I recently attended a one-day seminar called “Seeing is Believing” in November, hosted by the Bay Area SID chapter. I am now getting a sense that the fledgling AR glasses industry is nearing an inflection point. Driven by the success of the Meta Ray-Ban AR glasses release, there now seems to be a heartfelt urgency in developing the tools, technology and supply chain to enable a volume consumer market. Certainly, the major AR system integrators have always stressed the need to develop a consumer-capable mass production supply chain, but it feels like they are taking much more concrete steps to really enable it now.

The keynote presentation from Meta VP XDO Jason Hartlove was particularly compelling. Meta sold 50K of the Meta Ray-Ban glasses on day one and now has a 50K backlog of orders. Given this success, many of the big players are now likely to rethink their roadmaps and accelerate plans to bring AR products to market. AI and the maturing of the technologies in the AR glasses means the “AR is just around the corner,” thinks Hartlove.

But what is the timeline for the various display engines competing for dominance in AR glasses. At the recent Frontiers in Optics Laser Science conference in October, Meta heled crate a new work group within the AR Alliance to focus on laser-based AR engines. Organizer Barry Silverstein showed a chart that described the adoption of LCOS, microLED and laser-based solutions for AR glasses, but he gave no time scale in his graphic.  At the Seeing is Believing seminar, Hartlove added a timeline to this graphic. In his view, LCOS may be peaking soon with microLED peaking around 2034 and laser-based solutions not peaking until maybe 2043. If he is right, we are going to be talking about AR glasses for quite a long time.

But Hartlove was also quite candid in documenting some of the issues in this roadmap. Not only are we facing performance and efficiency challenges, but there are big issues in creating a supply chain for AR glasses that can produce millions of devices a year.

For example, waveguides are fabricated on 200 or 300mm wafers in semiconductor-based machines. Hartlove then did a little math. To make 10 million dual-display AR glasses, he assumed 20 waveguides can be patterned on the substrate with future goal of 90% yield. That will require 1.1M 300mm wafer starts per year – which is a lot of capacity. If the goal is 100M waveguides, you need 11M wafer starts. For comparison, the entire capacity of TSMC is 17M wafer starts per year, so we are talking about a staggering amount of investment. And this does not address the performance or cost challenges.

Hartlove sees similar challenges up and down the supply chain for which he does not have the answers. His advice is to solve the hard problems now and move fast. Once AR reaches scale, it will be hard to displace the incumbents.