A 10GbE point-to-point link costs $62.68 today; a 25GbE one costs $367.97 - about 5.9x the price for 2.5x the link speed. Both figures are two cards plus one matching cable, no switch. This page prices the link and states what the vendors document; whether your storage can use the difference is a question answered on the hardware pages linked below, not here.
Prices updated · live figures update every 4-5 hours
$/Gb divides each build's total cost by its link speed in Gb/s (10 or 25) - the method this site applies to matched price comparisons elsewhere (see the recertified-vs-new methodology), adapted so a higher-speed, pricier link is not read as simply "worse value" for costing more in total.
What actually changes, documented per card
SFP28, the 25G cage, is backward compatible with SFP+, the 10G one - the same relationship 10GBASE-T's multi-rate negotiation has one tier down. A 25G-capable card can run a 10G link with a 10G-rated cable; a 10G-only card cannot run at 25G no matter what cable is attached to it. That asymmetry is worth knowing if a build might grow: buying the 25G-capable card now and running it at 10G today keeps the door open, at a price premium this page states rather than hides.
PCIe requirements shift by generation and SKU rather than by a single 10G-vs-25G rule. The ConnectX-3 (10G) runs PCIe 3.0; ConnectX-5 (this cluster's 25G/100G card) runs PCIe 3.0 x8 or x16 depending on SKU, per NVIDIA's own manual. Check the specific card's own page for its sourced PCIe line rather than assuming one generation always needs more lanes than the other.
Driver generation also moves: the 10GbE cluster's older cards split between mlx4 (ConnectX-3) and ixgbe/i40e (Intel X520/X540/X550/X710); this cluster's ConnectX-5 and ConnectX-4 use mlx5, the actively maintained NVIDIA driver family. Both ConnectX-4 Lx (10G-capable, SFP28-native) and ConnectX-5 (this cluster) now carry NVIDIA's own end-of-life notice, confirmed 2026-09-13 - EOL is a commercial status, not evidence the in-kernel driver has stopped working, and NVIDIA's own separately dated kernel release notes confirm mlx5 still actively supports both lines.
What this page does not answer
Whether the jump from 10G to 25G matters for your specific setup depends on whether your storage can produce or absorb data above 10GbE's ceiling in the first place - a measurement this site has not made for your array and will not assert a number for. The 10GbE hub's own arithmetic section walks through how to check that against your own drives' published sustained-transfer figures; the same logic applies one tier up, just against a higher ceiling.
What sits at the other end of a link like this matters more than the link itself in most home and homelab setups, and that is a hardware question this page deliberately leaves to the pages built to answer it - see the rack disk shelf and NAS device links below for the storage facts this page does not restate.
This page prices both links and states what the vendors document - whether the extra bandwidth is worth it depends entirely on whether the storage or workload on either end of your specific link can produce or absorb data above 10GbE's ceiling, which is a question we don't answer here. See /enclosures/jbod and the NAS device pages for the hardware facts on what a specific array can actually do.
What actually changes going from 10GbE to 25GbE?
The physical layer and the driver generation, documented rather than assumed. SFP28 (25G) is backward compatible with SFP+ (10G) - the same cage, so a 25G card can run a 10G link with a 10G cable. The reverse isn't true: a 10G-only card cannot run at 25G regardless of cabling. PCIe requirements and driver generations also shift - see the per-card pages linked below for the sourced specifics.
Can I mix a 10GbE card and a 25GbE card on the same link?
The link runs at whichever speed both ends negotiate, which in practice means the lower of the two - a 25G SFP28 card talking to a 10G SFP+ card over a 10G-rated cable runs at 10G. You do not get a 25G link unless both ends and the cable all support it.
Do I need a new switch to go from 10GbE to 25GbE?
If you're doing a direct two-machine link, no switch is needed at either speed - see the point-to-point figures on this page. If you already have a 10GbE switch, it will not pass 25G traffic at 25G; the 25GbE hub's switch table has the used-enterprise and current-SMB switches that do.
What is $/Gb of link, and why use it instead of just comparing total cost?
Total build cost alone makes a higher-speed link look artificially expensive, because it is buying more link capacity, not just costing more for the same thing. Dividing each build's total cost by its link speed (in Gb/s) gives a per-Gb figure that can be compared across speed classes honestly - the matched-vs-shelf doctrine this site applies to price comparisons elsewhere, applied to networking. Both figures - total cost and $/Gb - are shown below, since they answer different questions.
Sources
Specification claims on this page were checked against the pages below on 4 August 2026. Where a widely-repeated figure did not survive that check, we report what the source says now and say so in the text rather than repeating the familiar number.
Used for: that ConnectX-3 has reached end-of-life status with NVIDIA, and the MCX311A / MCX312A / MCX314A part-number family. EOL means no new firmware or driver work from the maker; it does not mean the card stops working, and the page says so in those terms.
Used for: that mlx4 (ConnectX-3), mlx5 (ConnectX-4 and later), ixgbe (Intel 82599/X520, X540, X550) and i40e (X710) ship in the mainline Linux kernel. This is why an end-of-life card can still be a sane purchase, and it is a documented fact rather than a forum claim.
Used for: every price on this page, read from our catalogue at request time. Seeded 11 August 2026: 24 network cards across 6 models, 24 DAC cables, 8 transceivers. Every card is split tier 1 (the listing is that part) or tier 2 (a third-party board asserting that silicon).
Used for: We publish no per-port wattage for either cabling path. The relationship - 10GBASE-T draws more power and runs hotter than SFP+ direct attach - is documented at product level and is stated here. The magnitude is not, because we have not fetched and logged a source we can cite for it. Separately: every 'compatible with' phrase shown on this site is stored as a seller claim at tier 2 and badged wherever it renders. We have not tested any module against any switch.
Used for: that NVIDIA's own manuals state "The ConnectX-5 product line has moved to end-of-life" and "The ConnectX-4 Lx product line has moved to end-of-life" in those exact words (connectx5enhw and connectx4lxhw respectively, both dated 2026). The standard ConnectX-4 EN manual (connectx4enhw) carries an "EOL'd Ordering Part Numbers" table for every MCX413A/414A/415A/416A SKU but not the same plain-English banner sentence - stated as that asymmetry rather than smoothed into one claim. Also the per-SKU PCIe generation and lane count (PCIe 3.0 x8 or x16 depending on SKU; PCIe 3.0/4.0 on ConnectX-5 Ex variants) from the same manuals.
Used for: that ConnectX-4, ConnectX-4 Lx and ConnectX-5 all use the mlx5 driver family per NVIDIA's own "Supported Uplinks to Servers" table, dated June 2026 - current documentation, confirming the in-kernel driver is still maintained for hardware NVIDIA itself calls end-of-life for ordering. Also that NVIDIA's current DOCA-Host device list (docs dated September 2026) includes ConnectX-5 and ConnectX-4 Lx but not the standard ConnectX-4 - a real difference in ongoing software-package support between the two, stated as found rather than smoothed over.
Used for: that the i40e driver explicitly lists Intel Ethernet Controller XXV710 as supported hardware alongside X710/XL710/X722 (same page also notes XXV710's speed cannot be forced via ethtool - auto-negotiation only). The companion ice driver page (device_drivers/ethernet/intel/ice.html) is titled for the "800 Series" - the family E810 belongs to - without naming E810 verbatim in the rendered portion, which is stated as the limit of what this specific page confirms rather than stretched into a direct quote. Intel's own ARK product pages did not render in this sourcing pass (see the 100GbE and NIC pages' own note); PCIe generation and EOL status for XXV710 and E810 parts are marked not sourced for that reason.
Used for: the 7050X family's own published figures: "10GbE Typical Power: 1.2W per port", and for the 7050QX-32S specifically "Typical Power Draw: 150W (4.5W per port)" - the only per-port power figures this cluster publishes, because they are the only ones a vendor datasheet actually rendered and stated. Also Arista's own End-of-Sale advisory archive (arista.com/en/support/advisories-notices/endofsale), which carries a dated 2016 end-of-sale notice for "Select Models of the Arista 7050SX, 7050TX and 7050QX Series" and, separately, 2026-dated end-of-sale notices for the newer 7050SX3-48YC12 and 7050SX3-48YC8/48C8 models - a used 7050-family switch may be running silicon its own maker has since end-of-saled, stated with the notice's own date rather than ours.
Used for: every price in this cluster's tables, read from our catalogue at request time. Seeded 13 September 2026: 12 NICs across 3 models (ConnectX-5, ConnectX-4, Intel E810), 40 DAC cables (25G/40G/100G including breakout), 18 transceivers, 11 used-enterprise and current-SMB switches. Two vendor documentation gaps stated plainly rather than filled: NVIDIA/Mellanox's per-model switch datasheets (SN2010/SN2100/SN2410/SN2700) did not render in this sourcing pass beyond a series-level marketing page, so those four switches carry no sourced power, acoustic or EOL figures; Intel's ark.intel.com did not render at all in this pass, so XXV710 and E810 PCIe generation and formal EOL status are marked not sourced rather than assumed from the 700/800-series family pattern.