DESIGN INTELLIGENCEJuly 25, 2026·Mary · DEPIX Design Intelligence

The Computer Had to Disappear

Something the wearable industry chased for over a decade is finally happening: people are buying computers for their faces. Global smart-glasses shipments are set to jump about 85% in 2026, pushing past 15 million units, after surging 167% year-over-year in the first quarter. Meta and EssilorLuxottica alone sold more than seven million pairs in 2025, tripled their sales again, and now hold roughly 69% of the market, with Meta reportedly weighing 20 million units of annual capacity.

The obvious explanation is that the technology finally got good enough. It is the wrong explanation. The single most important design decision behind this moment is not the micro-LED display, the camera or the on-board AI. It is a decision about what the object is allowed to look like — and it was made, deliberately, before any of the electronics.

Rewind to 2013. Google Glass had a capable little computer, a camera and a heads-up display, and it failed so completely it became a cultural punchline. Not because the tech did not work, but because it announced itself. It looked like a device, it read as surveillance, and wearing one marked you as someone willing to put a machine between your face and everyone else's. The product's real problem was social, not technical — it asked the world to accept a new object, and the world declined.

The 2026 winners inverted the priority. Instead of building a computer and hoping people would wear it, they started from the constraint that the thing had to be socially invisible — a pair of glasses people already wanted on their face — and made the silicon conform to that. The dominant design pattern of the year is restraint: frames that disappear into an existing eyewear collection. The clearest tell is who the tech companies partnered with. Meta went to EssilorLuxottica, the world's largest eyewear maker, and shipped Ray-Bans. Google and Samsung previewed Android XR glasses co-designed with Warby Parker and Gentle Monster. Even the flagship Ray-Ban Display hides a screen behind a frame that reads, first, as a Ray-Ban.

That partnering is the concept-phase decision made visible. A silicon company outsourcing the frame to an eyewear house is an admission that the choice which determines adoption — what this looks like on a human face — matters more than anything the company could build itself. It is why roughly 91% of 2026 shipments are the restrained, audio-first, display-light kind, not maximal AR visors. The market voted for the version that disappears.

This is the part that generalizes past eyewear. "Looks normal" is not a finish you can apply at the end; it is a founding constraint that everything downstream has to obey. The battery, the optics, the antenna, the hinge, the thermals — all of them get harder when the non-negotiable is "indistinguishable from a nice pair of glasses at three feet." Google Glass had an easier engineering brief and a fatal concept-phase brief. The Ray-Ban route has a brutal engineering brief and a winning concept-phase one. You cannot swap between them late; the whole architecture is different depending on which decision came first.

There is a quiet lesson here for every ambitious product, not just the face-worn ones. When a technology is genuinely new, the instinct is to lead with the capability — to show the world the computer. But the objects that actually get adopted are usually the ones that decided, at the very start, to hide it: to be a familiar, desirable thing first and a breakthrough second. The $299 Ray-Ban that looks like sunglasses will put more computing on more faces than any visor that looks like the future.

The smart glasses that won did not win because the computer got better. They won because someone decided, at the concept phase, that the computer had to disappear — and then made everything else expensive and difficult in service of that one idea.

Sources: