What Is a DAC Cable? Direct Attach Copper Explained

By Dustin Guttadauro, Product Line Manager - Telecom & Fiber, Infinite Electronics  

A DAC cable (direct attach copper) is a short, fixed-length twinax copper cable with a transceiver-style plug, such as SFP+, SFP28 or QSFP28, permanently attached to each end. You plug it straight into a switch or server port, and it takes the place of two optical transceivers plus a fiber jumper. On links under about 5 to 7 m, that gets you lower cost, almost no added power draw and very low latency. 

Most DACs end up in the same place: inside a rack, between servers and a top-of-rack switch, or between two switches a few rack units apart. This guide covers passive vs active DACs, speed and reach limits for each form factor, a typical ToR wiring layout, and how DAC compares with AOC and Cat8. If you're planning the wider build, you can also browse L-com data center solutions. 

Key Takeaways 

• A DAC is twinax copper with fixed SFP or QSFP ends. There's no optical conversion, so a passive DAC draws close to zero power. 

• Passive SFP+ DACs reach up to 7 m at 10G, while passive SFP28 (25G) and QSFP28 (100G) DACs top out at 5 m, with 3 m the safer practical limit. 

• Active DACs add signal-conditioning chips that stretch reach to roughly 7 to 15 m depending on speed, in exchange for some power and a higher price. 

• For server-to-ToR links of 1 to 3 m, DAC is usually the cheapest, lowest-power and lowest-latency option you can buy. 

• Past about 5 m, or where cable bulk is already choking airflow, AOC is the better fit. Cat8 only makes sense if your gear has 25GBASE-T or 40GBASE-T RJ45 ports. 

What Is a DAC Cable? 

A DAC cable is a complete, factory-terminated link: two transceiver-shaped connector housings joined by shielded twinax copper. The housings match the port form factor (SFP+, SFP28, QSFP+, QSFP28, and newer QSFP-DD and OSFP versions), so the switch handles the cable much like an installed module. Each end holds a small EEPROM that tells the host what the cable is, how long it is and whether it's passive or active. 

The electrical signal from the switch ASIC travels directly over copper. There's no laser, no photodetector and no conversion between light and electricity, and that missing hardware is where most of the DAC's cost, power and latency advantage comes from. The trade-off is reach. High-speed electrical signals lose strength quickly in copper, so DACs are built for short hops. 

Twinax Cable: What's Inside a DAC 

Twinax cable has two inner conductors per pair, held at a fixed spacing inside a dielectric and wrapped in foil shielding. Each pair carries one differential signal lane. An SFP+ or SFP28 DAC needs two pairs (one to transmit, one to receive), and a QSFP DAC needs eight pairs for four lanes in each direction, which is why QSFP cables are noticeably fatter. 

Conductor size runs from about 30 AWG on short, flexible cables to 24 AWG on longer passive runs. Heavier gauge lowers loss but makes the cable stiffer and harder to dress in a crowded rack. For a closer look at construction and latency, see our twinax cabling deep-dive. 

Passive vs Active DAC Cables 

Passive DACs contain no signal-processing electronics, and active DACs add chips at one or both ends to clean up the signal. Reach, power, price and switch compatibility all follow from that one difference. 

Passive DAC 

A passive DAC is copper and connectors, nothing else. It relies on the SerDes in the switch or NIC to equalize the signal, which works well at short lengths and keeps power draw negligible. Passive cables are the default for in-rack links, and they're usually the first thing switch vendors qualify. 

Active DAC (ACC and AEC) 

An active copper cable (ACC) uses redriver or equalizer chips to boost the signal. An active electrical cable (AEC) goes further, with retimers that recover the signal and transmit a clean copy. Both allow thinner wire or longer lengths. You pay for that with a small power draw at each end and a higher price, though an active DAC usually still costs less than two optics and a fiber patch cord. 

Power and reach figures are approximate and vary by vendor and generation. 

Tip: Some older 10G switch ports accept only passive DACs, and others accept only "limiting" active cables. Check the switch vendor's compatibility list before you order active cables in volume. 

SFP+ DAC, SFP28 and QSFP: Speed and Reach Limits 

Each DAC form factor maps to a lane rate, and reach shrinks as lane speed rises. The table lists typical maximums based on the IEEE and SFF specifications and common vendor offerings, so confirm the exact figure on the datasheet for the cable you buy. 

The SFP+ DAC is still the most common copper link in enterprise racks. 10G server ports are everywhere, and 7 m covers almost any in-rack or next-rack run. At 25G the picture tightens: the 5 m figure assumes the host runs RS-FEC (Reed-Solomon forward error correction). Without it, plan on 3 m. 

Breakout DACs 

A breakout DAC splits one QSFP port into four SFP ends, either QSFP+ to 4 x SFP+ (40G to 10G) or QSFP28 to 4 x SFP28 (100G to 25G). It's a cheap way to connect four servers to one high-speed switch port, as long as the switch supports breakout mode on that port. 

Recommendation: Order the shortest length that reaches the port with a little slack. A 5 m cable coiled behind a 1U server adds bulk, blocks exhaust air and eats signal margin you didn't need to spend. 

Why Choose a DAC Cable Over Optics? 

On a short link, a DAC cable is usually cheaper, lower in power and lower in latency than a pair of transceivers with a fiber patch cord. Across a full row of racks, those differences add up quickly. 

Cost 

A DAC is one part instead of three. You skip two optical transceivers and a jumper, and nobody has to clean or inspect fiber end faces during the install. The exact saving depends on speed and vendor, but at 25G and 100G the gap per link is usually large, and it repeats on every server port in the room. 

Power and Cooling 

A passive DAC draws almost nothing. A typical 10G or 25G short-reach optic uses around 1 W, and a 100G QSFP28 SR4 module around 2.5 to 3.5 W (approximate; varies by vendor). Cable a 48-port 25G ToR switch with DAC instead of optics and you save roughly 50 W at the switch, a similar amount across the server NICs, and the cooling needed to remove that heat. 

Latency 

Passive DACs add essentially no processing delay. The signal only pays propagation time, a few nanoseconds per meter, which is a big reason DACs are common in trading and HPC racks. They also beat 10GBASE-T copper, where PHY encoding and DSP add on the order of 2 microseconds per link, against a small fraction of that for an SFP+ DAC link. 

A Typical Top-of-Rack DAC Wiring Layout 

In a top-of-rack (ToR) design, every server connects to a switch mounted in the same rack, so the longest server link is rarely more than 2 to 3 m. That short distance is where passive DAC fits best. 

Typical ToR layout: passive DACs inside each rack, AOC or optics on the uplinks to the spine. 

1. Servers connect to the ToR switch over 1 to 3 m passive DACs: SFP28 for 25G NICs, SFP+ for 10G. 

1. ToR switches in adjacent racks can link to each other (for an MLAG peer link, for example) with a short QSFP28 DAC. 

1. Uplinks from the ToR to the spine or aggregation layer usually run 10 m or more, so they use AOCs or optical transceivers over structured fiber. 

1. Where switch ports are tight, a breakout DAC can feed four 25G servers from one 100G ToR port. 

Tip: Route DACs down one side of the rack and keep them out of the server exhaust path. Follow the minimum bend radius on the datasheet, because thick passive twinax doesn't take tight bends at the port well. 

DAC vs AOC vs Cat8 

All three handle short data center links, but they suit different distances and different port types. 

When AOC Is the Better Choice 

Pick active optical cables once a run passes about 5 m, or when you're routing dozens of cables through tight overhead trays where copper weight and bulk become a real problem. AOCs plug into the same SFP and QSFP ports, so you can mix both on one switch: DAC inside the rack, AOC for row and spine runs. 

When Cat8 Makes Sense 

Cat8 cabling only pays off if your switches and servers have 25GBASE-T or 40GBASE-T RJ45 ports, and few data center switches ship with them. It's field-terminable and reaches 30 m, but BASE-T PHYs at those speeds draw more power and add more latency than an SFP-based link. On SFP28 and QSFP28 hardware, DAC is the simpler copper choice. 

Decision Checklist: Is a DAC Cable Right for Your Link? 

Sourcing DAC and AOC Assemblies 

L-com's direct attach copper (DAC) cables cover SFP and QSFP form factors for in-rack server and switch links, and the active optical cables above take over on the same ports once runs get longer. For the rest of the build, from patch panels to fiber trunks, start with the L-com data center cabling lineup. 

Frequently Asked Questions (FAQs) 

What is a DAC cable? 

A DAC cable (direct attach copper) is a short, fixed-length twinax copper cable with a transceiver-style plug, such as SFP+, SFP28 or QSFP28, permanently attached to each end. You plug it straight into a switch or server port, and it takes the place of two optical transceivers plus a fiber jumper. 

Is a DAC cable the same as a twinax cable? 

Not exactly. Twinax is the cable type: two inner conductors per pair, held at a fixed spacing inside a dielectric and wrapped in foil shielding. A DAC is a complete, factory-terminated assembly that joins shielded twinax copper to SFP or QSFP connector housings fixed to each end. 

Does a DAC cable need separate transceivers? 

No. The transceiver-style ends are part of the cable, so it plugs directly into the switch or NIC port. That's the main reason it costs less than an optical link, which needs two modules plus a fiber patch cord. 

What is the difference between a passive and an active DAC? 

A passive DAC is copper and connectors with no signal-processing electronics, so it draws near-zero power and relies on the host to equalize the signal. An active DAC adds redriver or retimer chips at one or both ends, which extends reach to roughly 7 to 15 m in exchange for some power draw and a higher price. 

How far can a DAC cable reach? 

Passive SFP+ DACs reach up to 7 m at 10G, while passive SFP28 (25G) and QSFP28 (100G) DACs top out at 5 m, with 3 m the safer practical limit. Active DACs extend reach to roughly 7 to 15 m depending on speed. Confirm the exact figure on the datasheet for the cable you buy. 

When should you use an AOC instead of a DAC? 

Use an active optical cable (AOC) once a run passes about 5 m, or when dozens of cables routed through tight overhead trays make copper weight and bulk a problem. AOCs plug into the same SFP and QSFP ports, so you can use DAC inside the rack and AOC for row and spine runs.