Marvell Wins EDGE Award for COLORZ 1600 2nm 1.6T ZR/ZR+ Pluggable Optical Platform - Marvell Technology

Marvell's push of coherent optical DSPs into the 2nm node is not a product iteration—it is a structural re-slicing of the optical interconnect value chain. Technically, 2nm compresses the coherent transceiver power envelope into the range where it can compete with direct-detection at 1.6T. This makes ZR/ZR+ viable for intra-cluster AI interconnect, a domain previously locked by 800G DR/FR pluggables. Upstream, this locks TSMC (Taiwan, China) 2nm capacity for photonics-adjacent silicon—a resource historically monopolized by pure logic. Downstream, hyperscalers will see coherent optics cannibalize existing 800G pluggable deployments. On compliance, single-source 2nm dependency is a structural vulnerability. Under US export controls on advanced nodes, any geopolitical friction directly threatens Marvell's roadmap. ZR/ZR+ coherent tech occupies a gray zone between civilian telecom and high-assurance communications; regulatory arbitrage windows are closing fast. Competitively, Broadcom will likely accelerate its 1.6T coherent roadmap to defend its switch-chip-plus-optical-DSP bundling advantage. Ciena and Infinera face sustained disintermediation pressure as hyperscalers internalize DCI. Outlook: within 12–24 months, 1.6T ZR becomes the de facto standard for AI cluster DCI. Marvell's 2nm pluggable bet is a defensive hedge against CPO at 3.2T—buying time with power efficiency.