Microsoft's Project Silica Stores 4.8 Terabytes in a Glass Slab That Could Last 10,000 Years

Overview

Microsoft Research’s Project Silica team published findings in Nature on 18 February showing that ordinary borosilicate glass — the same material used in Pyrex cookware and oven doors — can store 4.84 terabytes of data per slab and withstand at least 10,000 years at room temperature without any power or active maintenance. The results mark a significant advance in the project’s eight-year development by extending the technology from expensive fused silica to a far cheaper and more widely available material.

How It Works

The system uses femtosecond lasers, which fire pulses lasting quadrillionths of a second, to encode data as “phase voxels” — three-dimensional structures within the glass that alter the optical properties of the material. Each voxel can store multiple bits by encoding information in varying phase properties, according to Gizmodo’s coverage. Reading data back relies on phase contrast microscopy paired with a convolutional neural network that interprets the encoded voxel patterns.

A 12-centimetre-wide, 2-millimetre-thick glass square achieves a density exceeding one gigabit per cubic millimetre, yielding 4.84 TB per slab, as reported by Slashdot. The reading process generates insufficient power to alter the glass, making the medium tamper-resistant and immune to accidental overwriting — what Microsoft describes as a true physical air gap.

The Write Speed Problem

The technology’s primary current limitation is throughput. Four operating lasers achieve a combined write rate of approximately 66 megabits per second, meaning a full 4.84 TB slab requires around 150 hours to fill, Slashdot noted. The researchers believe adding more parallel laser beams could substantially improve throughput, but the current speed makes the technology most practical for write-once archival scenarios rather than active storage.

Why Ordinary Glass Matters

Project Silica’s original prototypes used fused silica, a high-purity material that is expensive and difficult to manufacture at scale. The shift to borosilicate glass — commercially produced in massive quantities for consumer cookware and laboratory equipment — substantially lowers the potential cost of the medium and opens a path to industrial-scale manufacturing. The Microsoft Research blog frames this as a key step toward making glass storage viable beyond research settings.

Addressing the Digital Dark Age

The motivation for Project Silica is a straightforward archival problem. Magnetic tape, the dominant medium for long-term enterprise data storage, degrades within decades and requires climate-controlled facilities and periodic re-encoding. Hard drives and optical discs fare similarly. As a result, digital records face what archivists call the “Digital Dark Age” — the paradox that modern societies generate vast amounts of data yet have no storage medium that reliably preserves it across centuries.

According to the Nature coverage, an optical physicist at the University of Southampton described the research as addressing the fact that “our current records are kept on fragile magnetic platters that are constantly decaying.” Project Silica consulted with the National Archives, museums, and entertainment industry stakeholders during development.

What We Don’t Know

The Nature paper demonstrates durability through accelerated aging tests, but real-world preservation across millennia remains unverified. Cost per terabyte at production scale has not been disclosed, and format obsolescence — whether future systems will be able to read phase voxels stored today — poses a challenge the research does not directly address. The path from laboratory demonstration to commercially deployed archival systems typically involves several additional engineering and manufacturing milestones.