DDN CEO On The Company’s AI Mission And The ‘Essential Role’ Partners Play

‘At the end of the day, the key is data intelligence, and so the underlying hardware and infrastructure is commodity. I think how Snowflake and AWS look at it is the right way. They deliver services. The infrastructure is commodity,’ says DDN CEO Alex Bouzari.

DDN is typically classified as a storage vendor, but that’s not how CEO Alex Bouzari describes his company.

Instead, Bouzari, in an exclusive conversation just before the start of this week’s big Nvidia GTC event, told CRN that DDN is a data intelligence company, and that that data intelligence is nearly completely focused on providing the data needed to build AI models and AI inference capabilities.

Bouzari discussed a couple of recent technologies introduced Monday, including IndustrySync, a new suite of one-click AI-engineered solutions for financial services, life sciences, and autonomous driving, and xFusionAI, which merges AI model training and inference into a single optimized platform.

Bouzari also took time to clear up any idea that the Chatsworth, Calif.-based firm is just a storage vendor. He noted that the underlying hardware for storage, as well as for compute and networking, are all commodities today, with the value all in the software.

“[The enterprise pivot] from on-prem to the cloud was all about the data,” he said. “It’s not about storage. It’s not about infrastructure. AWS is delivering services powered by some infrastructure, but the value is in the services. The infrastructure is commodity. Same thing here.”

With its shift to data intelligence, DDN’s business is now focused on AI, Bouzari said.

“It’s most of our focus,” he said. “Look, even our HPC customers are transforming their workloads into AI. We’re in a world economy pivot being enabled by AI in everything we do at work, at play, health, security. It’s crossing everything. And AI is certainly our very high growth engine. I think our AI business in 2024 quadrupled over 2023. It’s significantly more than half of our revenue now. It will most probably double again this year, at least.”

There’s a lot going on at DDN and its push to be a prime provider of data intelligence for AI. To learn more, read CRN’s entire discussion with Bouzari, which has been lightly edited for clarity.

DDN is introducing some big technologies at Nvidia GTC. What’s new?

Basically, three things. What we’re increasingly seeing is that in order for AI to really get in its stride, it requires enterprise adoption. Enterprise adoption means it has to be easy to deploy, easy to manage for enterprises regardless of size, it has to be non-disruptive to the IT organization, and it has to add business value to the organization. So we’ve put together what we call IndustrySync. Our customers time and again ask us to package a solution that can get deployed more easily and faster to provide the benefits of AI transformation and the business outcomes from AI without having to get into the weeds. We started out with three industries where the AI ROI is the most obvious: financial services, life sciences, and autonomous driving. In all three industries, we have lots of customers, but they have a certain level of infrastructure, a certain level of sophistication, a certain level of staffing. For every one of those, there’s another 10 or 20 [potential customers] who want to deploy but they don’t have the staffing or the budgets or the time to do it themselves. And so packaging and integrating the solutions is something that they’ve been asking us for. We’re starting with these three industries, and then will broaden it.

For these industries, the business outcome benefits are obvious. In financial services, with algo (algorithmic) trading, we give a 10x improvement in processing speed. It’s substantial. That’s why we have so many financial services industry customers for AI. Life sciences, same thing. Pharmaceuticals companies doing drug discovery significantly shrink the time and cost it takes to bring a new drug to market, costs that are measured at the low end in the tens of millions of dollars and at the high end billions of dollars. So it’s very significant. And with autonomous driving, you have to capture lots of sensor data in real time. You have to process it and analyze it. And everybody’s racing to get to Level 5 autonomy. …

IndustrySync is a set of integration services. DDN is not a systems integrator. We want to partner with systems integrators and VARs and resellers and so on. We definitely don’t want to take business away from them. This is just a way to facilitate the deployment of enterprise AI, the same way Nvidia is not a systems integrator, yet they get involved with reference architectures and best practices. They send people to customer sites to help them basically integrate Nvidia technologies into their environment.

We are doing the same thing. We’ve done hundreds and hundreds of these deployments, and in these three industries in particular. So we’re like, ‘Okay, we have a huge amount of knowledge about how to make these deployments easier, faster, less painful for customers. So let’s just step in and do it for them.’

Where do channel partners come to play?

Channel partners will do the physical deployment. We will be there alongside them in addressing the challenges a channel partner won’t necessarily be able to handle. We don’t want to take money away from the channel partners. Channel partners play an essential role in this.

As a matter of fact, we’re increasingly shifting to become a channel-driven organization. We do not want to sell direct. We do not want channel partners to be a fulfillment vehicle. We want the relationship to be symbiotic, to be value-add. We want our channel partners, our VARs, our resellers, our systems integrators, to get value, deliver value to customers with the services that they’re bringing to bear, the integration of the solutions. These integrations are oftentimes in the customer’s data center. Then you need the domain experts to come in and ensure that everything is working perfectly. We are the domain experts. They are the integrators and the deployers of the solution. We bring domain expertise the same way Nvidia brings in domain expertise. We bring the domain expertise so that the customer says, ‘Oh, wow, this was phenomenal. I’m going to step it up.’ It’s a way to accelerate the adoption of AI by enterprises.

What is the second new technology DDN is introducing at Nvidia GTC?

Yes, DDN xFusionAI. We have our EXAScaler parallel file system, and we’ve done hundreds of millions of dollars of AI revenue with it. It’s solving the problems of scale, efficiency at scale, pre-training and post-training of large language models, accelerating all of that, helping GPUs run at 100 percent efficiency. The EXAScaler software gives customers the ability to do it in a smaller footprint, because there is not enough data center space with enough power and GPUs in the world.

We also just launched DDN Infinia few weeks ago to significantly improve the latency attributes on the inference side of things. So think agentic AI, multi-modal. The data is increasingly multi-modal. It’s text and images and audio and video and so on. AI models have grown substantially in size. We have some customers who are north of a trillion parameters in their models. The listing of the models is something that requires a certain type of technology. That’s what Infinia does, while addressing inference.

xFusionAI integrates these two together to solve the full end-to-end AI workflow. A couple of years ago, people were saying, ‘Well, you need one infrastructure for model and model training, and another for inference.’ Now those worlds are coming together. It’s not realistic to ask organizations to have two different sets of infrastructure for two different parts of AI. So by bringing EXAScaler and Infinia together, integrating the two together under xFusionAI, we’re basically covering all the elements and all of the aspects of the AI framework, from the listing, pre-training, post-training, inference, and insight, addressing latency requirements, acceleration, and so on.

That’s basically what xFusionAI is doing. Because the investments are so substantial, you don’t want to invest in one infrastructure for inference and one for model training. You have to extract more value. You need flexibility in on-prem and multi-cloud approaches because of the scarcity of GPUs, data center space and power, and so on. It has to be a hybrid model, where they can grab the data wherever it might reside. They have to be able to process it in the most cost-effective locations, at the right level of SLA (service level agreements), and then they need to get insight. So customers need a very flexible hybrid model where you’re doing on-prem and multi-cloud, and you’re hitting all of the pieces of the pipeline so you can decide what to do where. xFusionAI brings all that together so organizations can maximize the business outcome value from their AI journey, their AI initiative, their AI transformation.

OK, let’s zoom out a bit. How do you define DDN?

We maximize business outcomes for any organization that has an AI initiative underway. DDN is a software capability that accelerates, enhances, and delivers flexible outcomes driven by AI for businesses in every industry and use case. Historically, we delivered appliances in data centers that maximized efficiency in terms of a given infrastructure. Today, the value we’re delivering is really in two places. It is above the stack, meaning the acceleration of the AI frameworks on both the model training and inference sides. But it’s also use cases. So let’s say you’re a bank, and fraud detection is a very big deal. Fraud detection requires real-time catching of fraud and determining what is a false positive, is a positive, or a negative. We are helping organizations extract value from AI in their business outcomes, powered by AI framework acceleration and enabled by efficiency in the data center, both on-prem and in the cloud. …

Our job is to deliver the highest value to the business layer in industries driven by AI. That’s what our whole focus is. All the technological developments we’re doing are focused on that. The acceleration of the AI frameworks is very important, because it’s an industry which is moving so quickly. You have to be able to make TensorFlow or PyTorch more efficient. I think Jensen [Huang, Nvidia CEO] was saying that Nvidia has 150 developers just on TensorFlow alone. All they do is optimize that. So optimization of the AI frameworks is very important in order to drive the business outcome value that each industry and each company is pursuing. That also means you have to make it easier for organizations to deploy these things.

We deliver data intelligence. AI without data intelligence is meaningless. All these things that people are doing, all these investments in GPUs and infrastructure, in networking and storage and what have you, the whole point is to gain insight to deliver business value. We are in the data intelligence business. We’re not in the infrastructure business, we’re not in the storage business, we’re not in the on-prem or cloud business. We’re in the data intelligence business. We maximize the value that organizations get from their data.

So it’s incorrect to call DDN a storage vendor?

Would you call Snowflake a storage vendor? It’s data, right? It’s the same thing. But it’s data that is delivering value to enterprises and organizations. It’s their way of delivering data intelligence and data enablement for IT organizations. We’re the same. We’re delivering data intelligence in AI initiatives.

Look, storage is a commodity. Networking is a commodity. Compute is a commodity. It is the underlying hardware. It’s commodity. I was talking about it with Jensen a couple of weeks ago. I said, ‘OK, do we all agree that all of the hardware is commodity? He was like, ‘Yeah, absolutely, it’s all commodity.’ You have to extract value that is business-enabled and do it non-disruptively for the IT organization. Otherwise, the IT organization won’t embrace it.

[The enterprise pivot] from on-prem to the cloud was all about the data. It’s not about storage. It’s not about infrastructure. AWS is delivering services powered by some infrastructure, but the value is in the services. The infrastructure is commodity. Same thing here. The difference, in the world of AI, is not in the shift from on-prem to cloud. You need both on-prem and multi-cloud, simply because you cannot consume some things or move massive amounts of data in the cloud because it’s way too expensive. [For example,] in a multi-modal world, you’re not going to move videos around in the cloud. AWS would be very happy, but you will go bankrupt if you try to do that. So it has to be hybrid, meaning both on-prem and multi-cloud. At the end of the day, the key is data intelligence, and so the underlying hardware and infrastructure is commodity. I think how Snowflake and AWS look at it is the right way. They deliver services. The infrastructure is commodity.

Our conversation has focused on AI. Is AI now the entire focus of DDN?

It’s most of our focus. Look, even our HPC customers are transforming their workloads into AI. We’re in a world economy pivot being enabled by AI in everything we do at work, at play, health, security. It’s crossing everything. And AI is certainly our very high growth engine. I think our AI business in 2024 quadrupled over 2023. It’s significantly more than half of our revenue now. It will most probably double again this year, at least. But even our HPC customers, forget AI workloads, what they’re doing is being AI-enabled. I think AI is permeating everything. And so we are an AI company, not because it’s cool to be AI, but because everything is becoming AI-enabled. It’s permeating everything.

[Like Nvidia is not really a semiconductor company,] I would argue that we’re not a storage company. But I think this will need to evolve. We came up with the term ‘data intelligence,’ and then six months later, Databricks adopted it. They call themselves a data intelligence company as well. Somehow what we came up with is getting traction. It’s all goodness. They are a data intelligence company. I think Palantir is data intelligence. We’re all looking at it differently, but it’s all about getting intelligence from data in different ways.

[Also,] increasingly we’re selling software. We’re deploying software in the cloud and hyperscalers. We now have Infinia deployed in hyperscalers. The underlying hardware, it’s commodity. If some customers want us to provide a turnkey solution because it’s easier, absolutely we will do it. But the real value is in software because you can deploy it more easily. You have your own hardware. Here is a DDN Infinia. Do whatever hardware. We don’t want to be in the hardware business.

Anything else we need to know about DDN?

We’re pushing very hard to accelerate our channel programs. We’re investing heavily in that we want to become, first and foremost, a channel-friendly company. We’ve hired somebody to run our reseller and VAR channels. We’ve hired somebody to run our GSI channels. We’ve hired somebody to run our server vendor channels. We’re pivoting into full-on channel. We don’t want to sell direct. This industry is moving too fast, and to really deliver value to customers, we need partners. Doing it direct makes absolutely no sense.