Breakout Concept in High-Speed Networking<\/strong><\/h3>\nA breakout cable is a single cable assembly that divides one high-speed port into multiple lower-speed ports. In optical networking infrastructure, this concept has existed since the 40G era, when QSFP+ ports first broke out to four 10G SFP+ connections. The same principle applies today at 400G and 800G speeds.<\/p>\n
Breakout cables matter because data center upgrades rarely happen all at once. A network engineer might install a 400G spine switch this quarter while leaf switches remain at 100G for the next 12 to 18 months. Breakout cables allow the spine to operate at full capacity while still connecting to existing leaf infrastructure. They also enable flexible port allocation, letting a switch administrator reconfigure a 400G port into four independent 100G logical ports through software.<\/p>\n
<\/p>\n
QSFP112 Form Factor Basics<\/strong><\/h3>\nQSFP112 (Quad Small Form-Factor Pluggable 112)<\/a> is a 400G optical transceiver form factor defined by the\u00a0QSFP112 MSA. It uses four electrical lanes operating at 112 Gbps PAM4 signaling, delivering 400 Gbps aggregate bandwidth. The physical dimensions match earlier QSFP generations, which means QSFP112 modules fit into the same port cages as QSFP28 and QSFP56 modules on compatible switches.<\/p>\nThis backward-compatible footprint is important for breakout deployments. A switch designed for QSFP112 can often accept QSFP28 or QSFP56 modules in the same physical slot, though the port must be configured for the correct speed and signaling mode. The QSFP112 form factor is also the native interface for NVIDIA ConnectX-7 network interface cards and Quantum-2 InfiniBand switches, making it the de facto standard for AI and HPC clusters.<\/p>\n
<\/p>\n
Common QSFP112 Breakout Configurations<\/strong><\/h3>\nAs 800G networking becomes increasingly common in AI clusters, cloud data centers, and HPC environments, breakout connectivity plays a critical role in bridging next-generation infrastructure with existing 400G deployments. Among various breakout options, 800G OSFP to 2\u00d7400G QSFP112 has emerged as one of the most widely adopted configurations.<\/p>\n
<\/p>\n
800G OSFP to 2\u00d7400G QSFP112<\/strong><\/h3>\nThis is currently the most common breakout deployment for organizations migrating from 400G to 800G networks. A single 800G OSFP port is divided into two independent 400G QSFP112 links, allowing new 800G switches to connect directly to existing 400G switches, NICs, and AI servers.<\/p>\n
Typical applications include:<\/p>\n
\n- \u2022<\/span><\/strong>800G spine switches connecting to 400G leaf switches<\/li>\n
- \u2022<\/span><\/strong>NVIDIA AI and HPC clusters using ConnectX-7 adapters<\/li>\n
- \u2022<\/span><\/strong>Gradual migration from 400G to 800G infrastructure<\/li>\n
- \u2022<\/span><\/strong>High-density cloud data center deployments<\/li>\n<\/ul>\n
<\/p>\n
Because both interfaces use high-speed PAM4 signaling and 400G QSFP112 remains widely deployed across AI and data center networks, this breakout configuration provides an efficient upgrade path while protecting existing infrastructure investments.<\/p>\n
Understanding\u00a0OSFP breakout cable<\/u><\/a>\u00a0options helps network architects design flexible migration paths that minimize stranded capacity during speed upgrades.<\/p>\n <\/p>\n
![\"800G](\"https:\/\/ascentoptics.com\/blog\/wp-content\/uploads\/2026\/06\/800G-OSFP-to-2\u00d7400G-QSFP112-Breakout-Architecture.png\")
<\/p>\n
<\/p>\n
<\/p>\n
Other QSFP112 Breakout Configurations<\/strong><\/h3>\nDepending on switch architecture and platform support, additional breakout configurations may also be available:<\/p>\n
400G QSFP112 to 2\u00d7200G QSFP56<\/strong><\/p>\nA single 400G QSFP112 port is split into two independent 200G QSFP56 connections.\u00a0This configuration is commonly used in mixed-speed Ethernet networks where 200G equipment remains in production.<\/p>\n
400G QSFP112 to 4\u00d7100G QSFP28<\/strong><\/p>\nSome platforms support breaking a 400G port into four 100G connections through ASIC-level breakout functionality.\u00a0Availability depends on the switch ASIC, firmware, and breakout implementation.<\/p>\n
Key Takeaway<\/strong><\/p>\nFor modern AI, HPC, and cloud networking deployments, 800G OSFP to 2\u00d7400G QSFP112 breakout cables represent the most practical and widely deployed breakout architecture, enabling seamless interoperability between emerging 800G platforms and existing 400G ecosystems.<\/p>\n
<\/p>\n
<\/p>\n
QSFP112 Breakout Cable Types: DAC, AEC, and AOC<\/strong><\/h2>\nChoosing the right cable type for a QSFP112 breakout deployment requires understanding the trade-offs between reach, power, cost, and signal integrity. Here is how the three main categories compare.<\/p>\n
<\/p>\n
Passive Breakout DAC<\/strong><\/h3>\nPassive Direct Attach Copper (DAC)<\/a> breakout cables use pure twinax copper wiring with no active electronics in the connector housing. The QSFP112 MSA and IEEE 802.3ck standards define the electrical interface, but the cable itself is a passive transmission medium.<\/p>\nFor QSFP112 breakout configurations, passive DAC reach is typically limited to 0.5 to 2 meters. The 112G PAM4 signaling rate experiences significant attenuation in copper, and the additional trace complexity of splitting one port into four legs further reduces maximum reach. Power consumption is negligible at less than 0.1W per end, making passive breakout DAC the most energy-efficient option.<\/p>\n
Passive breakout DAC works best for intra-rack connections where the spine switch and leaf switches sit in adjacent rack units. A 1-meter passive breakout DAC can connect a top-of-rack 400G switch to four 100G servers mounted directly below it.<\/p>\n
<\/p>\n
Active Breakout ACC and AEC<\/strong><\/h3>\nActive Copper Cables (ACC) and Active Electrical Cables (AEC) add signal conditioning electronics inside the connector housing. These devices use redrivers or retimers to compensate for copper attenuation, extending reach beyond what passive cables can support.<\/p>\n
For QSFP112 breakout cables, active copper options reach 2 to 5 meters. Power consumption increases to 1 to 3W per end, which is still significantly lower than optical alternatives. Active breakout cables are the practical choice when you need to cross a row or connect devices within the same rack group, but cannot keep everything within 2 meters.<\/p>\n
The key difference between ACC and AEC is the type of signal conditioning. ACC cables use redrivers that amplify the signal but do not retime it. AEC cables use retimers that recover and regenerate the clock, providing a cleaner signal output. For QSFP112 breakout applications, AEC is generally preferred because retimers handle PAM4 signaling more reliably at 112G per lane.<\/p>\n
<\/p>\n
Breakout AOC<\/strong><\/h3>\nActive Optical Cable (AOC)<\/a> breakout assemblies convert electrical signals to optical signals inside the connector housing and transmit them over fiber. A QSFP112 to 4x100G breakout AOC uses four independent optical channels, each with its own transmitter and receiver.<\/p>\nBreakout AOC reach extends from 1 meter to 30 meters, depending on fiber type and optical power budget. This makes AOC the only practical choice for inter-rack connections spanning more than 5 meters. Weight and bulk are also significantly lower than copper breakout bundles. A 4-leg copper breakout DAC is thick and stiff, while an AOC breakout uses thin fiber ribbons that route easily through cable managers.<\/p>\n
The trade-off is power and cost. A QSFP112 breakout AOC consumes up to 10W at the host end and approximately 2.5W per breakout leg. For a 64-port deployment, that adds over 600W of heat compared to a passive DAC. Cost per link is also 5 to 10 times higher than copper.<\/p>\n
<\/p>\n
![\"800G](\"https:\/\/ascentoptics.com\/blog\/wp-content\/uploads\/2026\/06\/800G-OSFP-to-2\u00d7400G-QSFP112-Cable-Options.png\")
<\/p>\n
<\/p>\n
<\/p>\n
Cable Gauge and Bundle Considerations<\/strong><\/h3>\nBreakout DAC cables use multiple coaxial pairs bundled together. Common wire gauges range from 26 AWG to 30 AWG. Thicker 26 AWG cable supports longer reach and better signal integrity, but is less flexible. Thinner 30 AWG cable bends more easily but has higher insertion loss.<\/p>\n
When deploying breakout bundles in dense racks, cable management becomes critical. A single 4-leg breakout cable contains eight individual connectors, four on each end. In an AI cluster with hundreds of breakout links, unmanaged bundles block airflow and complicate maintenance. Use horizontal cable managers and strain relief brackets to keep breakout legs organized.<\/p>\n
<\/p>\n
\n\n\nCable Type<\/b><\/strong><\/td>\n\nMax Reach<\/b><\/strong><\/p>\n<\/td>\nPower (QSFP112 End)<\/b><\/strong><\/td>\nPower (Per Leg)<\/b><\/strong><\/td>\n\nBest Use Case<\/b><\/strong><\/p>\n<\/td>\n<\/tr>\n\nPassive Breakout DAC<\/td>\n \n0.5 to 2 m<\/p>\n<\/td>\n
< 0.1W<\/td>\n Near zero<\/td>\n \nIntra-rack, adjacent devices<\/p>\n<\/td>\n<\/tr>\n
\nActive Breakout AEC<\/strong><\/td>\n\n2 to 5 m<\/p>\n<\/td>\n
1 to 3W<\/td>\n 0.5 to 1W<\/td>\n \nCross-rack, row-level<\/p>\n<\/td>\n<\/tr>\n
\nBreakout AOC<\/td>\n \n1 to 30 m<\/p>\n<\/td>\n
Up to 8W<\/td>\n Up to 2.5W<\/td>\n \nInter-rack, longer reach<\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
<\/p>\n
<\/p>\n
Signaling and Why Some QSFP112 Breakouts Need Active Electronics<\/strong><\/h2>\nPAM4 vs NRZ Signaling Mismatch<\/strong><\/h3>\nThe electrical interface inside a QSFP112 port uses 112G PAM4 signaling. PAM4 encodes two bits per symbol by using four distinct voltage levels, which doubles the data rate compared to traditional NRZ signaling without doubling the baud rate. This is how QSFP112 achieves 400G over only four lanes.<\/p>\n
QSFP28 modules, by contrast, use 25G NRZ signaling. NRZ uses two voltage levels to encode one bit per symbol. A passive copper cable cannot convert PAM4 to NRZ. It simply transmits whatever electrical signal it receives. If you connect a QSFP112 PAM4 signal directly to a QSFP28 NRZ receiver through passive copper, the receiver cannot decode the four-level PAM4 waveform.<\/p>\n
<\/p>\n
When Active Conditioning Is Required<\/strong><\/h3>\nNot every QSFP112 breakout configuration needs a retimer. The requirement depends on the signaling format on both ends:<\/p>\n
\n- \u2022<\/span>QSFP112 (PAM4) to QSFP28 (NRZ)<\/strong>: The speed adaptation and lane mapping are typically handled by the switch ASIC or NIC. The breakout cable serves as the physical interconnect and does not normally perform protocol or modulation conversion.<\/li>\n
- \u2022<\/span>QSFP112 (PAM4) to QSFP56 (PAM4)<\/strong>: Passive copper may work. Both sides use PAM4, though the lane rates differ. Many switch ASICs can handle the rate adaptation internally.<\/li>\n
- \u2022<\/span>800G OSFP (PAM4) to QSFP112 (PAM4)<\/strong>: Passive copper often works for short reach. Both ends use the same signaling format at the same or similar lane rate.<\/li>\n<\/ul>\n
<\/p>\n
When evaluating breakout cables, always verify whether the cable includes a retimer. Product descriptions that say “passive breakout DAC” for a QSFP112 to 4x100G configuration are technically suspect unless the switch itself handles the signaling conversion internally.<\/p>\n
<\/p>\n
QSFP56 Breakout Simplification<\/strong><\/h3>\nThe 400G QSFP112 to 2x200G QSFP56 breakout is simpler from a signaling perspective. QSFP56 uses 50G PAM4 per lane. Both the source and destination use PAM4, so the breakout cable does not need to perform modulation format conversion. The switch ASIC or cable redriver handles lane rate adaptation from 112G to 50G.<\/p>\n
Because of this simplification, passive breakout DAC cables for QSFP112 to 2x200G are more widely available and less expensive than their 4x100G counterparts. If your endpoints support 200G, this configuration reduces cabling complexity and cost.<\/p>\n
<\/p>\n
<\/p>\n
QSFP112 Breakout vs QSFP-DD Breakout: Key Differences<\/strong><\/h2>\nForm Factor Architecture Comparison<\/strong><\/h3>\nQSFP112 and QSFP-DD are both 400G form factors, but their internal lane architectures differ significantly. QSFP112 uses four lanes at 112G PAM4. QSFP-DD supports multiple generations of lane architectures, including 8\u00d750G PAM4 for 400G Ethernet and 8\u00d7100G PAM4 for 800G Ethernet. This architectural difference has direct implications for breakout cable design.<\/p>\n
QSFP112 (Quad Small Form-Factor Pluggable 112)<\/a> is a 400G optical transceiver form factor defined by the\u00a0QSFP112 MSA. It uses four electrical lanes operating at 112 Gbps PAM4 signaling, delivering 400 Gbps aggregate bandwidth. The physical dimensions match earlier QSFP generations, which means QSFP112 modules fit into the same port cages as QSFP28 and QSFP56 modules on compatible switches.<\/p>\n This backward-compatible footprint is important for breakout deployments. A switch designed for QSFP112 can often accept QSFP28 or QSFP56 modules in the same physical slot, though the port must be configured for the correct speed and signaling mode. The QSFP112 form factor is also the native interface for NVIDIA ConnectX-7 network interface cards and Quantum-2 InfiniBand switches, making it the de facto standard for AI and HPC clusters.<\/p>\n <\/p>\n As 800G networking becomes increasingly common in AI clusters, cloud data centers, and HPC environments, breakout connectivity plays a critical role in bridging next-generation infrastructure with existing 400G deployments. Among various breakout options, 800G OSFP to 2\u00d7400G QSFP112 has emerged as one of the most widely adopted configurations.<\/p>\n <\/p>\n This is currently the most common breakout deployment for organizations migrating from 400G to 800G networks. A single 800G OSFP port is divided into two independent 400G QSFP112 links, allowing new 800G switches to connect directly to existing 400G switches, NICs, and AI servers.<\/p>\n Typical applications include:<\/p>\n <\/p>\n Because both interfaces use high-speed PAM4 signaling and 400G QSFP112 remains widely deployed across AI and data center networks, this breakout configuration provides an efficient upgrade path while protecting existing infrastructure investments.<\/p>\n Understanding\u00a0OSFP breakout cable<\/u><\/a>\u00a0options helps network architects design flexible migration paths that minimize stranded capacity during speed upgrades.<\/p>\n <\/p>\n <\/p>\n <\/p>\n Depending on switch architecture and platform support, additional breakout configurations may also be available:<\/p>\n 400G QSFP112 to 2\u00d7200G QSFP56<\/strong><\/p>\n A single 400G QSFP112 port is split into two independent 200G QSFP56 connections.\u00a0This configuration is commonly used in mixed-speed Ethernet networks where 200G equipment remains in production.<\/p>\n 400G QSFP112 to 4\u00d7100G QSFP28<\/strong><\/p>\n Some platforms support breaking a 400G port into four 100G connections through ASIC-level breakout functionality.\u00a0Availability depends on the switch ASIC, firmware, and breakout implementation.<\/p>\n Key Takeaway<\/strong><\/p>\n For modern AI, HPC, and cloud networking deployments, 800G OSFP to 2\u00d7400G QSFP112 breakout cables represent the most practical and widely deployed breakout architecture, enabling seamless interoperability between emerging 800G platforms and existing 400G ecosystems.<\/p>\n <\/p>\n <\/p>\n Choosing the right cable type for a QSFP112 breakout deployment requires understanding the trade-offs between reach, power, cost, and signal integrity. Here is how the three main categories compare.<\/p>\n <\/p>\n Passive Direct Attach Copper (DAC)<\/a> breakout cables use pure twinax copper wiring with no active electronics in the connector housing. The QSFP112 MSA and IEEE 802.3ck standards define the electrical interface, but the cable itself is a passive transmission medium.<\/p>\n For QSFP112 breakout configurations, passive DAC reach is typically limited to 0.5 to 2 meters. The 112G PAM4 signaling rate experiences significant attenuation in copper, and the additional trace complexity of splitting one port into four legs further reduces maximum reach. Power consumption is negligible at less than 0.1W per end, making passive breakout DAC the most energy-efficient option.<\/p>\n Passive breakout DAC works best for intra-rack connections where the spine switch and leaf switches sit in adjacent rack units. A 1-meter passive breakout DAC can connect a top-of-rack 400G switch to four 100G servers mounted directly below it.<\/p>\n <\/p>\n Active Copper Cables (ACC) and Active Electrical Cables (AEC) add signal conditioning electronics inside the connector housing. These devices use redrivers or retimers to compensate for copper attenuation, extending reach beyond what passive cables can support.<\/p>\n For QSFP112 breakout cables, active copper options reach 2 to 5 meters. Power consumption increases to 1 to 3W per end, which is still significantly lower than optical alternatives. Active breakout cables are the practical choice when you need to cross a row or connect devices within the same rack group, but cannot keep everything within 2 meters.<\/p>\n The key difference between ACC and AEC is the type of signal conditioning. ACC cables use redrivers that amplify the signal but do not retime it. AEC cables use retimers that recover and regenerate the clock, providing a cleaner signal output. For QSFP112 breakout applications, AEC is generally preferred because retimers handle PAM4 signaling more reliably at 112G per lane.<\/p>\n <\/p>\n Active Optical Cable (AOC)<\/a> breakout assemblies convert electrical signals to optical signals inside the connector housing and transmit them over fiber. A QSFP112 to 4x100G breakout AOC uses four independent optical channels, each with its own transmitter and receiver.<\/p>\n Breakout AOC reach extends from 1 meter to 30 meters, depending on fiber type and optical power budget. This makes AOC the only practical choice for inter-rack connections spanning more than 5 meters. Weight and bulk are also significantly lower than copper breakout bundles. A 4-leg copper breakout DAC is thick and stiff, while an AOC breakout uses thin fiber ribbons that route easily through cable managers.<\/p>\n The trade-off is power and cost. A QSFP112 breakout AOC consumes up to 10W at the host end and approximately 2.5W per breakout leg. For a 64-port deployment, that adds over 600W of heat compared to a passive DAC. Cost per link is also 5 to 10 times higher than copper.<\/p>\n <\/p>\n <\/p>\n <\/p>\n Breakout DAC cables use multiple coaxial pairs bundled together. Common wire gauges range from 26 AWG to 30 AWG. Thicker 26 AWG cable supports longer reach and better signal integrity, but is less flexible. Thinner 30 AWG cable bends more easily but has higher insertion loss.<\/p>\n When deploying breakout bundles in dense racks, cable management becomes critical. A single 4-leg breakout cable contains eight individual connectors, four on each end. In an AI cluster with hundreds of breakout links, unmanaged bundles block airflow and complicate maintenance. Use horizontal cable managers and strain relief brackets to keep breakout legs organized.<\/p>\n <\/p>\n Max Reach<\/b><\/strong><\/p>\n<\/td>\n Best Use Case<\/b><\/strong><\/p>\n<\/td>\n<\/tr>\n 0.5 to 2 m<\/p>\n<\/td>\n Intra-rack, adjacent devices<\/p>\n<\/td>\n<\/tr>\n 2 to 5 m<\/p>\n<\/td>\n Cross-rack, row-level<\/p>\n<\/td>\n<\/tr>\n 1 to 30 m<\/p>\n<\/td>\n Inter-rack, longer reach<\/p>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n <\/p>\n <\/p>\n The electrical interface inside a QSFP112 port uses 112G PAM4 signaling. PAM4 encodes two bits per symbol by using four distinct voltage levels, which doubles the data rate compared to traditional NRZ signaling without doubling the baud rate. This is how QSFP112 achieves 400G over only four lanes.<\/p>\n QSFP28 modules, by contrast, use 25G NRZ signaling. NRZ uses two voltage levels to encode one bit per symbol. A passive copper cable cannot convert PAM4 to NRZ. It simply transmits whatever electrical signal it receives. If you connect a QSFP112 PAM4 signal directly to a QSFP28 NRZ receiver through passive copper, the receiver cannot decode the four-level PAM4 waveform.<\/p>\n <\/p>\n Not every QSFP112 breakout configuration needs a retimer. The requirement depends on the signaling format on both ends:<\/p>\n <\/p>\n When evaluating breakout cables, always verify whether the cable includes a retimer. Product descriptions that say “passive breakout DAC” for a QSFP112 to 4x100G configuration are technically suspect unless the switch itself handles the signaling conversion internally.<\/p>\n <\/p>\n The 400G QSFP112 to 2x200G QSFP56 breakout is simpler from a signaling perspective. QSFP56 uses 50G PAM4 per lane. Both the source and destination use PAM4, so the breakout cable does not need to perform modulation format conversion. The switch ASIC or cable redriver handles lane rate adaptation from 112G to 50G.<\/p>\n Because of this simplification, passive breakout DAC cables for QSFP112 to 2x200G are more widely available and less expensive than their 4x100G counterparts. If your endpoints support 200G, this configuration reduces cabling complexity and cost.<\/p>\n <\/p>\n <\/p>\n QSFP112 and QSFP-DD are both 400G form factors, but their internal lane architectures differ significantly. QSFP112 uses four lanes at 112G PAM4. QSFP-DD supports multiple generations of lane architectures, including 8\u00d750G PAM4 for 400G Ethernet and 8\u00d7100G PAM4 for 800G Ethernet. This architectural difference has direct implications for breakout cable design.<\/p>\nCommon QSFP112 Breakout Configurations<\/strong><\/h3>\n
800G OSFP to 2\u00d7400G QSFP112<\/strong><\/h3>\n
\n
<\/p>\nOther QSFP112 Breakout Configurations<\/strong><\/h3>\n
QSFP112 Breakout Cable Types: DAC, AEC, and AOC<\/strong><\/h2>\n
Passive Breakout DAC<\/strong><\/h3>\n
Active Breakout ACC and AEC<\/strong><\/h3>\n
Breakout AOC<\/strong><\/h3>\n
<\/p>\nCable Gauge and Bundle Considerations<\/strong><\/h3>\n
\n\n
\n Cable Type<\/b><\/strong><\/td>\n \n Power (QSFP112 End)<\/b><\/strong><\/td>\n Power (Per Leg)<\/b><\/strong><\/td>\n \n \n Passive Breakout DAC<\/td>\n \n < 0.1W<\/td>\n Near zero<\/td>\n \n \n Active Breakout AEC<\/strong><\/td>\n \n 1 to 3W<\/td>\n 0.5 to 1W<\/td>\n \n \n Breakout AOC<\/td>\n \n Up to 8W<\/td>\n Up to 2.5W<\/td>\n \n Signaling and Why Some QSFP112 Breakouts Need Active Electronics<\/strong><\/h2>\n
PAM4 vs NRZ Signaling Mismatch<\/strong><\/h3>\n
When Active Conditioning Is Required<\/strong><\/h3>\n
\n
QSFP56 Breakout Simplification<\/strong><\/h3>\n
QSFP112 Breakout vs QSFP-DD Breakout: Key Differences<\/strong><\/h2>\n
Form Factor Architecture Comparison<\/strong><\/h3>\n
![\"Migration](\"https:\/\/ascentoptics.com\/blog\/wp-content\/uploads\/2026\/06\/Migration-from-400G-to-800G-Networks.png\")
<\/p>\n![\"Where](\"https:\/\/ascentoptics.com\/blog\/wp-content\/uploads\/2026\/06\/Where-800G-to-2\u00d7400G-Breakout-Cables-Are-Used.png\")
<\/p>\n