Tape’s future may depend less on how many bits fit on a cartridge than on how quickly organizations can put data away and retrieve it. That tradeoff is shaping the latest Linear Tape-Open (LTO) roadmap: HPE, IBM and Quantum now target 365TB of native capacity for LTO-14, down from an earlier 576TB ambition, while emphasizing reliability, affordability and manageable recovery times.
The revision doesn’t suggest that magnetic tape has run out of room to grow. Instead, it reflects a practical constraint for organizations storing petabytes: a cartridge that takes a long time to fill can also take a long time to restore. Tape remains a useful tier for data that needs durable, low-cost retention but doesn’t require the quick access of SSDs or hard drives.
Capacity keeps climbing
LTO has increased cartridge capacity 400-fold since its first generation arrived in 2000 with 100GB of native storage and a transfer rate of 20 MB/s. The intervening generations raised capacity to 200GB with LTO-2, 400GB with LTO-3, 800GB with LTO-4, 1.5TB with LTO-5, 2.5TB with LTO-6, 6TB with LTO-7, 12TB with LTO-8 and 18TB with LTO-9.
The latest generation, LTO-10, has two native capacity points. Its original media, introduced in 2025, holds 30TB; an extended-length cartridge announced later raises that to 40TB. Both work in the same LTO-10 drives. For context, a 256TB SSD is approaching the market, and hard drives now reach more than 40TB. Those devices serve different access and performance needs, but their capacities make tape’s slower pace visible.
Across roughly the same 25 years, 3.5-inch hard drives grew from about 75GB to as much as 44TB, a roughly 585-fold increase. Tape has expanded more slowly, yet its capacity gains haven’t required a dramatic change to the cartridge form factor. Its continued appeal rests on that combination of density and a storage medium designed for long-term retention.
Two ways to fit more data
LTO can increase capacity by recording more bits on each area of tape and by packing more tape into a cartridge. LTO-10 uses both approaches. Its 15,104 data tracks represent a 68.5% increase over LTO-9’s 8,960, supported by a strontium-doped barium ferrite (BaFe) recording layer and higher linear density.
Tracking those narrower paths accurately calls for a redesigned, high-bandwidth servo system and a low-mass mechanism. The system is built to keep the head aligned even when shock or vibration affects the drive. For the 40TB cartridge, a thinner, smoother Aramid base film also makes room for more tape in the same shell. MagStor puts the tape length at about 1,335 meters, compared with approximately 1,035 meters in the 30TB version.
There’s considerable technical headroom beyond current products. In 2020, IBM and Fujifilm demonstrated strontium ferrite tape with an areal density of 317 Gb/in². They said that density could theoretically support a 580TB cartridge. The prototype used 56.2nm-wide tracks and a 29nm TMR read sensor. That demonstration shows what the medium might support; it doesn’t set a commercial LTO capacity target.
A slower roadmap, by design
When the LTO Program’s three companies extended the roadmap through LTO-14 in 2022, they projected near-doubling capacity with each generation: 72TB for LTO-11, 144TB for LTO-12, 288TB for LTO-13 and 576TB native for LTO-14. The 2025 update reset those targets to 70TB, 120TB, 210TB and 365TB, respectively. The earlier projection also put LTO-14 at 1.44PB with compression.
The revised figures, announced alongside new cartridge specifications in November 2025, put less weight on maximum density. The consortium says the targets are intended to support reliability, affordability, efficiency and lower cost per terabyte. It also argues that limiting cartridge capacity can reduce the time required to write and recover data. That rationale deserves attention: storage capacity has little operational value if restoring an archive takes longer than a business can tolerate.
Throughput, meanwhile, has not kept pace with cartridge capacity. LTO-8 transfers native data at 360 MB/s, LTO-9 at 400 MB/s, and LTO-10 remains at 400 MB/s. Compression can raise effective throughput to as much as 1,200 MB/s when data compresses at the assumed 2.5:1 ratio. But compression depends on the data, and the native rate is the more dependable baseline.
At 400 MB/s, writing 40TB in sequence would take roughly 28 hours in theory, before overhead. LTO-10’s top native capacity is 3.33 times LTO-8’s 12TB, while native transfer performance has risen by only about 11%. The LTO Program hasn’t published transfer-rate targets for LTO-11 or later generations, leaving performance growth an open question.
LTO-10 changes archive planning
LTO-10 also breaks with the format’s historical backward-compatibility pattern. Through LTO-7, new drives generally read cartridges from two generations earlier and wrote media from one generation earlier. LTO-8 and LTO-9 narrowed that support to one previous generation. Because LTO-10 uses a redesigned head, its drives read and write only LTO-10 media. The 30TB and 40TB LTO-10 cartridges remain interchangeable with each other.
That change matters to operators planning archives that must remain accessible for decades. Moving to LTO-10 means migrating data from LTO-8 or LTO-9 cartridges, or retaining older drives to read them. Maintaining a mixed fleet can extend access to legacy archives, but it also adds equipment and operational requirements to long-term storage plans.
Why tape still fits
Demand figures point to tape’s continued role even as solid-state and disk capacities rise. LTO capacity shipments reached 160.3 exabytes in 2025, down 9% from the year before. The consortium reported that shipments in the first quarter of 2026 were up 57% year over year, citing AI-driven data growth, traditional archiving, cyber-resilient backups, ongoing LTO-9 adoption and the start of the LTO-10 transition.
AI systems produce large volumes of training data, checkpoints, models, logs and generated assets, but much of that information doesn’t need to sit indefinitely on higher-cost storage. Tape offers high capacity at low cost per terabyte, consumes virtually no power while offline and can be physically removed from a drive to create an air gap from network-based attacks. Hardware encryption and the LTFS file system, which keeps indexing information on tape, add useful management and protection features.
As HPE’s Stephen Bacon put it, “AI has turned archives into strategic assets.” Quantum CEO Hugues Meyrath likewise pointed to rising data volumes alongside cost, energy and cyber-resilience pressures. Those comments reflect the industry’s case for tape, but the technology’s practical role is clearer in its tradeoffs: low-cost, power-efficient retention for data that doesn’t demand random access.
Sequential access remains tape’s defining limitation. It can’t match the random-access performance of HDDs or SSDs, and restoring large datasets can take time. The LTO roadmap’s more measured capacity targets recognize that reality. As archives expand, the useful measure of progress will be not just how much data a cartridge holds, but whether organizations can store and recover it at a cost and pace their operations can support.