Project Silica’s advances in glass storage technology

2026-02-18 · Source: Microsoft Research · Field: Technology & Digital — Cloud Computing & IT Infrastructure, Emerging Technologies & Innovation · Depth: Expert, short

Summary

Microsoft Research has published a breakthrough in *Nature* regarding glass-based data storage, extending Project Silica's femtosecond laser encoding technology from expensive fused silica to ordinary borosilicate glass. This innovation enables data preservation for at least 10,000 years, addressing critical commercialization barriers like cost and media availability. Key advancements include faster parallel writing, simplified data readers requiring only one camera instead of three, and easier manufacturing processes. The new phase voxel method significantly reduces complexity and cost by requiring only a single laser pulse for data encoding. This research also details optimized symbol encodings using machine learning and a nondestructive optical method for longevity testing.

Key takeaway

For AI Scientists developing long-term data archival solutions, this breakthrough in glass-based storage using borosilicate glass presents a compelling, durable alternative to traditional media. You should explore integrating this technology for immutable data preservation, especially for critical datasets requiring millennia-scale longevity. Consider the reduced cost and simplified reader/writer systems as significant advantages for future infrastructure planning.

Key insights

Glass-based data storage using borosilicate glass offers durable, cost-effective, and long-lived information preservation.

Principles

• Borosilicate glass is a viable, lower-cost storage medium.
• Single-pulse phase voxels simplify data encoding.
• Machine learning can optimize symbol encoding and mitigate interference.

Method

Data is stored in borosilicate glass using femtosecond lasers to create phase voxels with a single pulse. Parallel writing is achieved via multi-beam delivery and pre/post-heating models, with data read by a single camera.

In practice

• Utilize borosilicate glass for archival storage.
• Implement single-pulse phase voxel writing.
• Apply machine learning for error mitigation in storage systems.

Topics

• Glass Data Storage
• Project Silica
• Femtosecond Lasers
• Borosilicate Glass
• Archival Storage

Best for: AI Scientist, AI Researcher, Research Scientist, Data Engineer

Related on AIssential

• Scientists just found a way to store massive data using light in 3 dimensions — A new holographic storage method uses light's amplitude, phase, and polarization with AI for higher data density.
• Future AI chips could be built on glass — Glass substrates offer superior thermal stability and connection density for next-generation AI chips, enhancing performance and energy efficiency.
• Nvidia BlueField-4 STX adds a context memory layer to storage to close the agentic AI throughput gap — Nvidia's BlueField-4 STX architecture optimizes AI agent performance by accelerating KV cache access.
• Meta's Groundbreaking Glasses Initiative — AI is driving significant shifts in defense, cybersecurity, energy infrastructure, and consumer technology.
• The Memory Wall Is Real, Here Is the Door — Persistent DRAM shortages and rising costs necessitate hardware memory compression for AI product roadmaps.
• Liquid-cooled AI systems expose the limits of traditional storage architecture — Hybrid air/liquid cooling for AI infrastructure is inefficient and creates thermal liabilities for storage.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Microsoft Research.