In a world where magnetic tape degrades within decades and hard drives spin into obsolescence, Microsoft has quietly deployed a storage technology that could outlast civilizations. The company’s Project Silica, which encodes data into quartz glass using femtosecond lasers, has moved from research curiosity to operational reality — and the implications for cloud storage, archival science, and the economics of data preservation are profound.
Microsoft announced that its glass-based storage medium, developed under the Project Silica banner, is now being used internally to store what the company describes as critical archival data. The system writes data by firing ultrafast laser pulses into small squares of quartz glass, creating nanoscale three-dimensional structures called voxels within the material. Each voxel encodes data through changes in the polarization of light passing through it, and a single glass plate — roughly the size of a drink coaster and just two millimeters thick — can hold multiple terabytes of information. The data is then read back using a polarization microscope and machine learning algorithms that decode the complex optical patterns.
A Storage Medium That Could Outlast the Pyramids
The headline number — 10,000 years of durability — is not marketing hyperbole but rather a conservative engineering estimate. Quartz glass is chemically inert, resistant to electromagnetic interference, and unaffected by water, heat up to several hundred degrees Celsius, and virtually all forms of environmental degradation that plague conventional storage media. As Slashdot reported, Microsoft’s internal testing has subjected the glass plates to baking, boiling, microwaving, scouring with steel wool, and even demagnetization — none of which destroyed or degraded the stored data.
This stands in stark contrast to the fragility of current archival technologies. Magnetic tape, the workhorse of cold storage for decades, must be replaced every 10 to 30 years. Hard disk drives have similar lifespans and require climate-controlled environments. Even optical discs like Blu-ray, used by Facebook’s parent company Meta for cold storage, degrade over time. The cost of periodically migrating data from dying media to fresh media — a process the industry calls “data migration” — represents one of the largest hidden expenses in cloud computing. Microsoft’s glass storage, if it performs as promised at scale, could eliminate that recurring cost entirely.
How Femtosecond Lasers Write Permanent Data
The physics behind Project Silica are as elegant as they are complex. A femtosecond laser — one that fires pulses lasting just quadrillionths of a second — focuses energy into a tiny point within the glass, creating a voxel. Unlike traditional storage, which writes data on a surface, Project Silica writes throughout the entire volume of the glass, stacking dozens of layers of voxels at different depths. Each voxel can encode multiple bits of data, not just through its presence or absence but through the specific way it alters the polarization, retardance, and orientation of light passing through it. This multi-dimensional encoding dramatically increases storage density.
Reading the data requires a custom-built polarization microscope that captures images of the voxels layer by layer. Machine learning models then interpret the optical signatures to reconstruct the original data. Microsoft Research has published peer-reviewed papers on the system, and the company has been refining the read speeds — which initially lagged far behind conventional storage — to make the technology practical for large-scale deployment. The write process, while slower than writing to a hard drive, is a one-time operation for archival data, making speed less of a concern.
From the Warner Bros. Vault to Microsoft’s Own Data Centers
Project Silica first gained public attention in 2019, when Microsoft partnered with Warner Bros. to store a copy of the 1978 film “Superman” on a single piece of glass roughly 75 by 75 millimeters. That demonstration proved the concept but left open the question of whether the technology could scale beyond a single movie. Since then, Microsoft has continued developing the platform, and the company’s latest disclosures indicate that glass storage is now being used for real workloads within its Azure cloud infrastructure — specifically for data that must be retained for regulatory, legal, or historical reasons but is rarely accessed.
The economics of so-called “cold” and “archival” storage tiers are significant. According to industry estimates, more than 60% of all data stored in the cloud is rarely or never accessed after creation. Enterprises pay billions of dollars annually to store compliance records, medical images, scientific datasets, and government archives that must be kept but are almost never read. For this category of data, the cost of the storage medium itself is often dwarfed by the ongoing costs of power, cooling, and periodic media replacement. A write-once medium that requires no power to maintain and never degrades could fundamentally alter the cost structure of archival storage.
The Engineering Challenges That Remain
Despite the optimism, significant hurdles remain before glass storage can compete with tape and disk at hyperscale. Write speeds are still slow by data center standards. The custom reading hardware — the polarization microscopes and associated compute for machine learning decoding — is expensive and not yet mass-produced. And while a single glass plate can hold terabytes, the largest data centers manage exabytes, meaning millions of plates would need to be manufactured, indexed, and retrieved by robotic systems.
Microsoft has acknowledged these challenges and has been developing automated library systems — essentially robotic arms that can retrieve specific glass plates from storage racks and position them under readers. The company has drawn comparisons to existing tape library systems, which use similar robotic retrieval mechanisms. But the analogy only goes so far: tape libraries benefit from decades of manufacturing optimization and a mature supply chain. Glass storage has neither, at least not yet.
A Race Against Data Decay
The urgency behind projects like Silica is driven by a problem that the data storage industry has been grappling with for years: the sheer volume of data being created is growing far faster than the capacity to store it affordably and durably. According to IDC, the global datasphere is expected to exceed 180 zettabytes by 2025. Much of that data is ephemeral — social media posts, streaming buffers, temporary caches — but a growing share consists of records that organizations are legally or ethically obligated to preserve for decades or longer.
The healthcare industry, for example, is required in many jurisdictions to retain patient imaging data for the life of the patient plus several years. Financial institutions must keep transaction records for regulatory periods that can stretch to 25 years or more. Government agencies archive everything from census data to satellite imagery. For these use cases, the prospect of a storage medium that simply does not decay is enormously attractive — provided the cost per terabyte can be brought down to competitive levels.
Competition and the Broader Archival Storage Market
Microsoft is not the only organization exploring long-duration storage. Researchers at the University of Southampton have developed a similar technology using nanostructured glass, which they call “5D optical data storage,” capable of theoretically lasting billions of years. The French company Millenniata previously marketed M-DISC, a stone-like optical disc rated for 1,000 years, though the company’s commercial success was limited. DNA-based data storage, pursued by companies like Twist Bioscience and research groups at Harvard and the University of Washington, offers extraordinary density but remains prohibitively expensive and slow for practical deployment.
What distinguishes Microsoft’s effort is the backing of one of the world’s largest cloud infrastructure providers. Unlike academic projects or startups, Microsoft has the financial resources, the engineering talent, and — critically — the immediate use case within its own Azure data centers to push glass storage from prototype to production. The company’s willingness to deploy the technology internally, rather than waiting for external customers, signals a level of confidence that few competitors can match.
What 10,000-Year Storage Means for the Future of Information
If Project Silica fulfills its promise, the implications extend well beyond corporate data centers. Libraries, museums, and national archives have long struggled with the problem of digital preservation — the ironic reality that a clay tablet from ancient Mesopotamia has proven more durable than a CD-ROM from the 1990s. A commercially viable, millennia-scale storage medium could provide institutions with a way to preserve cultural heritage, scientific records, and historical documents without the constant anxiety of format obsolescence and media decay.
Microsoft has reportedly been in discussions with cultural preservation organizations about potential applications, though the company has not disclosed specific partnerships beyond the earlier Warner Bros. demonstration. The technology also raises philosophical questions about what data deserves to be preserved for millennia and who gets to make that determination — questions that the storage industry has traditionally left to its customers but that take on new weight when the time horizon stretches to 10,000 years.
For now, Microsoft’s glass storage remains a niche technology deployed for specific archival workloads. But the trajectory is clear: as write speeds improve, manufacturing costs decline, and automated retrieval systems mature, glass could become a standard tier in the cloud storage hierarchy — sitting below hot and warm storage as the ultimate cold archive. The company that once built its empire on software is now, quite literally, etching its future in glass.