The storage company that fit 100 terabytes into a drive the size of a paperback - and then, remarkably, declined to raise a fortune to tell you about it.
Here is a fact that sounds made up: in 2018, a company with roughly the headcount of a mid-sized restaurant announced a solid state drive that held 100 terabytes, packed it into a standard 3.5-inch case, priced it at about $40,000, and then watched as, for years, nobody built a bigger one. The company is Nimbus Data. The drive is the ExaDrive DC100. And the reason you may not have heard of either is, in its way, the most interesting part of the story.
The interesting part is that Nimbus Data did not do the thing that companies with a world-record product are supposed to do. It did not raise a giant venture round, hire a hundred salespeople, plaster the internet with its logo, and race to be acquired. Its founder and CEO, Thomas Isakovich, self-funded more than 90% of the company's growth. The math there is worth sitting with: a global storage business, 700-plus installations across 200 customers in a dozen countries, largely paid for out of the founder's own conviction rather than someone else's fund.
This is an unusual choice, and unusual choices in the storage industry are rare because the storage industry is not, culturally, a place for whimsy. It is a place for uptime, warranties, and spreadsheets about total cost of ownership. Enterprises buy storage the way hospitals buy generators - not because it is exciting but because the alternative is catastrophe. So when a vendor shows up and says, in effect, "we are privately funded, we intend to be profitable, we will not lock you in, and here, incidentally, is the largest drive on the planet," the pitch lands differently than a startup's.
Nimbus Data was founded in 2003. Isakovich is the kind of founder biographers love because the through-line is almost too clean: he designed and sold his first storage system as a teenager, worked at IBM and Oracle while at Stanford, then founded TrueSAN Networks before starting Nimbus. Four decades, one obsession. When someone spends that long thinking about where data lives, they tend to notice things the market has stopped noticing.
What Nimbus noticed was capacity. Most of the flash industry spent the 2010s racing on speed - more IOPS, lower latency, better benchmarks. Nimbus asked a quieter question: what if some customers don't need the fastest drive, they need the one that holds everything? A media archive, a genomics dataset, an AI training corpus - these workloads care less about microseconds and more about how many petabytes fit in a rack, and how much the rack costs to power.
"Nimbus Data aims to rewrite the rules of data storage by combining high-performance flash technology with powerful software and a radically customer-centric business model."
— Nimbus Data, company missionThe obvious question about a 100TB drive is: why can't everyone just do that? Add more flash chips, ship a bigger number. The answer is that a conventional SSD is bottlenecked by its controller - the chip that manages all that flash. Push a single monolithic controller past a certain capacity and it chokes. Nimbus's answer was a patented multiprocessor architecture, effectively several controllers working in parallel inside one drive, which is why it could reach densities the competition's single-chip designs could not.
The other half of the pitch is power. Data centers do not pay for storage once; they pay for it every month on the electric bill and again on the cooling bill. Nimbus claimed the ExaDrive drew roughly 85% less power per terabyte than competing enterprise SSDs, which it translated into about a 42% lower total cost of ownership per terabyte. Those are the company's figures, and they are exactly the kind of number a data-center buyer underlines twice.
There is a small joke buried in the form factor. A 3.5-inch drive is the same size as the spinning hard disks it is meant to replace - the boring beige rectangle that has lived in servers for thirty years. Nimbus took that exact silhouette and stuffed it with enough flash to hold something like 20,000 HD movies. The shape says "nothing to see here." The capacity says otherwise.
Nimbus is not a one-drive company, even if the DC100 is the headline. Its lineup spans the individual SSD up to full racks, all stitched together by its own software.
The high-capacity solid state drives, including the record-setting 100TB DC100, built on patented multiprocessor controllers.
Billed as the first multiprotocol Ethernet SSD - flash connected straight to the network for disaggregated storage.
All-flash systems and racks aimed at cloud, AI, enterprise, and edge infrastructure.
A multiprotocol all-flash data platform for mission-critical enterprises and cloud providers, unveiled in 2026.
The operating system and data-management layer: snapshots, replication, data reduction, encryption, high availability.
Zero vendor lock-in and flat-rate support - contract terms pitched as a feature, not fine print.
The business model underneath is straightforward B2B: sell dense, efficient hardware bundled with capable software, to people who move a lot of data and hate their electric bill. The twist is the go-to-market. Where rivals lean on the standard enterprise storage playbook of aggressive contracts and renewal traps, Nimbus made "we won't lock you in" part of the sales pitch - a small act of aggression aimed squarely at how the rest of the industry behaves.
The reported customer list reads like a spread from three different magazines. There is the entertainment tier - Walt Disney Animation Studios, DreamWorks - whose render farms generate absurd volumes of data. There is the defense and enterprise tier - Raytheon, Lockheed Martin, Citrix. There is eBay, whose scale speaks for itself. And there is, delightfully, WWE. Somewhere a wrestling storyline and a fighter jet's simulation data may be sitting on drives from the same 14-person company in Irvine.
"The ExaDrive DC100 offers more than three times the capacity of the closest competitor - 100 terabytes in a standard 3.5-inch drive."
— From the 2018 launch of the world's largest SSDThe common thread is workloads that are data-hungry rather than latency-obsessed: archives, media, high-performance computing, and increasingly AI, whose training pipelines have turned "how much can you hold, cheaply and coolly" into one of the more urgent questions in infrastructure. Which is the quietly fortunate thing about Nimbus's timing. It made its dense-flash bet in 2003, spent two decades looking early, and then the AI storage boom arrived and made the bet look prescient.
There is a version of this business that raised $200 million, spent it on brand and pipeline, and either flamed out or got quietly absorbed into a larger vendor's catalog. Nimbus chose the other version. Third-party estimates put its annual revenue in the tens of millions against a headcount you could seat at two dinner tables, which is another way of saying revenue per employee is the kind of number that makes venture investors wistful about the deal they didn't do. The tradeoff, of course, is reach: a self-funded company grows at the speed of its own cash, not someone else's, and there are corners of the market it will simply never sprint into. Nimbus appears to have decided that is a fair price for staying in control of its own story.
Thomas Isakovich founds Nimbus Data, largely self-funding the company's growth.
Early all-flash systems - the S-Class, E-Class, and Gemini lines - push array capacity into hundreds of terabytes with long warranties.
The ExaFlash platform and software-defined ExaDrive architecture lay the groundwork for extreme-density SSDs.
Launch of the ExaDrive DC100 - 100TB, the world's largest SSD, at roughly $40,000.
The AFX operating system and ExaFlash One array arrive; latest reported funding round (~$5M).
The company reports five consecutive years of profitability, expands to Miami, and rolls out the FlashMax platform.
The tidy way to end a profile is to declare a company brilliant or doomed. Nimbus Data resists both. What it offers instead is a small case study in a truth the tech industry keeps rediscovering: you do not have to be large to be indispensable, and you do not have to raise a fortune to build something no one else has. A 14-person company held the world capacity record for years by solving one hard controller problem that everyone else routed around, and by refusing to spend its way into the ground reaching customers who, it turned out, had real budgets and a real problem.
Whether that record holds forever is not the point. Rivals will eventually ship denser drives; the number always moves. The point is the shape of the bet - pick the lane no one is racing in, stay solvent, and let the market wander over to your side of the road. In 2026, with AI turning storage capacity into a bottleneck, the road is a little more crowded than it used to be. Nimbus Data has been standing there since 2003.