← Scout
OPTICAL
Sector teardown11 min read

Optical · CPO

The optical interconnect bottleneck: EML, CPO & the AI backbone

Scout Research · Jun 24, 2026 · 11 min read

The short version

  • 1Every GPU added to an AI cluster pulls in more optical interconnect. The hard parts don't scale at the same rate.
  • 2The binding constraint is the EML laser — not the GPU, not the GaN. A handful of firms can make it, and NVIDIA has largely bought out their capacity.
  • 3Three compound semiconductors gate everything: InP (the optical signal), GaN and SiC (the power to run it).
  • 4CPO — putting the optics on the same chip as the processor — is the architecture that breaks the copper wall. It's where the next decade of margin sits.
The chain

Six stages, upstream to AI

Data moves from raw substrates to a finished AI cluster through six stages. Value — and scarcity — concentrate at the laser and the packaging steps, not the ends.

01
Materials
02
IC design & foundry
03
Optical components
04
Module assembly
05
Packaging & test
06
AI infrastructure
The chokepoint

The real bottleneck is the EML laser

Two laser types feed AI optics. They could not be more different in how scarce they are. The EML (electro-absorption modulated laser) is the physical-layer wall for 800G and 1.6T modules. The VCSEL is plentiful and is not the constraint.

 EML — the hard bottleneckVCSEL — not the bottleneck
Where it's usedLong-haul, single-mode · 800G / 1.6TShort-reach, multi-mode
Why it's hardCW laser + modulator grown monolithically on InP — brutal yieldsMade on GaAs, tested at wafer scale — high yield, low cost
Who can make it~5 firms: Broadcom, Lumentum, Coherent, Mitsubishi, SumitomoMany, including consumer-sensor makers
Supply state~30% gap even after expansion; NVIDIA pre-bought much of itHealthy; single-channel 200G already shipping

The shortage runs upstream, too: EML output is throttled by an InP substrate shortage, where AXT's backlog is at record highs. Non-NVIDIA buyers are quoted lead times into 2027.

Outside of HBM memory and CoWoS packaging, the EML is the most profitable — and most supply-constrained — hidden throat in the entire AI infrastructure chain.
The materials

Three compound semiconductors gate it all

Silicon can't make an efficient laser. The optical era runs on three compound semiconductors instead — one for the signal, two for the power.

InP
Indium phosphide
The optical signal path. Only InP makes efficient lasers and detectors at telecom wavelengths.
GaN
Gallium nitride
High-frequency, high-efficiency power delivery and RF drive for modules, switches and servers.
SiC
Silicon carbide
High-voltage, high-power backbone for data-center and grid-side power.

One line: InP carries the light; GaN and SiC carry the power. They're separate chains that converge at the AI cluster.

The system constraint

Grid-to-chip: the power wall

The industry's own framing has shifted. The bottleneck isn't only the GPU — it's the whole system: power, cooling, supply chain. An AI factory is a power-constrained machine. On the grid-to-chip path, GaN and SiC become the system-level constraint — substrate supply, packaging and thermal limits, and foundry process all bind at once.

This doesn't contradict the optical story; it's a different axis. On the signal path, the bottleneck is EML + InP. On power and cooling, it's GaN/SiC. Same AI map, two different chokepoints.

The architecture

What CPO actually is

Three terms get used interchangeably. They're actually a hierarchy — a science, a hardware architecture, and a global system.

 What it isWhere it lives
PhotonicsThe science of engineering lightResearch, materials, components
CPOA manufacturing architectureInside one chip / server box
Optical networkA large-scale comms systemAcross buildings, cities, oceans

Traditionally a processor sends an electrical signal across a copper board to a separate front-panel transceiver that converts it to light. At 200 Gbps and above, copper means heat, power draw, and signal loss. Co-packaged optics (CPO) integrates the transceiver directly onto the silicon next to the processor — data leaves as light immediately, cutting power use by up to ~50%.

That's why the giants are committing billions to the shift: NVIDIA is expanding optical-connection capacity with Corning and investing in component leaders like Coherent. Moving data with photons instead of electrons is the only way to keep global compute from grinding to a halt.

The map

Who owns the chokepoints

Strip the full supplier table down to the names that actually gate each scarce step:

EML laserthe hard bottleneck
AVGO · LITE · COHR · Mitsubishi · Sumitomo
InP substrateupstream of EML
AXTI · COHR · JX Metals · Sumitomo
Photonic / electronic ICthe brains
NVDA · MRVL · AVGO · LITE · COHR
Connector + fiberthe plumbing
GLW · APH · Fujikura · Sumitomo · YOFC
Module assemblyintegration
COHR · LITE · CSCO · MRVL · FN
Power · SiC / GaNgrid-to-chip
WOLF · ON · NVTS · QRVO · SWKS
Bottom line

Where the basket points

The thesis is asymmetry: demand for AI compute pulls hardest on the steps the market under-owns — the EML laser, its InP feedstock, and the CPO integrators that replace copper with light. That's the spine of the Photonics basket.

InformsPhotonics Is the FutureView Photonics Is the Future →

Sources

  • · Company filings & investor materials: Broadcom, Lumentum, Coherent, AXT, NVIDIA, Corning
  • · Industry capacity & lead-time commentary (800G / 1.6T optical modules, InP substrate)

Scout Research is for information only and is not investment advice.