When Alex Bouzari walks into a room, people notice. His collar flared, peaking out from multiple jackets, the CEO of DataDirect Networks (DDN) stands out.

With a net worth in the billions, Bouzari is firmly placed in the ‘eccentric’ bracket of tech founders. But when it comes to the business of data, Bouzari is deeply serious.

DDN makes the storage hardware and software layer to bring data to bear on AI workloads, with the company growing alongside Nvidia - which Bouzari is keen to note has long been a customer.

Its platform is used by hyperscalers like Google and neoclouds like CoreWeave, as well as a wide array of enterprises.

Bouzari and DDN came from the world of supercomputers, "like Nvidia," he reminds us. "The problems you have to solve in AI are actually very similar to the ones you have to solve in high-performance computing. They are matters of scale efficiency, addressing the needs of performance."

The scale now is even larger, both in terms of individual workloads and the magnitude of opportunity.

"You don't have to be Elon Musk in order to do AI," Bouzari says, simultaneously noting Musk's xAI is a customer.

"I think the transformation is one of figuring out what the use cases are that are being enabled by AI. And today they're in financial services, life sciences, and autonomous driving. And then I think you will have things like the [Nvidia] Omniverse that are going to kick in across all industries with the ability to simulate things."

The simulation game

With simulation, businesses can theoretically "spin outcomes out in this parallel universe, and then see what works and what doesn't. It'll compress the time to get something implemented, and it'll significantly lower the cost."

This is easier said than done, with AI still an expensive proposition, and one that is difficult for most enterprises to experiment with. DDN, like a number of rivals, hopes to position itself as the guiding hand for these businesses, helping them deploy a full stack of tools.

Key to this, Bouzari believes, is understanding the different tribes within a traditional organization.

"When you look at enterprises, you have two constituencies to whom you have to speak and convince that there is a value," he says. "You have the IT organization, which is risk-based: 'Don't screw with my environment, don't bring something in that could have catastrophic consequences because then I'm going to get fired.'

"And then you have the business side of it, addressing the needs of the business stakeholders," he continues, referencing a pharmaceuticals company wanting to use AI to bring drugs to market faster.

That required a shift for DDN, which had historically worked with scientists and supercomputing experts. "We never dealt with the IT organization," Bouzari admits. "They have to be able to deploy an app on-device that's non-disruptive."

Once you can convince them that it won't break things or get them fired, Bouzari believes that you can sell them on more internal power.

"Our pitch to the IT organization is, 'guys, you've been focused on keeping the lights on, and so the C-suite takes you for granted. Now you have an opportunity to become strategic to the organization. Your value goes up significantly. Don't be afraid of it, embrace it.'"

The other part of winning over the enterprise community came in the form of fundraising, with the company bringing in $300 million from Blackstone at a $5 billion valuation this January.

DDN had turned down multiple other fundraising opportunities, Bouzari says, noting that the profitable business "does not need the money."

"Their pitch was that they invest in industry transformations," he explains, with Blackstone backing CoreWeave in the cloud and DDN at the data layer.

"Enterprises don't know who we are," he says, "or, [if they do], they think you're the guys who do these very massive scale things for governments and so on."

Bouzari hopes Blackstone will help change that perception, both through its brand and connections, and the fresh funds. The other way to increase awareness of DDN could be through going public.

"We could just push the button today, we've built up the infrastructure, the predictability, into the revenue and all that to do it," he says candidly. "But I think increasingly you see large companies stay private.

"For us, does it make sense to go public in order to increase the visibility of DDN, to accelerate the adoption of AI by enterprises? We're looking at it, but there's always a price to pay."

Nvidia CEO's advice

Bouzari recalls a conversation he had with Jensen Huang, founder and CEO of Nvidia, last year. "He said: 'Don't go public, you guys don't realize how good your life is. All you have to do is focus on the customer and develop the technology. That's all you have to do. My world is a whole other world.'"

Despite Huang's comments, DDN is "constantly going back and forth" on going public. "For every organization that knows DDN, you have 10 that don't know," he says, and notes that companies prefer to work with those where they can see that they're profitable and stable.

"In the public markets, you just get an opportunity to amplify your voice."

Whether DDN goes public or remains private, it aims to become an integral part of the enterprise shift, first experimenting with AI and then deploying it at scale.

"It's a fascinating time to be alive, because the world is pivoting into something completely different,” Bouzari says.

Pivots, however, come with pain.

The DeepSeek panic of January 2025 - where a cheaper AI model from China briefly caused markets to crash and tens of billions to be wiped from Nvidia, utilities, and data center companies - shows how markets do not understand AI.

On the other end of the spectrum, ever-growing valuations based on companies developing unproven AI models have increased the risk that the entire industry has overstretched. “People don't understand,” Bouzari says. “They just pour money into AI and then they freak out - it's the Dotcom Bubble all over again.”

"What I really focus on is the AI return on interest has to pencil out, right?" Bouzari says. "If people are buying GPUs, just because everybody's buying GPUs, stay away from them. Don't engage.

"[Some companies are like] 'I'm just going to raise money very easily at a stupid valuation.’ Then they get the GPUs, and it's ‘shit, shit, shit, I don't have the customers.'"

Instead, he is working with companies "where the ROI is obvious,” returning to the examples of financial services and pharmaceuticals. Elsewhere, the company is happy to take its time.

DDN, he says, is encouraging customers not to do huge AI deployments and chase trends. "Be cautious; implement a use case where you will get a quick win, see how it works, and then step it up."

Neoclouds and the risk factor

The pace of investment in the data center space has meant that operators cannot move slowly. Neoclouds like CoreWeave are developing at a rapid pace in an effort to stay ahead.

While Bouzari complemented the fellow Blackstone investment, he cautioned that neoclouds more broadly were focused just on deploying hardware. "The risk factor there is that, if you do not add services like the hyperscalers have, eventually, with Nvidia continuing to rev the generations of GPU, each new generation of GPU lowers the cost."

That then opens a space for a newer neocloud without legacy GPUs to come in cheaper, and cut into margins.

“They have to add the managed services, they have to add the software suites. They have to use the infrastructure as a way to monetize the services - right now they're only doing infrastructure.”

If they don’t do this, Bouzari predicts “consolidation, where many of the ones who are only looking at it as infrastructure will get absorbed by others.” Others, he says, “will go bankrupt.”

That said, given the current market fervor, “if you can raise money, life is good.”

The pace of change will start to stabilize in three to five years, Bouzari forecasts - at least for cloud and data centers. "But then the other thing you have is Edge devices are going to come in, with robots and autonomous cars. It will take longer for Edge devices to be available at a price point and at a reliability level where there will be hundreds of millions of them."

This represents "another massive opportunity," he says, with DDN currently developing technology to deal with the anticipated coming wave. That said, Bouzari admits he did not think it would take as long as it has for autonomous cars to become a reality.

For this wave and the next, having a flexible technology stack is key, he says. “You don't know how these things are going to evolve - training and inference have converged. The technologies are moving so fast, so you need a set of technologies that can address large scale, efficiency, and low latency, and you have to be able to consume your AI in the cloud, in multi-cloud, and on-prem.

“There is so much money pouring into this industry. Innovation is happening, but you don't know from where, so I think the important thing is developing technologies where it is flexible enough for what comes.”