New Technology Spotlight: From lab to rack: the real work of CPO
Download our eBook on co-packaged optics testing, deployment timelines, and standards, sponsored by VIAVI Solutions
Co-packaged optics (CPO) promises lower power, lower latency, and higher density for AI clusters. Moving from early devices to production at scale brings practical problems: testing optical devices to semiconductor standards, deciding when CPO makes sense compared with pluggables, and setting standards fast enough to keep up with AI hardware cycles.
This Technology Spotlight looks at the practical work behind CPO. Experts explain why testing is moving earlier in manufacturing, where CPO is in relation to volume deployment, and how standards work is speeding up to keep pace with AI.
What you’ll learn from reading “From lab to rack: the real work of CPO”
- Why optical testing has to reach semiconductor-fab levels of automation, uptime, and density, including fitting more functional testing into half the floor space.
- Why connector management is one of the most underestimated parts of CPO deployment, when a single front panel can carry 300 MMC connectors.
- Why CPO will start with scale-up AI clusters before wider deployment, and why pluggables, linear optics, and silicon photonics will continue alongside it.
- How standards work is speeding up, from the 3.2T module implementation agreement to expanded-beam connectors and 12.8T near-package optics.
Who should read “From lab to rack: the real work of CPO”?
Network architects, data centre planners, optical component and system engineers, test and manufacturing teams, and procurement leads will find practical detail on where CPO is ready today and where it isn’t. Anyone planning AI interconnect for the next three to five years will benefit from understanding how testing, timelines, and standards will decide which architectures can scale.
Why download this Technology Spotlight?
The move to CPO depends as much on manufacturing, testing, and standards as on the optics themselves. This Spotlight shows where the risks lie, what realistic timelines look like, and what to ask suppliers before you commit to an architecture.
By downloading this Technology Spotlight, you will discover:
- Testing at semiconductor scale: How functional testing is moving earlier in production, and why equipment reliability matters as much as measurement precision.
- Realistic deployment timelines: Why CPO will start with scale-up AI clusters, and how it compares with DSP-based pluggables, linear pluggable optics, and silicon photonics, which already serves an estimated 30–40% of the 800G market.
- Reliability in AI clusters: Why a single link failure can stall a training run, and how telemetry and self-healing links help keep failures from stopping the job.
Standards on AI timelines: How standards processes can keep pace with 18-month hardware cycles, and why moving to common connector standards matters for cost and competition.
