QSFP112 (Quad Small Form-factor Pluggable 112)<\/a> is a hot-pluggable optical transceiver module that delivers 400 Gbps of aggregate bandwidth using four electrical lanes operating at 112 Gbps per lane with PAM4 signaling. It maintains the same physical dimensions as the QSFP28 and QSFP56 form factors, which allows data centers to upgrade bandwidth density without replacing switch cages or physical fiber infrastructure.<\/p>\n <\/p>\n The “112” in QSFP112 refers to the 112 Gbps data transmission rate per electrical lane. This distinguishes it from earlier generations: QSFP28<\/a> uses 4 x 25 Gbps lanes, and QSFP56 uses 4 x 56 Gbps lanes. QSFP112 doubles the per-lane rate again to reach 400 Gbps total in the same compact package.<\/p>\n QSFP112 modules comply with several industry standards that ensure interoperability across vendors:<\/p>\n <\/p>\n Inside a QSFP112 optical transceiver module, four electrical lanes carry 112 Gbps each using PAM4 (Pulse Amplitude Modulation 4-level) signaling. PAM4 encodes two bits per symbol by using four distinct voltage levels, which doubles the throughput at the same baud rate compared to traditional NRZ (Non-Return-to-Zero) signaling.<\/p>\n This higher signaling complexity requires advanced signal processing. Each QSFP112 module contains:<\/p>\n <\/p>\n The optical engine then modulates the corrected electrical signals onto laser carriers for transmission over fiber optic cable. On the receive side, the process reverses: photodetectors convert optical signals back to electrical, and the DSP\/CDR\/FEC chain recovers the original data.<\/p>\n <\/p>\n <\/p>\n <\/p>\n The QSFP112 electrical interface operates at 112 Gbps per lane using PAM4 modulation. This represents a significant leap from the 25 Gbps NRZ signaling used in QSFP28 modules. The electrical specifications follow the OIF-CEI-112G-VSR-PAM4 standard, which defines the channel loss budgets, crosstalk limits, and jitter tolerances required for reliable operation.<\/p>\n Because PAM4 signals are more susceptible to noise than NRZ, FEC is mandatory for QSFP112 links. Most 400G Ethernet implementations use Reed-Solomon FEC (RS-FEC, RS(544,514)) as defined in IEEE 802.3. RS-FEC introduces additional latency, typically in the hundreds of nanoseconds range depending on the switch ASIC and implementation.Some high-performance computing environments may use lighter FEC modes or even firecode FEC if the channel quality supports it, though this requires careful validation.<\/p>\n <\/p>\n QSFP112 optical transceiver modules typically consume 8 to 12 watts per module, depending on the optical variant and reach. Short-reach multimode modules such as SR4 tend toward the lower end of this range, while long-reach single-mode modules with complex DSP such as LR4 may approach the upper limit.<\/p>\n This power consumption is roughly 30 percent lower per module than equivalent QSFP-DD 400G modules, which typically draw 10 to 14 watts. For a 32-port 400G switch fully populated with QSFP112 modules, the power savings can reach 64 to 128 watts of optical power alone. In dense data center racks where every watt translates to cooling load, this difference directly affects operational expenditure.<\/p>\n QSFP112 modules use a flat-top thermal design that maximizes contact area with the switch heatsink. The module case temperature must remain within the MSA-specified range, typically 0 to 70 degrees Celsius for commercial-grade units. Data center operators should verify that their switch platforms provide adequate airflow across the QSFP112 cage array, especially in fully loaded 1RU chassis.<\/p>\n <\/p>\n The QSFP112 form factor maintains the same physical envelope as QSFP28: approximately 18.4 mm wide, 89.4 mm long, and 8.5 mm high. This dimensional consistency is deliberate. It allows network equipment vendors to design 400G switches that reuse the same front-panel cage and PCB keepout areas as their 100G predecessors.<\/p>\n In practice, this means a 1RU switch can house up to 32 QSFP112 ports, delivering 12.8 Tbps of switching capacity in a single rack unit. This port density is essential for hyperscale data centers that need to maximize bandwidth per square meter of floor space.<\/p>\n <\/p>\n <\/p>\n QSFP112 optical transceiver modules are available in several variants optimized for different transmission distances and fiber types. Selecting the correct type depends on your link budget, fiber infrastructure, and connector compatibility.<\/p>\n <\/p>\n For short-reach connections inside data centers, QSFP112 VR4 and SR4 modules transmit over multimode fiber (MMF) using 850 nm vertical-cavity surface-emitting lasers (VCSELs).<\/p>\n Both variants use an MPO-12\/APC (Angled Physical Contact) connector. The APC polish is critical because flat-polished MPO connectors can reflect light back into the transmitter, causing signal degradation at these power levels. The twelve-fiber MPO ribbon cable carries four transmit and four receive fibers, with four fibers remaining unused.<\/p>\n These modules are ideal for intra-rack and adjacent-rack connections in data centers where the switch and server or storage arrays are within the same row.<\/p>\n <\/p>\n For longer reaches and data center interconnect applications, single-mode fiber (SMF) variants typically use four parallel 1310nm optical lanes implemented through EML or silicon photonics technologies.<\/p>\n <\/p>\n Deploying QSFP112 modules requires attention to connector and fiber compatibility. Here is a quick checklist for network engineers:<\/p>\n Connector type verification should be part of every 400G deployment plan, as MPO polarity and connector polish mismatches can significantly delay installation.<\/p>\n <\/p>\n <\/p>\n <\/p>\n Network engineers evaluating 400G optical transceiver options must choose between three primary form factors: QSFP112, QSFP-DD, and OSFP. Each has distinct trade-offs in power, density, backward compatibility, and upgrade path.<\/p>\n <\/p>\n QSFP112<\/b><\/strong><\/p>\n<\/td>\n OSFP<\/b><\/strong><\/p>\n<\/td>\n<\/tr>\n 4 x 112 Gbps<\/p>\n<\/td>\n 8 x 50 Gbps or 8 x 100 Gbps<\/p>\n<\/td>\n<\/tr>\n 400 Gbps<\/p>\n<\/td>\n 400 Gbps \/ 800 Gbps \/ 1.6 Tbps<\/p>\n<\/td>\n<\/tr>\n Same as QSFP28<\/p>\n<\/td>\n Larger than QSFP28<\/p>\n<\/td>\n<\/tr>\n 8 – 12W<\/p>\n<\/td>\n 12 – 16W<\/p>\n<\/td>\n<\/tr>\n QSFP28\/QSFP56 cages<\/p>\n<\/td>\n Requires an adapter for QSFP28<\/p>\n<\/td>\n<\/tr>\n No<\/p>\n<\/td>\n Yes<\/p>\n<\/td>\n<\/tr>\n Power-sensitive 400G, AI\/HPC<\/p>\n<\/td>\n Greenfield 800G+ deployments<\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n <\/p>\n <\/p>\n <\/p>\n QSFP112 optical transceiver modules excel in scenarios where power efficiency and physical compatibility with existing QSFP infrastructure are priorities. Choose QSFP112 when:<\/p>\n <\/p>\n <\/p>\n QSFP-DD and OSFP are better choices when future-proofing for 800G is a requirement. QSFP-DD achieves 400G using eight 50G lanes and can scale to 800G using eight 100G lanes. OSFP offers the largest thermal envelope and is the preferred form factor for 800G and 1.6 Tbps switch platforms.<\/p>\n Choose QSFP-DD when you need backward compatibility with QSFP28 modules in a mixed 100G\/400G\/800G environment. Choose OSFP for new greenfield data centers where maximum thermal headroom and the longest upgrade path are priorities.<\/p>\nQSFP112 Naming and Standards<\/strong><\/h3>\n
\n
How QSFP112 Works<\/strong><\/h3>\n
\n
![\"How](\"https:\/\/ascentoptics.com\/blog\/wp-content\/uploads\/2026\/06\/How-QSFP112-Works\uff1a4-\u00d7-112G-PAM4.png\")
<\/p>\nQSFP112 Technical Specifications<\/strong><\/h2>\n
Electrical Interface and Signaling<\/strong><\/h3>\n
Power Consumption and Thermal Design<\/strong><\/h3>\n
Physical Dimensions and Port Density<\/strong><\/h3>\n
Types of QSFP112 Optical Transceivers<\/strong><\/h2>\n
Multimode Fiber Variants (VR4 \/ SR4)<\/strong><\/h3>\n
\n
Single-Mode Fiber Variants (DR4 \/ FR4 \/ LR4)<\/strong><\/h3>\n
\n
Connector and Fiber Infrastructure Requirements<\/strong><\/h3>\n
\n
![\"Connector](\"https:\/\/ascentoptics.com\/blog\/wp-content\/uploads\/2026\/06\/Connector-and-Fiber-Infrastructure-Requirements.png\")
<\/p>\nQSFP112 vs. QSFP-DD vs. OSFP: How to Choose<\/strong><\/h2>\n
\n\n
\n Specification<\/b><\/strong><\/td>\n \n QSFP-DD<\/b><\/strong><\/td>\n \n \n Electrical lanes<\/td>\n \n 8 \u00d7 50G or 4 \u00d7 100G<\/td>\n \n \n Aggregate bandwidth<\/td>\n \n 400 Gbps \/ 800 Gbps<\/td>\n \n \n Form factor size<\/td>\n \n Slightly deeper than QSFP28<\/td>\n \n \n Typical power<\/td>\n \n 10 – 14W<\/td>\n \n \n Backward compatibility<\/td>\n \n QSFP28\/QSFP56 cages<\/td>\n \n \n 800G upgrade path<\/td>\n \n Yes<\/td>\n \n \n Best use case<\/td>\n \n Enterprise 400G\/800G migration<\/td>\n \n ![\"QSFP112](\"https:\/\/ascentoptics.com\/blog\/wp-content\/uploads\/2026\/06\/QSFP112-vs-QSFP-DD-vs-OSFP.png\")
<\/p>\nWhen to Choose QSFP112<\/strong><\/h3>\n
\n
When to Choose QSFP-DD or OSFP<\/strong><\/h3>\n
![\"QSFP112](\"https:\/\/ascentoptics.com\/blog\/wp-content\/uploads\/2026\/06\/QSFP112-in-AI-HPC-Cloud-Networks.png\")
<\/p>\n