The future of long-term data storage is clear and will last 14 billion years
Summary
SPhotonix is commercialising 5D Memory Crystal technology that stores data as tiny voxels in fused silica glass, written with femtosecond laser pulses. The technique encodes information using voxel position (x, y, z) plus optical orientation properties (birefringence), creating five-dimensional storage that the company says is extremely durable and dense.
The medium can hold up to 360 TB on a 5-inch glass platter and, according to SPhotonix, voxels etched in silica should remain readable for about 13.8 billion years at 190°C. The company has raised $4.5m to move the tech from TRL 5 to TRL 6 and is engaging hyperscalers to trial prototypes in data centres. Current speeds are modest (write ~4 MBps, read ~30 MBps) but SPhotonix aims for ~500 MBps read/write within a few years to match archival tape performance.
Key Points
- 5D Memory Crystal encodes data in fused silica using femtosecond lasers to create birefringent voxels (x,y,z + optical orientation).
- A 5-inch glass platter can store up to 360 TB; silica’s stability underpins a claimed lifespan of ~13.8 billion years at 190°C.
- SPhotonix raised $4.5m and plans to advance from TRL 5 to TRL 6 and deploy prototypes into hyperscaler data centres for cold storage trials.
- Target use case is cold data (60–80% of stored data): archival, compliance, and long-term preservation rather than low-latency hot storage.
- Current read/write speeds are slow (4 MBps write / 30 MBps read); the company aims for ~500 MBps to be competitive with tape archives.
- Initial read/write devices will be expensive (approx. $6,000 read, $30,000 write); SPhotonix plans a licensing/consortium model rather than manufacturing itself.
- Data retrieval today is lab-based; a field-deployable reader is expected within ~18 months.
Context and relevance
The announcement matters because data volumes keep exploding (IDC and industry forecasts), driven further by AI, and current storage media consume energy, fail more frequently and have limited lifespans. A stable, low-energy archival medium that can scale and survive for millennia (or longer) would change how organisations handle cold data, long-term archives and cultural preservation. Hyperscalers and enterprises under pressure to reduce operational and environmental costs are obvious early adopters.
Why should I read this?
Look — if you care about where we stash the digital detritus of civilisation, this is proper sci-fi made real. It promises archival density, insane longevity and a chance to stop dumping petabytes onto energy-hungry spinning rust when most of it never needs instant access. Read it if you work in storage, archives, compliance or run large-scale data platforms; otherwise, bookmark it for when your legal team starts demanding millennia-long retention.
Author style
Punchy — the piece flags a potentially industry-changing archival tech and pushes the urgency: data volumes are ballooning and current media won’t cut it. If the tech scales as claimed, it’s a strategic shift for data-centre architects and storage buyers. Worth digging into the detail.