Fujifilm to 5x tape production in $63M AI archiving bet

Fujifilm is spending about $63 million to quintuple its magnetic tape production capacity, a bet that the AI era's data explosion will keep archives growing. The investment, announced Thursday, targets the company's Kanagawa Factory in Odawara and will run from fiscal years 2026 through 2028, with the fivefold increase measured against fiscal 2025 levels. That makes it "the largest investment Fujifilm has made to date in dedicated LTO tape facilities," according to the company's announcement.

The driver is generative AI, which Fujifilm says is producing "exponential growth" in data needs worldwide. That growth is pushing companies to store more data for longer, and tape is stepping back into the spotlight because it offers low costs and offline security - meaning it's immune to cyberattacks that hit always-connected storage. For decision-makers, the message is clear: the AI boom is not just about compute; it's about where all that data lives, and tape is becoming a first-class citizen again.

For decades, magnetic tape was written off as a dying medium, relegated to the back of IT closets. But the math has flipped. Tape costs a fraction of hard drives or flash per terabyte, and because it's offline, it's invisible to ransomware and other network-based threats. That combination is exactly what enterprises need as AI models generate terabytes of training data, logs, and synthetic content that must be retained for compliance, audit, and future model training. The low cost per gigabyte makes tape the natural home for cold data - the stuff that's rarely accessed but must be kept for years.

Fujifilm's move is a direct response to that demand. The company will expand manufacturing and add personnel at its Kanagawa site, primarily to produce the latest-generation "FUJIFILM LTO Ultrium10" tape. LTO, or Linear Tape-Open, is the industry standard for tape storage, and the Ultrium10 generation is the newest iteration, offering higher capacity and faster transfer speeds than its predecessors. By scaling up fivefold, Fujifilm is positioning itself to capture a wave of orders from hyperscale data centers, cloud providers, and enterprises that are rethinking their storage architectures. The investment also signals something broader: the AI boom isn't just about GPUs and data centers; it's creating a downstream storage economy that spans tape, disk, and flash.

While Nvidia and others race to build compute, the data those systems produce has to live somewhere. Tape's economics make it the natural home for the coldest data - the stuff that's rarely accessed but must be kept. That's why Fujifilm's bet is less about nostalgia and more about a structural shift in how data is managed. For executives, the implications are practical. If you're running a company that generates large volumes of data - whether from AI models, IoT sensors, or customer transactions - your storage costs are about to become a bigger line item. Tape can cut those costs dramatically, but it requires a different operational mindset: you need to classify data, decide what's cold, and build workflows that move it to tape. The offline security angle is equally important, as ransomware attacks continue to target always-on storage.

Fujifilm's investment also raises the stakes for competitors. Sony, the other major LTO tape manufacturer, will likely need to respond with its own capacity expansion, and storage vendors like IBM and Quantum that build tape libraries will benefit from a more robust supply chain. For CIOs and CFOs, this is a signal to revisit their storage strategies and negotiate long-term tape supply agreements before demand outstrips capacity. The project runs from fiscal years 2026 through 2028, so the capacity won't come online overnight. But the direction is clear: tape is back, and it's not a niche play.

As generative AI continues to multiply data, the companies that treat archiving as a strategic asset - not an afterthought - will have a cost and security edge. Fujifilm is betting $63 million that you'll need that edge. The investment is a reminder that the AI supply chain extends far beyond chips and servers, into the unglamorous but essential world of storage. For boards and executives, the takeaway is to start treating data retention as a competitive lever, not a back-office chore. The companies that lock in tape capacity and build efficient archival workflows now will be the ones that avoid the storage cost crunch later.