Bridging data
and intelligence
We take a radically different path to photonic integration—our cantilever waveguide coupling fuses III-V performance with silicon scalability in a way no one else can. By combining breakthrough technology with flawless execution, we’re building the optical engines that will drive the next era of AI and telecom.
AI datacenters are hitting a wall
GPUs keep getting faster, but they sit idle — waiting for data. The problem isn’t compute anymore. It’s communication.
Optical links could fix this. They’re faster, they reach farther. But today’s photonics are too costly and too complex to keep up with AI.
Below: Utilization of GPU vs. data-feed rate.
Fundamentally
different
Photon Bridge removes the interconnect wall – making interconnects affordable, dense, and scalable for AI fabrics.
Our optical engines are:
Cost Optimized
— Silicon scale
— Backend/OSAT integration
Power-efficient
— Up to <1pJ/bit optical
— Thermally stable
Performance optimized
— III-V speed; up to >200Gbps/lane
— Multiple colors
— Optical I/O density 50um pitch; self aligned
Cost Optimized
— Silicon scale
— Backend/OSAT integration
Power-efficient
— Up to <1pJ/bit optical
— Thermally stable
Performance optimized
— III-V speed; up to >200Gbps/lane
— Multiple colors
— Optical I/O density 50um pitch; self aligned
How we make
things simpler
Photon Bridges make photonics click. We combine the best building blocks (III-V semiconductors to create, manipulate, and detect light) with the strongest foundation: thick silicon waveguides to guide it.
Our patented solution shifts integration to packaging houses (OSATs), for the first time delivering scalable, affordable, high performance photonics.
Unlimited optical IOs
Any material combination
Low coupling loss
Unlimited optical IOs
Any material combination
Low coupling loss
How we manufacture
Our OSAT compatible pick and place process fuses only burn-in tested III-V dies with low loss silicon waveguides, achieving <1dB coupling loss and high yields, delivering unmatched density, reliability and scalability for AI datacenters and telecom.
Flip chip assembly
Applications
The race to deliver the next generation of AI applications has begun. Our technology is uniquely designed to unlock the cost and performance breakthroughs needed to make these solutions succeed at scale.
Photonic Fabrics
— 5x lower cost
— 2x lower power
— Improved thermal headroom
LPO/CPO Chiplets
— Copper2Fiber Module
— 1.6Tbps @ <2 pJ/bit
— Thermally Stable
Photonic Fabrics
— 5x lower cost
— 2x lower power
— Improved thermal headroom
LPO/CPO Chiplets
— Copper2Fiber Module
— 1.6Tbps @ <2 pJ/bit
— Thermally Stable
