. First, were massive training, because this is really still the behemoth of the AI clusters being deployed. So training of foundation models.

>> And spend too? - And spend, absolutely. We have to survive. We have to make a buck at the end of the day to be able invest in R&D and delight our customers as well. So training, foundation models, training for startup. Fine-tuning of models has been pretty much the workhorse. And our technology, Exoscaler is really unbeatable there, proven at full scale. And even when customers pick some of our competition, they come right back to us as soon or come to us once they need to scale because they understand the deficiencies of competition. And then now you're basically getting into inferencing on- prem in the cloud, multi-cloud. Where are you going to run? Data intelligence is becoming everything. It's no longer data storage anymore. It's data intelligence. Every AI breakthrough is driven by data intelligence and it's becoming mission-critical. So we have a new technology. Actually, interesting because NVIDIA, we launched it earlier this year with you guys as well. And it's extremely fast , call it a data ocean, but very low latency, multi-tenant ready. So really ready for the next generation of inference. And you can basically instantiate or manage your data, leverage your data, make money based on your data on-prem in the cloud. And it's a software instantiation.

>> I mean, you guys get serious customers. You have a great enterprise business, but when it comes down to brass tacks, the big payloads. You're talking about the hyperscalers, you're talking about NVIDIA. Early on, that's a good bet.

Paul Bloch

>> Yep. - You were there early, present at creation, really, in this wave. And I think one highlight of that is that, in the past year, Intel has went from 73% market share to 13 in one year.

Paul Bloch

>> That's crazy. - Which when you think about it, that's mind-blowing. That means NVIDIA won.

Paul Bloch

>> Absolutely. - I'm oversimplifying it, but what it means is that tide has shifted.

Paul Bloch

>> Sure. - High scale, large scale, high volume, low latency demands on the data is absolutely the demand.

Paul Bloch

>> You need performance and accuracy. And so really, if you look at it historically, we have the large customers. NVIDIA was our largest customer in '21, '22, '23. In '24, NVIDIA became more second-largest customer. XAI was our top customer. But it's really not at that scale. We deliver at that scale, which is super important. But we also deliver at very small scale. We deliver to enterprise AI. We deliver to manufacturing companies. We deliver to banks. We deliver at any scale. We actually start with a very single little unit of 2U. But the notion is that we give them the security and the safety and the knowledge that they can scale seamlessly as they ramp from a POC to production. But yeah, clearly, we've demonstrated that the largest cluster we support is 200,000 GPU today, and it'll probably be about a million GPU at the end of the year.

>> Well, I want to get into some of the words you said earlier, sovereign this, sovereign that. You got sovereign everything. Sovereign cloud is the hottest trend that nobody's talking about.

Paul Bloch

>> Yep. - It will come on probably second half of the year, we predict. We think it'll be the top conversation in 2026. So when you talk about sovereign cloud, that brings back the roots of DDN, which is, as we know following you guys, there's engineering involved. You mentioned R&D. You guys do a lot of R&D investments.

Paul Bloch

>> We do. - So that work that's got to get done to do sovereignty, whether it's for a military application or some sort of use case that needs some bounds around it but yet stay open and frictionless traversal, this is to be an even larger scale sovereignty problem. Can you share your thoughts on what that means for Sovereign cloud, that experience, how it translates the NVIDIA piece?

Paul Bloch

>> Sure. So in essence, like I mentioned before, John, sovereign HPC, we actually did it. We've been doing it for 20 years. So the largest government labs, defense, intelligence agencies, smart cities, and so forth, we deployed system. Now what you need to add is the AI portion of it, the more social media, all the rich content. You basically are just expanding this. And becoming also, what you're doing now, is you're creating more intelligence at the level of the company of the country, actually. So you're talking about enterprise, you're talking about company businesses. So the difference is that it used to be only the labs and the energy and a few per se organization customers. Right now, really, the sovereign AI is super important because you want to basically make your intelligence in country, keep it in country, and basically leverage and make dollars out of this. And this is what you're talking about, AI factories. So, for example, Europe has announced a program for 13 AI factories and four giga factories. And if you look at the budget, they're significant, but again, they need to step it up to meet the U. S. or even China.

>> I'm doing a whole feature on sovereignty because I think this has got multiple threads. I'll just get into a few of them. One is the technical challenges.

Paul Bloch

>> Sure. - You laid out a few of them. And we talk about on-prem, on-premises. That's in the data center for an enterprise. We know what that means. But hybrid distributed computing, there's a notion I'm calling on-country, which is like a prem, it's a premise, but it's the country. So that's technical. Then you got the geopolitical. I think sovereignty around the assets of a nation is data as well. So it's not just an IT-like problem. This is a business country sovereign problem. Share your thoughts on this. This is a super important point.

Paul Bloch

>> You just mentioned that historically we're used to having IT making decisions. You're used to IT running your business. This is completely different at this point. That's really the transformation and the paradigm shift. AI is no longer driven by it. AI is driven by the business, by the data scientists, by the chief data officer, and so forth to actually make money to have turned to advanced technology and so forth. But at the end of the day, it becomes, there's an ROI and a TCO to it, right?

>> Exactly. - And if you don't do it, someone else will be doing it. And so, again, I think it's an opportunity for a lot of the countries that have engineers or that have actually knowledge to really create that intelligence, maintain it, leverage, and grow out of it.

>> Paul, I was talking to a friend and we were talking about metaphors and all kinds of ways. It's like, yeah, we're sailing to that island and they'll say, "Wait, that's a better one over there. " Sovereignty was GDPR. That's old school compliance stuff, kind of antiquated, outdated, but still relevant. You still have to have a compliance checklist.

Paul Bloch

>> You have to. - And then marry the technology shift with that. Otherwise, it doesn't work. At the beginning, what's your thoughts on that?

Paul Bloch

>> Again, when you're talking about country, it's also ease of data, ease of dealing with the government. Formless. You don't have to go to DMV anymore. That's antiquated. Why would you go to DMV? I don't need to see the people in the DMV. Do you?

>> They don't want to see you either. it's a win-win.

Paul Bloch

>> So this is all technology that basically add intelligence and data at checking on data. And this is really what we're deploying now with our customers. We're deploying country-wide sovereign AI systems that are powered, for example, by NVIDIA with a network, DDN for the data intelligence platform at full scale, and then you basically have the suite of software and application that is required to connect all the various government agencies as well as the enterprise. So you're talking about telco. You're talking about passport. You're talking about ID. You're talking about everything integrated in a very simple way for the common citizen to basically interface with the government in a simple way.

>> And they got to unify that. - And then smart city. You're talking about smart city.

>> Computer vision, it's the biggest data ingest of all time. All right, so are you having conversations, Paul, around governments around competitive advantage? Because there's two trends we're watching. Energy, everything's bounded by energy. We believe, and this is our thesis on theCUBE, theCUBE research, that alliances may change between countries based upon, one, their sovereignty capabilities now with AI power.

Paul Bloch

>> Sure. - Hey, if I'm a country that doesn't have a lot of power, hey, nuclear. France has got a lot of nuclear power. Maybe I'll align with France.

Paul Bloch

>> Absolutely. - I'm just making this up, but that's the kind of thing I'm imagining. What are some of the leaders, are they there yet? Are we having those conversations?

Paul Bloch

>> Clearly, if you look at it, there's about probably 50 sites with a gigawatt or more being set up over the world over the next three years. And this started. Basically, it's already in motion. So really what's cool is that the governments are actually accelerating the permits, the requirements, the land. All this is being accelerated, which is actually appreciated. And then you mentioned basically France and nuclear. Absolutely, this is massive. EDF is actually one of our large customers in France, and so we do all the nuclear actually research and data on DDN, which is kind of cool, for the past 15 years. And so clearly we are access to power, access to land, access to infrastructure, but also the knowledge and the people that basically make this run, all is going to impact.

>> Well, I'm super happy for DDN. You guys have one founder-led company.

Paul Bloch

>> My partner Alex and I, we met about 40 plus years ago at Caltech.

>> You guys did pretty good for yourself.

Paul Bloch

>> We tried. - Made a lot of money, but more importantly, you guys stayed the course. Now we're at another inflection point. As you look forward now, you're founder-led, now founder-inspired. What's the view as you scale the company because now you have massive demand? This isn't your grandfather's storage market anymore.

Paul Bloch

>> No. - This is a whole other level of activity and people are trying to squint through them. People are making generational decisions right now on the systems that they're building.

Paul Bloch

>> Absolutely. We did a couple of things. One, obviously, technology, actually, we're ahead of the game. So it really was more a notion of beefing up the management team at DDN, expanding the marketing, expanding the visibility of the company so customers know that we're there. And then once they know, they see the differences with other technologies. So that's one part whereby over the past year we've expanded our team. We've hired 300, 400 people over the past 12 months, which is significant for us. The other portion of it is we partnered with Blackstone six months ago. They invested $300 million in DDN. First round ever, right?

>> Yeah. A round, seed round.

Paul Bloch

>> Out of a small valuation of 5 billion at the time.

>> Not a bad seed round. - No, but the idea was not really for the money because the company operates profitably. We actually generate cash on an ongoing basis, even with the growth. But the idea was really validation of the company, that we are someone that is becoming the data intelligence play for Blackstone, tier one investor. And the other thing, obviously, having access to some of their connections and being able to create a financial package altogether.

>> You guys didn't need the money, just for the record. You guys were doing great. And that was well-known in the industry. This is more about expanding the stakeholder base, right?

Paul Bloch

>> It's coming out and saying, "DDN is no longer the old DDN. It's no longer your old Oldsmobile. " We've never been an Oldsmobile, but, really, DDN is basically, yes, we're proud of our past, but we are in the future and we're morphing actively into the future .

>> Let's get into some of the cool things around some of the deals you've done. I know a few of the deals. I can't talk about them.

Paul Bloch

>> Or I'm going to kill you, right?

>> They're confidential. Well, I could report them, but that's not what we do. But there are big deals in there from customers that have a high bar technically and a high-scale bar. And as you look at DDN going forward, you got to have that tech advantage, but also have the speed. So as you scale out the offering, what are some of the use cases you're going after? Can you share a little bit more about that next 20- mile stair, the journey? And then, two, how do your customers squint through the hype? Because there are other people out there grandstanding saying they're number one, "We have the best technology."

Paul Bloch

>> Absolutely. So you asked me a few different questions, right?

>> Yeah. One is what's the future and then a hype squinting.

Paul Bloch

>> So if you look at it, historically, we've already spoken about foundation model art training. So who is basically nowadays doing on- prem on the cloud at scale? High frequency trading, financial banks, fraud detection, algorithmic trading, Monte Carlo analysis, massive influx of investment. And we are very present. To give you an idea, Jump Trading has close to an exabyte of our data today. And if you go automotive, Tesla, same story. Automotive autonomous applications, we're seeing it across the board. We're installed at Volvo. We're installed at Tesla. We're installed at Ford. We're installed at GM. This is all growing. Medical, Mayo Clinic, St. Jude. A lot of the research we're very proud of actually, of delivering value and technology, but also saving lives at the end of the day. And so medical genomics, very large image simulation. We're seeing a lot of adoption there. Manufacturing, this is coming, but we're seeing default detection, KLA, Bosch, people like that that are basically doing high-scale default detection. Factory automation as well. And then I'm missing obviously sovereign AI, which is just booming as well. So there's a lot of verticals that are now approaching. And then, two, competition. When customers look at, what I would encourage you is dig, ask for referencing. Ask for customers at scale. What have you really done? Where is it running for a long time? What are really the results of it? Because if you look at it and you look at a PowerPoint, everybody does everything great. And it's very difficult.

>> Looks good on paper. - And I will tell you to the detriment of our competition, in some cases, they don't understand the scale. They haven't done the scale. They come from enterprise. And so they imagine that you can do scale, and you can't. So they don't completely understand that.

>> They wish for scale. - They wish for scale anything. It's easy, but it's not.

>> Yeah, those examples you mentioned are all big use cases. One of the things Jensen talks about at GTC now multiple years in a row is the digital twins, the AI factories.

Paul Bloch

>> Absolutely. - Basically, you've laid out essentially AI factory use cases in all those examples. One, AI factories of the future, you seem to be the ingredient there at scale.

Paul Bloch

>> Yep, yep, yep. - And two, as digital twin technology comes in, how is that playing into some of the things you guys are engaging with?

Paul Bloch

>> Same thing, massive amount of data, massive access to data, real-time video inject. Because it's not just doing digital twin simulation. Once you've done it, you actually keep on optimizing it real-time. So once you've built that plant, you basically have cameras checking literally real-time what the workers do, what are the movements, where are you going? And you're doing online optimization. So it's a massive amount of data, real-time, which is exactly what DDN does.

>> I want to get your thoughts on digital twins, if you don't mind. The classic definition of digital twin was always grounded in manufacturing. Why? Because it's physical. It's a plant. It's core to the business. Now you're starting to see digital twins in other areas, retail, media. We're doing our own little digital twin project. So it's morphed into a representation of efficiency, simulation efficiency to get the main thing right, whether it's manufacturing or media.

Paul Bloch

>> Smart cities. That's a good example. You mentioned smart city. Dubai, UAE, for example, have imaged completely the area. Actually, I flew in the day before the Dubai storm last year, and we saw that Dubai wasn't as smart as they thought they were. The cars were literally driving into water. So it's not exactly smart. So there's tremendous advancement to be had, and that's really what digital twin will do. You'll be able to do real-time simulation of traffic, real- time location, optimization across the board.

>> It's about optimization really. At the end of the day, it's optimization.

Paul Bloch

>> Optimization, easier change, and efficiency.

>> Now, as you guys become the master of physical digital merging, which is essentially what digital twins is bringing us to, I have to just ask you, and this is off...

Paul Bloch

>> And then your physical AI. - Physical AI.

Paul Bloch

>> You're just starting your physical AI.

>> Well, it Gets really meta here.

Paul Bloch

>> And our avatars in 20 languages pretty soon.

>> We already have demos inside theCUBE where the hosts are literally bots that look like humans. I'm going to be replaced soon, so soon I'll be out of here. But the physical AI brings up this digital society. It's off-topic but I just want to get your thoughts as a leader and a founder. As crypto comes on, stable coins, as we start to see currency, money, programmable money, if software's eating the world 15 years ago, that memo from Andreessen, now the financial institutions are going to be impacted. So now you've got physical AI coming together with sovereign nations with digital money. Just your thoughts and your vision on how you see that playing out.

Paul Bloch

>> Very complex. If I knew I would be a trillion probably in the next .

>> You're already billionaire so that's all right.

Paul Bloch

>> Yeah. And again, it's about the fun. Both Alex and I are driven by the mission, the innovation. The money's great, but it's really not what's driving us. And so where do we see this? Again, you got to step-by- step plan the future and work with it. So crypto, I actually thought that crypto would be much more developed at this point, and it's still held by very few people. We do believe that crypto will have a play. Countries, hopefully, will keep on working together, but there's certainly a notion of clothing and keeping some of your value in the IP. So it's going to be a fine line. However, what we're finding still is that investments are coming in across, and that's potentially what's going to save us.

>> Yeah. We're seeing the plumbing and the systems have to get built. Now, going back more to NVIDIA's side, I have to ask about robots and robotics.

Paul Bloch

>> Yeah. - Because robotics has machine learning for many, many years. A lot of engineers working on robots. It's been in manufacturing, it's been around, but now you're starting to see the price point of the hardware, and now that software integrates with hardware, you're seeing a resurgence in robotics.

Paul Bloch

>> Sure. - And AI is certainly helping. Supercomputers helping, digital twins are helping. Just your vision on robotics.

Paul Bloch

>> Massive. That's the next wave. If you look at it, we've seen already separate waves from Gen. AI to Agenic.AI to Enterprise. AI and so forth, and RAG. But the next one, physical AI, is going to be extraordinary. Not so sure it's going to do for the human race, but for sure technologically it's going to be very interesting.

>> We had a joke on theCUBE last week. When did the population in San Francisco of robots surpass humans? Somebody was like, "2025. " I'm like, "Well, it depends on home penetration. I got a humanoid at home."

Paul Bloch

>> Not sure 2025, but pretty fast.

>> Pretty soon. - Right after.

>> Well, Paul, great to have you on.

Paul Bloch

>> Thank you very much. I really appreciate it. Always a pleasure.

>> Final question. To summarize DDN, for the folks watching, as you guys really go next level, you're expanding out the company. You guys are growing. You're opening up from the founder- led, founder-inspired, now with more stakeholders, bigger growth. What is DDN?

Paul Bloch

>> Customer satisfaction. At the end of the day, we care about our customers. We do everything for our customers. And when we make a commitment, we stick to it. We're not perfect, but we will basically make it right. So customer satisfaction, number one, innovation, number two, and obviously taking care of our people and having the best people in the industry.

>> Yeah, I love the mission over money. It's a great vision. Thanks for coming on theCUBE.

Paul Bloch

>> Thank you very much, John. - Appreciate you, Paul.

>> Appreciate it. - Say hello to Alex for me.

>> I will. . - All right. I'm John Furrier for

Paul Bloch

>> theCUBE. Thanks for watching.