Emerging Technologies · By Brian Fulcher
Prototype or demonstration. This article covers an integration test of experimental archival storage, with restricted access to pilot testing.
Last reviewed: November 7, 2026.
Keeping a file is only half the job. Someone still has to read it years later, after the software, storage hardware and people responsible for it have changed. A new ceramic-storage test asks a practical question: could a very different physical medium work with tools archive teams already understand?
What changed, and when
On November 5, 2026, Cerabyte announced successful S3 integration testing with CERN openlab. The company-issued release says the team wrote and retrieved CERN material, including publications, video and sample physics data. It quotes CERN Tape Archive Manager Vladimir Bahyl describing the results as encouraging and the evaluation as continuing.
That establishes a reported compatibility demonstration. It does not establish broad production deployment. The announcement comes from the vendor, with a named CERN participant’s account; it is not a published independent qualification report. Some republications carry November 6, but the original release is dated November 5.
How ceramic storage works
Cerabyte’s technical overview describes ceramic nanolayers on thin glass. Data become tiny patterns written with a laser and read optically using microscope-based equipment. Think of information represented by physical marks rather than held as an electrical charge. That comparison explains the idea; it does not mean an ordinary camera can read a finished archive system.
The separate software layer matters too. Amazon’s S3 documentation describes object storage: files and their descriptive information are stored as objects in containers called buckets. Applications use an interface, or API, to request operations such as writing and retrieving an object. A compatible interface can hide changes in the equipment underneath. This test concerns S3-style access to ceramic media; it does not mean CERN’s files were stored in Amazon’s cloud.
Who could benefit
The natural question is about data kept for long periods and read infrequently: scientific records, cultural collections or an organization’s historical documents. If the medium and surrounding system prove dependable, keeping familiar access tools could make evaluation less disruptive. These are potential applications, not verified deployments in each field.
CERN’s own project account gives useful context. It says tape remains its archival medium of choice and describes the ceramic collaboration as an evaluation. Its 2025 work included decoding patterned samples with a microscope. The listed next steps include reliability modeling and an open, self-describing format. Compatibility is one piece of a much larger preservation problem.
Availability and limits
Cerabyte’s current website restricts pilot tests to enterprise customers, research institutions and institutional investors. This is not a consumer storage product readers can simply buy. The reviewed sources do not establish general orderable availability, verified pricing, delivery geography or a production service commitment.
The release does not provide a detailed S3 conformance matrix, measured end-to-end throughput for this test, operating costs or a published failure-and-recovery study. Passing an integration exercise cannot fill those gaps. CERN’s earlier project page is background, not independent confirmation of every claim in the new announcement.
Cerabyte promotes very long media life and power-free retention. Those are vendor claims about the storage approach. The system still needs equipment and energy to write, locate and read data. Long-lived marks alone do not guarantee that future readers will have functioning hardware, usable software, intact encryption keys or permission to access the files.
Root Access Take
I’d start an evaluation with a restore exercise. Can a different operator recover a known collection, verify every file and explain a failure without help from the original project team? For infrastructure teams, that evidence would matter more than an impressive lifetime claim.
There is a human tradeoff too: durable records can preserve knowledge, but they can also preserve information people should no longer retain. Retention rules, access controls and a workable deletion policy belong in the design discussion. The site’s data lifecycle guide provides related context.
What to watch next
Look for a published test method and results that another archive team can repeat: supported operations, error rates, retrieval under load, recovery from damaged media and the cost of keeping readers available. An openly documented format and independent restore tests would help answer whether the data can outlast the people and systems that wrote it.
Primary sources
Cerabyte’s original announcement, November 5, 2026 · CERN openlab’s project account and 2025 results · Cerabyte’s mechanism overview · Cerabyte’s pilot-access statement · AWS documentation explaining S3 object storage. Company sources describe company claims; CERN’s project page supplies separate evaluation context.