QSFP28 delivers 100G per module using 4 lanes at 25G each. QSFP-DD delivers up to 400G per module using 8 lanes at 50G each, in a housing that fits the same port footprint as QSFP28.
That single architectural difference drives every tradeoff below.
QSFP28 runs 4 electrical lanes at 25G per lane for a total of 100G per port. It is the dominant form factor for 100G deployments and remains widely used in 400G architectures as a breakout target.
Key characteristics:
QSFP28 is mature, widely stocked, and priced competitively in the third-party market. If your 400G switch uses QSFP-DD ports, you can still populate them with QSFP28 modules where density requirements allow.
QSFP-DD (Quad Small Form Factor Double Density) doubles the lane count to 8 by adding a second row of electrical contacts. Each lane runs at 50G PAM4, giving a total throughput of 400G per module. The housing is mechanically backward compatible with QSFP28 ports on switches that support it.
Key characteristics:
QSFP-DD is the preferred form factor for hyperscale spine layers and AI/ML cluster interconnects where 400G per port is the baseline requirement, not the ceiling.
This is where the decision often gets made.
A 32-port QSFP-DD switch delivers 32 x 400G = 12.8T of switching capacity in a 1RU chassis. The same chassis populated with QSFP28 modules in breakout mode gives you 128 x 100G ports, useful for leaf-spine designs where 100G server connections are the norm.
If your goal is maximizing 400G uplink density in a fixed chassis, QSFP-DD wins. If you need high 100G port counts from a single switch, QSFP28 breakout from a QSFP-DD chassis is often the more practical path than buying a dedicated QSFP28 switch.
QSFP28 modules typically draw 3.5W to 4.5W. QSFP-DD 400G modules run 10W to 15W depending on variant and reach. That difference matters at scale.
A 32-port QSFP-DD chassis fully populated at 15W per port pulls 480W in optics alone. Factor that into your power budget and cooling design before committing. For edge deployments or enterprise aggregation layers where port count is moderate, the power delta is manageable. For hyperscale builds with hundreds of ports, it is a real line item.
QSFP-DD ports on compliant switches accept QSFP28 and QSFP+ modules. This means a QSFP-DD switch can serve as your 400G spine today while accepting 100G QSFP28 modules in ports connected to existing 100G leaf switches or servers. You do not have to rip and replace your entire fabric at once.
QSFP28 ports do not accept QSFP-DD modules. The lane count and electrical interface are incompatible in that direction.
| Variant | Form Factor | Reach | Media |
|---|---|---|---|
| SR4 | QSFP28 | 100M | MMF |
| CWDM4 | QSFP28 | 2KM | SMF |
| LR4 | QSFP28 | 10KM | SMF |
| ER4 | QSFP28 | 40KM | SMF |
| SR8 | QSFP-DD | 100M | MMF |
| DR4 | QSFP-DD | 500M | SMF |
| FR4 | QSFP-DD | 2KM | SMF |
| LR4 | QSFP-DD | 10KM | SMF |
| ER8 | QSFP-DD | 40KM | SMF |
Both form factors support OTN, Ethernet, and Fibre Channel encapsulation in appropriate variants. QSFP-DD ER8 and LR8 variants extend to 40KM and beyond, covering metro aggregation and DCI use cases without requiring amplification on shorter spans.
Choose QSFP28 when:
Choose QSFP-DD when:
For most enterprise IT teams upgrading from 10G or 25G access layers, QSFP28 at the leaf and QSFP-DD at the spine is the practical answer in 2026. It preserves existing cabling investments while giving the spine layer room to grow.
OEM transceivers from vendors like Cisco run $200 to $500-plus per module for QSFP28 and significantly more for QSFP-DD 400G variants. At scale, that pricing makes large fabric refreshes prohibitively expensive.
Third-party compatible modules deliver 70% to 90% cost savings on both form factors. For a 32-port QSFP-DD spine switch, the difference between OEM and third-party pricing on the optics alone can reach tens of thousands of dollars per chassis.
The compatibility risk that used to accompany third-party modules is well-managed today through proper vendor selection. Compatibility test videos, published datasheets, and verified interoperability with Cisco, Juniper, Huawei, and Arista platforms reduce that uncertainty significantly.
HYTOPTODEVICE stocks both QSFP28 and QSFP-DD modules across the full range of variants and reach distances. The catalog covers SR, LR, ER, CWDM, and DWDM options for both form factors, with compatibility test videos available before you commit to a purchase. If you need custom-programmed or white-label modules for an OEM/ODM run of 100 to 4,000 units, that program is available as well.
The most common mistake in 400G planning is buying QSFP-DD everywhere because 400G sounds future-proof, then discovering that half the ports sit at 100G for the next three years while the power bill reflects 400G optics.
Match the form factor to the actual traffic requirement at each layer:
Both form factors are available in the HYTOPTODEVICE catalog at hytoptodevice.com, where you can download datasheets and review compatibility test videos to validate your selection before ordering.
Conclusion
The right answer between QSFP28 and QSFP-DD depends on where in your fabric you are deploying, what your actual per-link traffic looks like, and how your power budget is structured. Neither form factor is universally superior. Match the spec to the layer, validate compatibility before ordering, and price the full BOM including chassis and power infrastructure. Learn more at hytoptodevice.com.
Q: Can I use a QSFP28 module in a QSFP-DD port? A: Yes, on switches that support backward compatibility. QSFP-DD ports are mechanically and electrically designed to accept QSFP28 and QSFP+ modules. Check your switch vendor's datasheet to confirm which port modes are supported, as not all platforms enable this on every port simultaneously.
Q: Is QSFP-DD always the right choice for a 400G upgrade? A: Not automatically. QSFP-DD makes sense at the spine layer or in high-density 400G server interconnect environments. At the leaf or access layer where 100G connections still dominate, QSFP28 is often the more cost-efficient and power-efficient choice.
Q: What is the power difference between QSFP28 and QSFP-DD at scale? A: QSFP28 modules typically draw 3.5W to 4.5W. QSFP-DD 400G modules run 10W to 15W. In a 32-port chassis fully populated with QSFP-DD modules at 15W each, optics alone account for 480W. That is a meaningful difference in power and cooling planning for large deployments.
Q: Are third-party QSFP-DD modules compatible with Cisco and Juniper switches? A: Third-party compatible QSFP-DD modules are available for Cisco, Juniper, Arista, Huawei, and other major platforms. Compatibility depends on correct programming and firmware. Reviewing published compatibility test videos and datasheets from your supplier before purchasing is the standard validation step.
Q: What reach distances are available for QSFP-DD 400G modules? A: QSFP-DD 400G variants cover 100M (SR8 on MMF), 500M (DR4 on SMF), 2KM (FR4 on SMF), 10KM (LR4 on SMF), and 40KM-plus (ER8/LR8 on SMF). The right variant depends on your link distance and fiber type.
Q: Can QSFP-DD modules support 800G? A: Newer QSFP-DD variants running 8 lanes at 100G PAM4 reach 800G per module. These are distinct from the 400G QSFP-DD modules discussed in this article and require switch ASICs and port hardware that support 800G operation. Standard 400G QSFP-DD switches do not support 800G modules.
Q: Where can I source both QSFP28 and QSFP-DD modules with verified compatibility? A: HYTOPTODEVICE at hytoptodevice.com stocks both form factors across SR, LR, ER, CWDM, and DWDM variants with compatibility test videos and downloadable datasheets available pre-purchase. The catalog spans 1.25G to 800G, so you can source your full 400G upgrade from a single supplier.
