GlobalFoundries announced on May 4, 2026, the SCALE (Silicon photonics Co-packaged Advanced Light Engine) optical module solution for co-packaged optics (CPO), becoming the first chipmaker to deliver a platform that exceeds the Optical Compute Interconnect Multi-Source Agreement (OCI MSA) standard for AI scale-up architectures. SCALE uses GF's silicon photonics technology with both coarse and dense wavelength-division multiplexing (CWDM and DWDM) for bi-directional data transmission over optical fiber, achieving significant bandwidth density and system scalability versus traditional copper interconnects. GF has demonstrated 8λ and 16λ bi-directional DWDM capability on the platform.
The push addresses a fundamental constraint: as AI hyperscalers build denser GPU clusters, power and cooling limits force a shift from copper to optical interconnects inside data centers. CPO reduces heat, cuts power draw per unit bandwidth, and increases density—critical as companies scale to multi-rack AI deployments. SCALE integrates 50Gbps and 100Gbps micro-ring modulators, through-silicon vias (TSVs), and copper pad pitches from 110μm down to sub-45μm for 2.5D/3D stacking, enabling customers to transition from design to volume production.
For architects, GF's SCALE arrival signals that CPO is moving from research to production-ready status. Because optical links are now standardized via OCI MSA, system integrators (Broadcom, others) and data center operators can build components with confidence in interoperability. GF stock rose 4% on the announcement, reflecting investor belief that specialized photonics is a growth driver as NVIDIA and AMD dominate compute chips. The next milestone is real-world hyperscaler design wins; industry observers should track first CPO production deployments in Q4 2026–Q1 2027.