Gaetano@crux_capital_
CEO of Kopin Corporation just dropped a bomb in this interview $KOPN
My two worlds are colliding. Data center interconnect, and microLEDs that have traditionally lived in defense.
He called it Project Shasta.
This post is going to be a bit technically dense, but stick with me and ask questions as needed!
Here is the concept as he described it. A microLED is usually a display. A ton of tiny pixels that emit light to form an image. Shasta is the same hardware idea, but you use the pixels to transmit data, not pictures. In other words, light becomes the link.
If that sounds familiar, it should. This is the same direction the interconnect stack has been moving that I cover exhaustively on this page. We are trying to move bandwidth with less power, less heat, and less loss. That is why optical keeps moving closer to the compute.
What really interested me is the framing. He is not talking about a normal optical transceiver. He is talking about microLED to microLED communication through air, at very high data rates, as a possible alternative to the traditional path of copper plus retimers, or fiber plus optics.
To be clear, he also said it is not ready for prime time. So I am not modeling this as a near term revenue driver. I am treating it as a real option on a future architecture shift.
Lets look at the implications incase this becomes real...
First, it is a direct attack on one of the hardest parts of modern compute. Moving data is becoming more expensive than computing on it. Every step of the journey adds power draw, heat, latency, and packaging complexity. The idea of turning the link into “light over distance” is an attempt to reduce how much hardware you need in between two endpoints.
Second, it is a shot at the power bottleneck. In dense compute, the system spends a lot of energy just pushing bits across short distances. If you can move information with light more efficiently than copper, you reduce the watts burned on the interconnect and you reduce the cooling burden that comes with it. That feeds back into higher compute density per rack.
Third, it changes where value lives. The industry has spent decades building an entire ecosystem around modules, connectors, fiber, cages, retimers, and board level plumbing. A credible optical “over the air” link would push value toward whoever controls the emitter, the receiver, and the integration know how. It becomes less about swapping parts and more about designing the platform.
Fourth, it would pull optics inside the box in a new way. We typically picture optics as something that leaves the chassis through fiber. Shasta is implying an internal optical fabric where components talk directly using light without a physical cable between them. That is a different mental model. It is closer to optical wireless inside a system than it is to what we call a transceiver today.
Fifth, it is a very high bar technically, which is why it is interesting. If you are sending data with microLED pixels, you need the light source to switch extremely fast. You need the receiver to interpret that signal cleanly. You need optics that collimate and aim the light. You need alignment that stays stable despite vibration, thermal expansion, and time. And you need it to work at scale, not as a lab demo. If someone clears those gates then it is a real architecture unlock.
To close, if compute keeps scaling, the interconnect cannot stay a side issue. It becomes the system. Any credible path that moves bits with less power and less complexity is worth paying attention to and will draw investor curiosity, even if it is early.
I'm pretty mind blown right now. Will absolutely stay on top of this
Video linked in comments