The Flamboyant Billionaire Behind DDN, the Secret Weapon Powering AI's Data Deluge

From Obscure Supercomputer Labs to AI's Insatiable Appetite

For nearly three decades, DDN (DataDirect Networks) operated in the rarefied world of supercomputing, quietly designing storage systems for national laboratories and NASA. The Chatsworth, California-based company, co-founded in 1998 by French-Iranian entrepreneur Alex Buzari and Paul Bloch, built a reputation for eliminating the most stubborn bottleneck in high-performance computing: getting data to processors fast enough. Their software orchestrates parallel data flows so that no GPU or CPU ever idles waiting for information—an approach that kept them alive on a lean diet of government and research contracts, generating roughly $300 million a year.

The AI revolution changed everything. When Nvidia began lashing its GPUs into massive clusters for training large language models, its engineers hit the same wall that had frustrated physicists for years: conventional storage couldn't feed the chips quickly enough. An Nvidia vice president recalled that during an early supercomputer build, the storage system failed under load. DDN's technology, called in as a replacement, simply worked. That endorsement, despite CEO Jensen Huang reportedly calling Buzari's flamboyant Regency-era fashion “inappropriate,” opened the floodgates. Today DDN counts Elon Musk’s xAI and multiple national governments as customers for their homegrown AI supercomputers.

The business is now scaling at a pace that has transformed Buzari into a billionaire. Revenue hit $400 million in 2024, climbed to $500 million in 2025, and is projected to double to $1 billion in 2026. A $300 million investment from Blackstone in January 2025 valued the company at $5 billion. Buzari and Bloch each own about 40% of the equity, worth roughly $2 billion apiece. After years of bootstrapping and a near-death experience in 2001, DDN has become one of the most consequential behind-the-scenes players in the AI infrastructure buildout.

Why DDN's Storage Tech Became Indispensable for the AI Gold Rush

The GPU Idle Problem That DDN Solves

AI training clusters are ghastly expensive. A single Nvidia server can cost half a million dollars, and that is before astronomical electricity bills. Yet DDN contends that such servers often sit idle nearly two-thirds of the time because storage cannot supply data fast enough. Buzari’s core pitch is that his software—which now can be purchased independently of hardware—eliminates those waits, effectively turning a fleet of expensive GPUs into a continuously utilized asset. “We are the data engine that makes GPUs profitable and maximally productive,” he says. This value proposition is why hyperscalers and deep-pocketed AI labs are willing to pay a premium.

A Late-Stage Software Pivot Finally Pays Off

For years DDN insisted on selling a combined hardware-software package, a strategy that limited its addressable market. Only in 2023 did the company unbundle and offer its storage software as a standalone product. The shift has been transformative: software gross margins exceed 90%, compared to about 60% on hardware. The move also freed DDN from competing against commodity disk manufacturers and allowed it to partner with server vendors whose customers already own the physical drives. Blackstone executive John Watson describes the change as DDN going through “a kind of puberty in its late twenties,” a belated but critical step to capture the scale of the AI opportunity.

The Competitive Field: Vast Data's $30B Valuation and Looming Giants

DDN is not alone. The same forces that propelled its growth have minted formidable rivals. New York-based Vast Data, which provides high-speed storage for AI, raised $1 billion in April at a $30 billion valuation, dwarfing DDN’s $5 billion figure. Younger players like Weka and Hammerspace are winning significant contracts, while legacy giants Dell, Pure Storage, and NetApp have been slow to adapt but are now turning their attention to AI storage. The market is red-hot—shares of memory maker SanDisk surged 3,100% in twelve months, and Micron nearly quadrupled its revenue—but competition will only intensify as the spoils become more visible.

Blackstone's Strategic Value Beyond the $300M Check

Blackstone’s investment delivered more than a billionaire-making paper valuation for the founders. It gave DDN access to the private equity giant’s vast network of data center clients and its $325 billion real estate portfolio, which includes some of the largest digital infrastructure assets in the world. The relationship could prove crucial as DDN embarks on its most ambitious project yet: a potential 1.3-gigawatt data center in Wheeler, Texas, with a total price tag that could exceed $60 billion. Financing is still being arranged, but the backing of Blackstone—which owns and leases data centers to many of the same AI operators DDN serves—creates a pathway for vertical integration and captive demand that standalone storage companies rarely enjoy.

What DDN's Ascent Means for the AI Infrastructure Stack

For organizations building or expanding AI compute capacity, DDN’s trajectory highlights three practical considerations, all rooted in the company’s own disclosures and market position:

• Prioritize data throughput, not just raw teraflops. DDN claims typical GPU clusters waste most of their time waiting for data. Ensuring storage infrastructure can saturate the chip’s input/output bandwidth can effectively multiply the return on an expensive Nvidia deployment. Enterprises should evaluate whether software-defined storage solutions, like DDN’s now-unbundled offering, can cut idle time and lower the real cost per training run.
• Watch the software-only model as a margin and integration advantage. Since 2023, DDN has allowed customers to use its orchestration software with third-party disks. For any storage vendor contemplating a similar move, the DDN case suggests that software margins above 90% dramatically improve unit economics while broadening the addressable market. Buyers can negotiate for software-only deployments to avoid hardware lock-in while still gaining the performance benefit.
• Blackstone’s involvement creates a data center pipeline. With Blackstone as a major shareholder, DDN has a direct line to the world’s largest private data center platform. AI operators co-locating in Blackstone-owned facilities may find DDN storage pre-integrated or offered as a managed service. This could compress procurement timelines and reduce integration risk—though it also ties customers to a single storage ecosystem unless they explicitly demand alternative solutions.