Archival storage positions great deals of difficulties. We desire media that is exceptionally thick and steady for centuries or more, and, preferably, does not take in any energy when not being accessed. Great deals of concepts have actually drifted around– even DNA has actually been thought about– however among the easiest is to engrave information into glass. Numerous kinds of glass are extremely physically and chemically steady, and it’s fairly simple to engrave things into it.

There’s been a great deal of initial work showing various elements of a glass-based storage system. In Wednesday’s problem of Nature, Microsoft Research revealed Project Silica, a working presentation of a system that can check out and compose information into little pieces of glass with a density of over a Gigabit per cubic millimeter.

Composing on glass

We tend to think about glass as delicate, vulnerable to shattering, and efficient in streaming downward over centuries, although the last claim is a misconception. Glass is a classification of product, and a range of chemicals can form glasses. With the best beginning chemical, it’s possible to make a glass that is, as the scientists put it, “thermally and chemically steady and is resistant to wetness ingress, temperature level variations and electro-magnetic disturbance.” While it would still require to be managed in a manner to decrease damage, glass offers the sort of stability we ‘d desire for long-lasting storage.

Putting information into glass is as basic as engraving it. That’s been one of the obstacles, as etching is generally a sluggish procedure. The advancement of femtosecond lasers– lasers that give off pulses that just last 10^-15 seconds and can give off countless them per second– can substantially lower compose times and enable engraving to be concentrated on a really little location, increasing possible information density.

To check out the information back, there are numerous alternatives. We’ve currently had fantastic success utilizing lasers to check out information from optical disks, albeit gradually. Anything that can choose up the little functions engraved into the glass might possibly work.

With the above factors to consider in mind, whatever remained in put on a theoretical level for Project Silica. The huge concern is how to put them together into a practical system. Microsoft chose that, simply to be careful, it would address that concern two times.

A real-world system

The distinction in between these 2 responses boils down to how a private system of information (called a voxel) is composed to the glass. One kind of voxel they attempted was based upon birefringence, where refraction of photons depends upon their polarization. It’s possible to engrave voxels into glass to produce birefringence utilizing polarized laser light, producing functions smaller sized than the diffraction limitation. In practice, this included utilizing one laser pulse to develop an oval-shaped space, followed by a 2nd, polarized pulse to cause birefringence. The identity of a voxel is based upon the orientation of the oval; considering that we can solve numerous orientations, it’s possible to conserve more than one bit in each voxel.

The alternative method includes altering the magnitude of refractive impacts by differing the quantity of energy in the laser pulse. Once again, it’s possible to determine more than 2 states in these voxels, permitting several information bits to be kept in each voxel.

Borosilicate glass uses severe stability; Microsoft’s sped up aging experiments recommend the information would be steady for over 10,000 years at space temperature level. That led Microsoft to state, “Our outcomes show that Silica might end up being the archival storage option for the digital age.”

That might be overselling it simply a bit. The Square Kilometer Array telescope, for instance, is anticipated to require to archive 700 petabytes of information each year. That would imply over 140,000 glass pieces would be required to keep the information from this one telescope. Even presuming that the compose speed might be increased by including considerably more lasers, you ‘d require over 600 Silica devices running in parallel to maintain. And the Square Kilometer Array is far from the only job creating massive quantities of information.

That stated, there are some functions that make Silica an excellent match for this sort of thing, most especially the total lack of energy required to maintain the information, and the truth that it can be recovered quickly if required (a sharp contrast to the days required to obtain info from DNA, for instance). Plus, I’m undoubtedly drawn to a system with a storage medium that appears like something right out of sci-fi.

Nature, 2026. DOI: 10.1038/ s41586-025-10042-w (About DOIs).