Webcast
Hear how always-hot storage is reshaping cloud economics and unlocking new AI use cases.
For years, storage teams categorized data into neat tiers:
Then AI changed everything.
In our latest webcast, Gleb Budman (CEO & Co-Founder of Backblaze, called out the shift:
“There’s all this data that maybe was thought to not be valuable anymore. But now suddenly it is — and you need access to it.”
Archived videos, logs, long-forgotten assets — once considered cold — are being pulled back into active workflows. Why? Because multimodal systems demand historical, diverse, and richly contextualized data.
“With AI workloads, you’re feeding massive datasets into GPUs… and latency kills performance,” Patel said.
“If you’re pulling that from cold storage, you’re adding delay at every step.”
Cold storage had a purpose: cheap, deep, slow. It made sense for disaster recovery, compliance archives, or backups you hoped never to touch. But today’s workloads are constantly retrieving, iterating, and feeding compute. That turns yesterday’s cold data into today’s bottleneck.
Patel shared one example from the field:
“We had a media customer who needed to re-index their library after updating their asset management system. The cost and delay to pull it from Amazon S3 Glacier was so bad, they said, ‘We’re never doing that again.’”
That’s the real-world impact of treating hot data like it’s cold.
Backblaze’s B2 Cloud Storage (an object storage service) was built around a different assumption: data that might be used should always be ready. There’s no cold tier to retrieve from. No class transitions. No delays or rehydration logic.
“We don’t believe in hot versus cold,” said Patel.
“We believe in storing it once — always ready, always available — and priced so it’s still economical.”
In AI, that means:
The podcast discussion outlined a new reality: storage lifecycle rules are no longer static. What was “rarely accessed” last quarter could be foundational next quarter. Here’s why:
“You don’t know when that data becomes valuable again,” Patel emphasized.
“If it’s stuck in cold storage, you’re at a disadvantage.”
AI gets the headlines, but always-hot architecture matters for other use cases:
All of it may be “old” — but in modern workflows, age ≠ inactivity.
If you’re rethinking your cold/hot model, here’s what matters:
🎧 In the webcast, you’ll learn:
