Samsung is reportedly developing nearline SSDs ranging from 250TB to 1PB. That's one petabyte—enough storage for 8,000 copies of GTA V, if that's your benchmark of choice.
The engineering achievement is genuine. Getting that much storage into a single drive form factor requires serious innovation in NAND density, controller architecture, and thermal management. Samsung's semiconductor team is among the best in the world, and this is the kind of thing they're good at.
The question isn't whether they can build it. It's who needs it, and whether this is solving a real problem or just flexing manufacturing capability.
As someone who's dealt with storage scaling at a startup, I'm genuinely curious about the use case. Data centers are the obvious answer—hyperscalers like Amazon, Google, and Microsoft that need to store exabytes of data. For them, consolidating storage into fewer drives means lower power consumption, less physical space, and simpler infrastructure management.
But the economics have to make sense. Nearline SSDs occupy a weird middle ground between performance and capacity. They're faster than hard drives but slower than performance SSDs. They store more than performance SSDs but less than a rack of hard drives. The question is whether that trade-off justifies the cost.
For most workloads, probably not. If you need fast storage, you want NVMe performance SSDs. If you need massive capacity, you want cheap spinning disks. The nearline SSD sweet spot is pretty narrow—warm data that's accessed occasionally but needs faster response times than hard drives can provide.
The 1PB number is particularly interesting. That's almost certainly a technology demonstration rather than a shipping product. Samsung is showing what's possible, probably with an eye toward securing contracts with hyperscalers who want to plan their data center roadmaps years in advance.
The actual products that ship will likely be in the 30-100TB range—still massive, but more economically rational. A 1PB drive would cost more than most companies' entire storage budgets.
What's driving this is the explosion in data generation. AI training creates enormous datasets. Streaming video needs massive archives. Genomics research generates petabytes per project. The demand for storage is real and growing exponentially.
But storage is also getting cheaper every year. The question is whether dense SSDs or cheap hard drives win the economics battle. Right now, for cold storage, hard drives are still unbeatable on cost per terabyte. For hot storage, SSDs win on performance. Nearline is the awkward middle child trying to justify its existence.
