Lumentum Showcases Optical Scale-Up Demo at OFC Using VCSELs

Why It Matters

The platform addresses AI data‑center bandwidth and power bottlenecks while diversifying the photonics supply chain, accelerating adoption of optical scale‑up solutions.

Key Takeaways

• Lumentum demoed 1060 nm VCSEL optical interconnect at OFC
• Solution co‑packs VCSEL array with ASIC for rack‑level AI
• Fan‑out wafer package enables standard semiconductor flow
• Operates above 150 °C, offering thermal resilience
• Provides supply‑chain alternative to silicon photonics

Pulse Analysis

The relentless growth of generative AI workloads is stretching traditional electrical interconnects, prompting data‑center architects to explore optical alternatives that can deliver higher bandwidth density with lower power. Lumentum’s 1060 nm VCSEL platform arrives at a pivotal moment, offering a wavelength that balances speed and thermal performance better than the legacy 850 nm datacom lasers. By integrating the VCSEL array directly with a host ASIC, the solution reduces the electrical‑to‑optical conversion distance, mitigating signal loss and enabling "slow and wide" protocols such as UCIe and PCIe to scale efficiently.

Technically, the platform leverages a fan‑out wafer‑level package that aligns with existing semiconductor manufacturing lines, shortening time‑to‑market and lowering cost. The two‑dimensional monolithic array incorporates backside lenses for precise alignment, while channel sparing adds redundancy, enhancing system reliability. Notably, the devices sustain temperatures exceeding 150 °C, a critical advantage for densely packed AI racks where thermal budgets are tight. Compared with silicon‑photonic transceivers, the 1060 nm VCSELs provide faster modulation speeds and longer‑term reliability, backed by Lumentum’s track record of shipping over 10 billion emitters for 3D sensing applications.

From a market perspective, Lumentum’s offering diversifies the optical interconnect ecosystem, reducing reliance on silicon‑photonic foundries and InP laser suppliers. This supply‑chain flexibility could accelerate adoption among hyperscale cloud providers seeking to mitigate component shortages. As AI accelerators continue to proliferate, the demand for scalable, high‑temperature‑tolerant optical links will rise, positioning Lumentum’s VCSEL scale‑up platform as a compelling alternative in the race to meet next‑generation data‑center bandwidth needs.