Traditional 100G QSFP28 modules—SR4, LR4, CWDM4, PSM4—all use four optical lanes, each carrying 25G over NRZ (non-return-to-zero) modulation. You get 100G total by aggregating those four lanes.
Single-lambda flips that architecture. It transmits 100G over a single wavelength using PAM4 (pulse amplitude modulation, 4-level) signaling. One lane, one wavelength, full 100G. That's the same PAM4 foundation that 400G and 800G modules rely on—which is exactly what makes single-lambda a genuine bridge technology rather than a dead end.
The 100G Lambda MSA defines the specification. It targets 500m reach over OM4 multimode fiber and up to 2km over SMF, using the QSFP28 form factor. That means single-lambda modules drop into existing QSFP28 ports on your current switch hardware with no forklift upgrade required.
Here's how the three approaches stack up across the specs that matter most for infrastructure planning:
| Parameter | Traditional 100G (SR4/LR4) | 100G Single-Lambda | 400G QSFP-DD / OSFP |
|---|---|---|---|
| Modulation | NRZ | PAM4 | PAM4 |
| Optical lanes | 4 | 1 | 4 or 8 |
| Fiber pairs | 4 (MPO-12) | 1 (LC duplex) | 2–4 (MPO-16 or MPO-12) |
| Typical reach | 100m–10km (varies by type) | 500m (MM) / 2km (SM) | 100m–10km (varies by type) |
| Form factor | QSFP28 | QSFP28 | QSFP-DD / OSFP |
| Switch hardware change needed | No | No | Yes (in most cases) |
| Typical power draw | 1.5–3.5W | ~1.5W | 8–15W |
The fiber architecture difference is worth calling out. SR4 and PSM4 require MPO-12 ribbon fiber. Single-lambda uses standard LC duplex—the same cabling plant you likely already have for 10G SFP+ links. That cuts both cabling cost and deployment complexity, especially in parallel rollouts.
The 400G QSFP-DD power figure also deserves attention. At 8–15W per port versus roughly 1.5W for single-lambda, a 48-port switch fully populated with 400G modules draws substantially more power. At scale, that difference shows up directly in your PUE and cooling costs.
Module cost is only one line item. The real upgrade cost depends on whether you need new switch hardware, new cabling, and higher power density.
Third-party compatible modules from a factory-direct supplier like HYTOPTODEVICE run 60–90% below OEM list price at any volume—whether you're buying 10 units or 1,000. That applies to 100G QSFP28 variants and 400G QSFP-DD alike. The OEM pricing gap is real: Cisco's list price for a single 100G LR4 module can exceed $1,000, while a switch-verified compatible alternative runs well under $200.
This is where the 400G path gets expensive fast. Most 400G deployments require new switch hardware. QSFP-DD and OSFP ports don't exist on current-generation 100G switches. If your Cisco Nexus 9300 or Arista 7050 is mid-lifecycle, adding 400G capacity means either a full switch replacement or deploying new spine hardware alongside existing gear.
Single-lambda 100G avoids this entirely. The module drops into any existing QSFP28 port—no new ASICs, no new chassis, no new line cards.
Single-lambda's LC duplex compatibility is a real cost advantage over SR4/PSM4 (MPO-12) and 400G SR8 (MPO-16). If your data center is already wired with LC duplex SMF or OM4 runs, single-lambda 100G requires zero recabling.
At 1.5W per port, single-lambda 100G draws roughly the same as a 10G SFP+ module. A 48-port switch at full single-lambda population draws around 72W in optics alone. The same switch populated with 400G QSFP-DD modules would draw 384–720W just in optics at 8–15W per port. At $0.10/kWh, that difference compounds significantly over a 3–5 year hardware cycle.
Many real-world data centers don't fit cleanly into either bucket. A common pattern: single-lambda 100G at the access and aggregation layers, 400G QSFP-DD at the spine. This preserves existing switch investments at the edge while giving you the bandwidth headroom you need at the core.
A practical hybrid architecture looks like this:
This approach lets you stage the 400G investment across capital cycles. The spine refresh happens on your schedule. Access and aggregation layers run single-lambda 100G until those switches reach end-of-life, at which point you refresh them to 400G-native hardware.
HYTOPTODEVICE stocks both sides of this architecture: 100G QSFP28 single-lambda modules verified for Cisco, Arista, Juniper, and Huawei platforms, alongside 400G QSFP-DD DR4, FR4, and SR8 for spine builds. Factory-direct pricing means no distributor markup on either side of the deployment.
One practical concern with any third-party module is the unsupported transceiver warning. Cisco, Arista, and Juniper all use vendor coding to flag non-OEM modules. The fix is proper firmware coding—not a workaround.
Switch-verified compatible modules from a supplier that codes firmware correctly will not generate persistent warning messages. HYTOPTODEVICE codes modules for Cisco, Arista, Juniper, and Huawei out of the box, with compatibility documentation available for procurement sign-off. That matters when you're justifying a third-party purchase to management or a compliance team.
Q1:Can 100G single-lambda and 400G modules coexist on the same switch?
A:Not on the same physical switch in most cases. Single-lambda 100G uses QSFP28 ports; 400G QSFP-DD and OSFP use physically larger, different port cages. A 400G-capable switch can often break out ports to 4x100G using QSFP-DD breakout cables, but a standard QSFP28 switch cannot host 400G QSFP-DD modules. In a hybrid deployment, single-lambda 100G and 400G coexist at different network layers—not on the same switch.
Q2:What hardware changes are required to move to 400G?
A:In most deployments, moving to 400G requires new switch hardware with QSFP-DD or OSFP port cages. Current-generation 100G switches with QSFP28 ports cannot accept QSFP-DD modules. You'll also likely need new cabling: 400G SR8 uses MPO-16, and 400G DR4 uses MPO-12. The exception is breakout configurations where a 400G QSFP-DD port splits into 4x100G QSFP28 links—but that still requires a 400G-capable switch at one end.
Q3:What are realistic deployment scenarios for 100G single-lambda today?
A:The most common use cases in 2026: upgrading access-to-aggregation uplinks on existing QSFP28 switches without hardware replacement; replacing 4x25G breakout links with cleaner 1x100G single-lambda connections over LC duplex; and extending 100G reach over existing SMF cabling where MPO infrastructure isn't in place. It's also well-suited for edge data center and colocation environments where power per port is a hard constraint.
Q4:How much power does 100G single-lambda save compared to traditional 4-channel 100G?
A:The savings are real, if not dramatic. A 100G SR4 module typically draws 2.5–3.5W; a single-lambda 100G module draws approximately 1.5W. Across a 48-port switch, that's a difference of 48–96W—roughly 420–840 kWh per year per switch. The bigger power story is the comparison against 400G QSFP-DD at 8–15W per port, where single-lambda 100G draws 80–90% less power per port.
Q5:Are 100G single-lambda modules backward compatible with existing QSFP28 ports?
A:Yes. Single-lambda 100G modules use the standard QSFP28 form factor and are electrically compatible with any QSFP28 port. The host switch sees a standard 100G interface. The difference is internal: single-lambda uses one optical lane with PAM4 instead of four lanes with NRZ. Your switch software doesn't need to distinguish between the two—it just sees a 100G link. That said, verify the module is coded for your specific switch platform to avoid unsupported transceiver warnings.
The choice between 100G single-lambda and 400G isn't about which is technically superior. It's about where you are in your hardware cycle, what your traffic patterns actually demand, and what your capital budget allows in 2026.
Single-lambda 100G is a technically sound bridge: same QSFP28 form factor, LC duplex cabling, low power draw, and PAM4 architecture that aligns with the 400G/800G direction. It's the right call for access and aggregation layers where a full 400G hardware refresh isn't justified yet.
400G QSFP-DD and OSFP belong at the spine, in AI cluster interconnects, and anywhere you're doing a planned hardware refresh.
For factory-direct pricing on both sides of that architecture—with switch-verified compatibility for Cisco, Arista, Juniper, and Huawei—see hytoptodevice.com.
