AI's data dandy becomes a billionaire

Alex Bouzari spent 28 years building DDN into a specialist storage company for supercomputers. AI demand now pushes the Chatsworth, California, company toward $1 billion in annual revenue.

The storage bottleneck behind AI

Modern AI systems depend on thousands of graphics processing units working in parallel. Those chips can process enormous workloads, but they need a steady stream of data. Slow storage leaves expensive computing equipment waiting for files, which raises costs for companies building AI models.

DDN addresses that bottleneck with storage hardware and software. Its system breaks large files into pieces, sends those pieces across many drives and brings them back to the processors as they need them. The design resembles a highway system that keeps traffic moving across every lane.

The company began in 1998, when Bouzari and his Caltech friend Paul Bloch focused on research laboratories running the world’s fastest supercomputers. Those machines could perform more than a quintillion operations per second, but storage systems often failed to keep pace.

DDN grew into a profitable niche business. Revenue reached about $300 million a year before the AI boom expanded its customer base. Nvidia became a key partner after the chipmaker encountered storage failures while building its own AI supercomputers.

Nvidia’s GPUs could handle billions of parallel calculations, yet the system stalled when storage could not deliver data. DDN solved the problem for Nvidia and later won contracts with AI companies, data center operators and national governments.

Revenue takes off

DDN generated about $400 million in revenue in 2024 and $500 million in 2025. The company projects $1 billion in revenue for 2026, helped by contracts with Elon Musk’s xAI and government-backed AI projects.

The surge reflects a wider data center spending cycle. SanDisk shares gained 3,100% over the prior 12 months, according to figures cited in the company’s coverage, while Micron reported $41.5 billion in quarterly revenue after a year-over-year increase.

AI operators face a costly problem. A new Nvidia server can cost about $500,000 before power and cooling expenses. DDN says those servers can sit idle for about two-thirds of their operating time when storage systems fail to supply data.

DDN sells software that helps customers keep GPUs working through more of each workload. The software carries gross margins above 90%, compared with about 60% for hardware. That margin gap encouraged the company to sell its software on its own in 2023.

For years, DDN bundled software with its storage racks. Customers began buying their own drives, which made the bundle harder to sell. The standalone software business gave DDN access to more AI customers and improved its economics.

Blackstone adds capital and reach

Blackstone bought a $300 million stake in DDN in January 2025 at a $5 billion valuation. Bouzari and Bloch each hold stakes of about 40%, putting the value of each founder’s interest near $2 billion at that valuation.

Blackstone also gives DDN access to clients across the investment firm’s data center portfolio, which totals about $325 billion. That network could help DDN reach customers that need storage for large AI clusters.

The investment also sets a demanding benchmark. DDN must maintain growth as AI companies reconsider capital spending, power availability and the cost of training new models. A slowdown in data center construction would affect storage suppliers with high exposure to AI infrastructure.

Rivals chase the same customers

DDN competes with established vendors such as Dell, NetApp and Everpure. Smaller companies, including Weka and Hammerspace, also sell software for high-performance AI storage.

Vast Data has become DDN’s most prominent rival. The New York company raised $1 billion in April at a $30 billion valuation and sells storage software to CoreWeave, Nebius and Mistral AI.

Customers may favor specialized storage firms because AI workloads place unusual demands on data systems. Large technology companies can also build custom infrastructure and negotiate lower prices with hardware suppliers. That pressure could limit the margins that attract investors to storage software.

A company built through setbacks

Bouzari entered the storage business after he saw a removable hard drive at a government trade show. The personal computer market already had established suppliers, but he saw government laboratories as an overlooked customer group.

Researchers later told him that storage failures caused them to lose experiment data. Bouzari and Bloch built a system that could move data fast enough to support the laboratories’ supercomputers.

The founders raised almost $10 million in venture capital in 2001. Investors pushed DDN toward corporate customers as the economy weakened after the Sept. 11 attacks. The strategy drained cash and sent investors searching for a sale.

Bouzari and Bloch cut staff and asked employees to accept delayed pay. The company returned to profitability, and Bouzari bought out the investors in 2002.

DDN reached about $100 million in revenue by 2008 and $220 million by 2018 without outside capital. That history gave the founders control through years when few investors paid attention to high-performance storage.

A $60 billion ambition

Bouzari now plans a data center in Wheeler, Texas. The site could scale to 1.3 gigawatts, enough power to match the consumption of about 1 million homes. The project could cost more than $60 billion.

Bouzari is still arranging financing. The plan would place him on the other side of the infrastructure market, where he would build and operate the kind of AI capacity that creates demand for DDN’s storage systems.

DDN’s rise follows a pattern across the AI economy. Founders who spent years serving specialized technical markets now find their products at the center of a much larger spending race. Bouzari built storage for researchers who needed speed and reliability. AI companies now need the same qualities at data center scale.