Pyrex for the archive

For ten years Project Silica has been etching data into glass with a femtosecond laser, betting that a slab of silica could outlast any tape or disk. The catch was the glass: it had to be optical-grade fused silica, expensive and finicky. A new paper in Nature, peer-reviewed rather than demoed, shows the same trick now works in plain borosilicate — the same family as Pyrex cookware.

The main design decision for future Silica systems — whether to deploy birefringent or phase voxels — has not been made.

Getting there meant a new way to write. Instead of the old two-pulse method that carves a tiny light-bending structure and needs three or four cameras to read back, the team used a single laser pulse to nudge the glass's refractive index, and wrote many tracks in parallel. One camera reads it. The payoff is manufacturability: cheap glass, simpler hardware.

It costs capacity. On a DVD-sized 2mm wafer the cheap glass holds about 2 terabytes, less than half the 4.84 TB the team still gets from pricey fused silica. So the headline figure and the headline material don't sit on the same disc — the breakthrough is making the glass ordinary, not making it bigger.

The pitch was never density anyway. A modern tape cartridge holds more and writes roughly ten times faster. What glass offers is forgetting-proof storage: write once, then leave it on a shelf for a projected 10,000 years with no copying, no migration, no rot — the maintenance treadmill that eats every archive of data meant to be kept forever. That number is inferred from accelerated aging, not observed, but it is the whole reason to care.