Yearly confirmation: Tape recording technology remains a viable option
LTO (Linear Tape-Open) tape storage, an open-standard technology co-developed by HPE, IBM, Quantum, Sony, and Fujifilm, has emerged as a reliable and cost-effective solution for storing and preserving the large volumes of data generated by AI applications.
High Capacity and Scalability
The latest LTO generations, such as LTO-9, provide up to 18TB native storage per cartridge. Future generations (LTO-10) are expected to offer 36TB, a scalability that suits the massive data volumes generated in AI and machine learning workflows.
Cost-Effectiveness
Compared to disk or cloud storage, LTO tape has a low total cost of ownership, making it an economical choice for long-term storage of large AI datasets which might not require frequent access.
Long-term Data Retention and Durability
LTO tapes can reliably preserve data for over 30 years, supporting compliance and archival needs in AI projects where historical data may be crucial.
Security Features
Newer LTO generations incorporate hardware encryption and WORM (Write Once Read Many) capabilities, preventing data tampering and supporting immutability required for regulatory compliance and data integrity in AI applications.
Offline/Offsite Storage Capability
Tape storage can be kept offline and offsite, reducing risks of cyberattacks, ransomware, and data loss—an important consideration for secure AI data archiving.
Compatibility and Flexibility
As an open-standard technology, LTO is supported by multiple vendors, allowing for a variety of solutions and future-proofing storage infrastructure.
Contextual Considerations
While tape storage offers these advantages, it has slower data access times compared to disk storage due to its sequential access nature, which can delay retrieval when rapid data access is needed.
The Future of LTO Tape Storage
Emerging technologies like holographic optical ribbon storage (HoloMem) promise even higher capacity and longevity than LTO, but LTO remains widely established and integrated into AI data storage strategies as of 2025.
IDC's research veep for infrastructure software platforms, Phil Goodwin, believes that tape plays an important role in modern data architectures. LTO 10 is set to debut in the current year, promising 36TB per tape. If the LTO team maintains its usual four-yearly release cadence, LTO 14 will arrive around 2040, predicted to pack 576TB on each tape.
The amount of tape shipped in 2024 represents a 15.4 percent increase from 2023's 152.9 Exabytes. The three companies HPE, IBM, and Quantum, who back the Linear Tape-Open (LTO) Format, shipped 176.5 Exabytes of tape storage media in 2024.
In summary, LTO tape storage provides a reliable, secure, and cost-effective solution for storing and preserving the large volumes of data produced by AI applications, particularly when long-term retention with low access frequency is required. Its evolving capacity and security features support the growing demands of AI and machine learning workflows while balancing budget and risk management considerations. The unique combination of scalability, cost-efficiency, and cyber resilience is highlighted as a reason for its continued relevance.
In the context of AI applications generating large volumes of data, LTO's scalability, as evident in the expected 36TB storage capacity of LTO-10, aligns well with the data needs of such workflows.
The cost-effectiveness of LTO tape storage, coupled with its low total cost of ownership, makes it a practical choice for long-term storage of AI datasets that may not require frequent access.
As AI projects often require compliance with stringent data retention and integrity regulations, LTO's hardware encryption and WORM capabilities offer essential security features for preserving the data's immutability and preventing tampering.
