AI and Data Center Networking



AI & Data Center Networking — Newlink

Solutions

AI & Data Center Networking

The optical layer now sits on the AI critical path. From GPU scale-up inside the rack to leaf-and-spine scale-out across the data hall, our 400G, 800G and 1.6T transceivers move data where your accelerators need it.

Generative AI is driving data center bandwidth higher and faster than any workload before it. Inside an AI cluster, thousands of optical links tie together GPUs, switches and storage — and every one of them has to deliver more throughput per watt than the generation before. Newlink builds the core technology platforms behind a complete 400G, 800G and 1.6T portfolio for AI and data center networking, so operators can deploy a coherent family today and scale compute and network together — without re-architecting the physical layer at every speed jump.

200G-per-lane ready

A clean electrical and optical path from 800G to 1.6T and on toward 3.2T — without redesigning the physical layer at every speed jump.

Lower power per bit

Efficiency engineered into every link, so each port carries more data for less of your facility’s power budget.

Broad interoperability

Validated across leading switch and host platforms for predictable, multi-vendor deployments.

Every footprint & reach

OSFP, OSFP224 and QSFP-DD in DR, FR and LR variants — spanning in-rack, cross-row and campus or metro DCI.

1.6T

Next-generation AI spine

200G-per-lane flagship optics for the highest-bandwidth scale-out and scale-up fabrics.

1.6T OSFP224 transceiver
1.6T
OSFP224

Flagship 1.6 Tbps module on 200G-per-lane electrical and optical I/O for next-gen AI spine and scale-out fabrics.

Lanes8 × 200G
ReachDR8 / 2×FR4

Learn more

800G

The scale-out workhorse

Today’s default rate for large AI clusters — available in OSFP and the familiar QSFP-DD footprint.

800G OSFP224 transceiver
800G
OSFP224

Single-lambda 200G optics in OSFP, balancing the highest density with strong thermal headroom for AI clusters.

Lanes4 × 200G
ReachDR4 / FR4 / LR4

Learn more

800G OSFP112 transceiver
800G
OSFP112

Proven 100G-per-lane 800G module for high-radix switches, with broad host and switch interoperability.

Lanes8 × 100G
ReachDR8 / 2×FR4

Learn more

800G QSFP112-DD transceiver
800G
QSFP112-DD

800G in the familiar QSFP-DD footprint, easing migration for operators standardizing on QSFP cages.

Lanes8 × 100G
ReachDR8 / 2×FR4

Learn more

400G

Proven, efficient, everywhere

Compact, low-power optics for leaf and access tiers, plus longer single-mode reach for DCI and metro links.

400G QSFP112 transceiver
400G
QSFP112

Compact 100G-per-lane 400G optics for leaf and access tiers, with low power draw and a small thermal envelope.

Lanes4 × 100G
ReachDR4 / FR4 / LR4

Learn more

400G OSFP112 transceiver
400G
OSFP112

400G in OSFP for deployments that prefer the larger cage and added heat-sink margin across mixed fabrics.

Lanes4 × 100G
ReachDR4 / FR4 / LR4

Learn more

400G QSFP56-DD transceiver
400G
QSFP56-DD

Mature 50G-per-lane 400G workhorse for installed networks and broad backward compatibility.

Lanes8 × 50G
ReachDR4 / FR4 / LR4

Learn more

400G OSFP DR4+ transceiver
400G
OSFP DR4+

100G-per-lane 400G in OSFP with a 1310 nm DR4+ optical path, extending single-mode reach to 2 km with sub-10 W power.

Lanes4 × 100G
Reach2 km (DR4+)

Learn more

Get in touch

Build your AI fabric on one platform.

Talk to our team