Fujifilm is investing 10 billion yen ($63 million) to quintuple production capacity at its Kanagawa Factory for LTO magnetic tape, the enterprise storage medium that never actually died despite consumer abandonment decades ago. The expansion, running from fiscal years 2026 through 2028, is the largest dedicated LTO facility investment in the company's history and will focus on manufacturing the latest-generation LTO Ultrium10 tape. The driver is straightforward: generative AI is producing data volumes that need to be stored somewhere cheap. Cloud hyperscalers and data center operators have been quietly expanding tape usage because the economics are unbeatable for cold and archival storage. Tape sits in a drawer, consumes zero power when idle, and costs a fraction of disk or flash per terabyte. The LTO roadmap projects capacity growth through Gen 14, suggesting the consortium sees decades more runway. Fujifilm's technical lineage here is real. The company has been in magnetic tape since the 1950s, released its first LTO Ultrium tape in 2000, and pioneered commercial barium ferrite tape in 2011. The new LTO Ultrium10 uses what Fujifilm calls "fine hybrid magnetic particles" — a barium ferrite and strontium ferrite combination — representing genuine materials science iteration, not rebranding. The cybersecurity angle is quietly significant. Fujifilm explicitly highlights tape's air-gap capability as ransomware defense. A tape cartridge physically disconnected from any network is immune to remote encryption attacks. As ransomware becomes an existential threat to enterprise data, the ability to pull backup media offline and lock it in a vault is a feature no always-connected storage medium can match. This is a case where an old technology's constraints became virtues. Tape's slowness (sequential access only) is irrelevant for archival workloads. Its disconnectability, once a limitation, is now a security feature. Its low cost per terabyte, always an advantage, becomes decisive when AI training runs generate petabytes of intermediate data that regulations or business logic require retaining. The competitive picture matters. Fujifilm and Sony are essentially the only two remaining LTO tape manufacturers of consequence. A 5x capacity increase from one of them signals genuine demand signals from hyperscaler customers, not speculative positioning. Cloud providers don't sign up for multi-year supply expansions on a whim. The 20-year view is interesting: if AI data generation continues compounding, tape's cost advantage widens with every generation. The LTO consortium roadmap already projects Gen 14 at dramatically higher capacities. Fujifilm is positioning itself as the picks-and-shovels supplier to an AI gold rush — not selling intelligence, but selling the cheapest way to remember everything intelligence produces.
