Molex Acquires Teramount to Scale Fiber-to-Chip Connectivity for Co-Packaged Optics

Fiber doesn’t get headlines. It gets overlooked, buried under GPUs and model hype, treated like plumbing until the pressure builds and the whole system starts coughing. Then everyone remembers what actually moves the signal. Teramount did not chase noise. They went straight for the constraint nobody outside the room could see, the messy, stubborn handshake between fiber and chip. While the market obsessed over compute, Hesham Taha and Avi Israel were down in the trenches solving the part that actually lets all that power talk. Not sexy on the surface. Absolutely lethal in impact.

Now Molex is stepping in with a roughly $430M acquisition, and if you understand where co packaged optics is heading, you can feel the subtext without anyone spelling it out. This is infrastructure gravity shifting in real time.

Teramount, founded in 2013 out of Jerusalem, built TeraVERSE around a deceptively simple idea: make fiber to chip connectivity detachable, passively aligned, and manufacturable at scale. No drama in assembly. No delicate dance every time you need performance. Just clean, repeatable physics doing its job. Turns out, when you remove friction at that layer, everything upstream starts to breathe easier.

That kind of thinking does not happen by accident. Two applied physics PhDs from Hebrew University, sharpened at Nanonics Imaging, deciding that the real opportunity was not another chip, but the bridge between them. Most people build louder. They built smarter.

The investor bench saw it early. Grove Ventures set the tone with the initial $8M. Then Koch Disruptive Technologies led a $50M Series A, with AMD Ventures, Hitachi Ventures, Samsung Catalyst Fund, and Wistron all pulling up chairs. That is not a casual cap table. That is a group that understands where bandwidth, heat, and density collide.

And here is the quiet lesson wrapped inside the headline. Teramount did not need to be everything. They needed to be essential. They picked a problem so fundamental that once solved, it becomes very hard to replace. When your tech sits at the junction of AI infrastructure, data centers, and high performance computing, you are not a feature. You are a dependency.

Molex is not buying a product. They are buying a lever into the future of optical interconnects, into a world where scaling compute without fixing connectivity is like putting a jet engine on a bicycle. Credit to Hesham Taha and Avi Israel for playing the long game, and to the teams at Molex and Koch Disruptive Technologies for recognizing signal before it gets obvious.