Every few years someone declares tape dead, and every few years the shipment data disagrees. The more useful question for a backup architect is narrower: for the one copy that must survive a full compromise of your network and credentials, is tape still the right medium, and has the latest generation changed the practical case?
The answer is a qualified yes, and the qualifications matter as much as the answer.
What current LTO actually offers
LTO-9 holds 18TB native per cartridge, up to 45TB assuming 2.5:1 compression. LTO-10, which began reaching customers in 2025, was first delivered at 30TB native (75TB compressed at 2.5:1), after early expectations of 36TB were revised down by vendors. Its native transfer speed is 400MB/s, the same as LTO-9.
In November 2025 the LTO Program’s technology providers (HPE, IBM, and Quantum) announced a 40TB native LTO-10 cartridge, up to 100TB at 2.5:1, with product expected to ship in the first quarter of 2026. The gain comes from drive head refinements and a thinner Aramid-based base film that allows more tape length in the standard cartridge shell, and the announcement says the 40TB media works in existing LTO-10 drives. The Program’s published roadmap runs through generation 14, targeting 913TB per cartridge.
Compression ratios are a marketing assumption. Encrypted or already-compressed data, which describes a lot of modern backup output, will land well under 2.5:1, so plan on native capacity.
The capacity math
Take one petabyte of native backup data. At 18TB per LTO-9 cartridge, that is 56 cartridges. At 30TB, it is 34. At 40TB, it is 25. Fewer cartridges means fewer slots in a library, fewer media to label, track, rotate, and ship offsite, and a smaller physical vault. For a team whose objection to tape was operational overhead, a 55 percent reduction in cartridge count against LTO-9 is the relevant change. We have not included media or drive pricing here, because the cost per terabyte depends on vendor quotes, drive and library purchase, and volume. Get current quotes before you build a business case rather than relying on generic figures.
Demand is rising, not collapsing
The LTO Program’s annual shipment reporting shows how the market behaves. Shipments reached a record 176.5 exabytes in 2024, up 15.4 percent. They settled at 160.3 exabytes in 2025, down 9 percent, which was still higher than any year other than 2024. The Program’s most recent report says total shipped capacity in the first quarter of 2026 rose 57 percent from the first quarter of 2025, attributed to continued LTO-9 volume and the early LTO-10 ramp.
Those numbers describe the whole market, including hyperscale and archive workloads, so they do not prove that tape suits your environment. They do show that the format is in active use and investment, which reduces the risk of buying into a dead platform.
Why offline copies matter for ransomware
The ransomware argument for tape is structural. A cartridge that has been written, verified, and ejected from the drive is inert: it draws no power, accepts no commands, and cannot be reached by malware or a stolen administrator credential. Attackers who spend days inside a network deleting snapshots, wiping backup repositories, and disabling retention cannot touch a cartridge sitting in a vault. Our companion piece on air-gapped backup covers the broader pattern, and our piece on why attackers target the backup repository first covers the threat.
Tape also supports write-once media (WORM) in LTO, which adds protection against overwriting even if a cartridge is mounted later. Recovery from a mounted cartridge still depends on trustworthy catalog and encryption key management, so protect those with the same rigor.
The honest limits are just as real.
- Restore speed. At a native 400MB/s per drive, restoring hundreds of terabytes takes days unless you run many drives in parallel. Tape is the last line of defense, not the first, so keep faster restore tiers for routine and partial recovery.
- LTO-10 broke backward compatibility. Reports on the LTO-10 generation, including from heise, note that for the first time LTO-10 drives work only with LTO-10 media, because of a revised servo format and head design. An archive on LTO-8 or LTO-9 needs a drive that reads it. If you keep long-term copies, keep readable drives, or plan a migration, and budget for that.
- Human process. Someone has to rotate, transport, track, and destroy media, and run periodic restore tests. A tape set that nobody has restored from in two years is a hope, not a control.
- Library and drive failures. Mechanical systems wear out. Keep spare drives and test with real data.
Where tape fits in a 3-2-1-1-0 design
The 3-2-1-1-0 rule extends the classic 3-2-1 with one copy offline, air-gapped, or immutable, and zero errors after verification. Tape is one way to meet the offline requirement. Immutable object storage and logically air-gapped cloud tiers are others, and they restore faster with no courier logistics, but they depend on a provider’s account and credential model remaining intact.
A reasonable pattern for many mid-size environments is a fast on-premises or cloud tier for daily recovery, an immutable cloud copy for ransomware recovery, and a monthly or quarterly tape set that is physically offline and stored offsite. The tape copy is smaller and slower, and it is the copy you can count on if every online credential is compromised.
A decision checklist
Choose tape if you hold hundreds of terabytes or more of data you must retain for years, if your threat model includes full credential compromise, or if regulators or customers expect a demonstrably offline copy. Think twice if your data set is small, your restore objectives are measured in hours, or nobody on staff will own media handling.
Whichever way you decide, put one test on the calendar: pull a cartridge from the offsite set, mount it on a clean system, and restore a file with checksum verification. The time it takes, and who had to find the encryption key, are your real recovery numbers.