Annual confirmation of the persistence of cassette tapes' existence
In the ever-evolving landscape of data storage, the Linear Tape-Open (LTO) format continues to play a significant role, particularly in supporting artificial intelligence (AI) applications. Co-developed by HPE, IBM, Quantum, Sony, and Fujifilm, the LTO format offers high-capacity, cost-effective, and scalable storage media that caters to the enormous data retention needs of AI training and inference.
Last year, shipments of LTO tapes reached an impressive 176.5 Exabytes, a 15.4% increase over 2023, demonstrating strong demand and ongoing relevance despite the growth of other storage technologies. This growth is expected to continue, according to IDC's research, indicating a likely annual "Tape is still not dead" update for years to come.
One of the key contributions of LTO tape for AI workloads is its high capacity per cartridge. The latest generation, LTO-10, debuted in 2024 with a native capacity of 36TB per tape, and the roadmap projects up to 576TB for LTO-14 around 2040. This scaling allows for storing vast amounts of training data and model checkpoints produced by AI systems efficiently.
Tape's cost-effectiveness and scalability are also crucial factors. Remaining among the most economical formats for storing petabytes to exabytes of data, tape is indispensable given the exponential rise in AI model sizes and datasets with tens to hundreds of exabytes of data supporting large-scale models training. Tape is particularly useful for cold or archival storage of large AI datasets and intermediate results that do not require real-time access.
Moreover, LTO tape is valued for its durability and long-term retention, making it suitable for retaining AI training data and compliance-related datasets over extended periods with low maintenance cost.
While AI model training increasingly demands high-bandwidth, low-latency storage solutions, LTO tape addresses a complementary niche: large-scale, long-term data storage to offload less time-sensitive data from more expensive active storage tiers.
Bruno Hald, General Manager for secondary storage at Quantum, and Phil Goodwin, Research Vice President for infrastructure software platforms at IDC, both expressed their sentiments about the importance of tape in modern data architectures. Hald stated that LTO tape technology is essential for organizations navigating the AI/ML era, while Goodwin emphasised its role in secure, sustainable long-term data storage for enterprises.
In summary, in 2024, LTO tape supports AI data storage by providing scalable, economical, and reliable archival and bulk storage. This enables organizations to manage the vast data volumes underpinning large AI models while balancing costs and storage efficiency. The unique combination of scalability, cost-efficiency, and cyber resilience makes LTO tape a valuable component for enterprises seeking secure, sustainable long-term data storage.
- The scalable storage capacity of LTO tape makes it ideal for AI applications, as demonstrated by the native capacity of 36TB per cartridge in LTO-10 and a projected capacity of up to 576TB for LTO-14.
- LTO tape also offers cost-effectiveness, remaining among the most economical formats for storing large petabytes to exabytes of data, which is crucial for AI systems dealing with exponential data growth.
- In addition to its data storage capabilities, LTO tape's long-term retention and durability make it a valuable tool for organizations in the AI/ML era, providing secure and sustainable long-term data storage for compliance-related datasets.
