PCIe lanes explained: Bandwidths, slots, and uses in 2025

PCIe lanes are data channels within a PCIe slot, which is used for transmitting and receiving data between the motherboard and the component. There are two sets of wires which is used to transmit data in both directions, and the amount of lanes you have available relates to the different slots on the motherboard.

There are four different PCIe slots that you’ll find on your motherboard. These are PCIe x1, x2, x4, x8, and x16; the more lanes you have available on a slot, the more data can be transmitted and received. The smallest PCIe slot (PCIe x1) has a single lane available and can be used for Wi-Fi cards, LAN cards, Bluetooth, and more.

Along the same lines, PCIe x4 slots have a total of four PCIe lanes available, which offers four times the bandwidth and is commonly used for things such as NVMe SSDs, expansion cards, RAID controllers, and internal capture cards, among other things. PCIe x8 doubles the available lanes again (with more bandwidth) for things such as high-performance network adapters, storage controllers, and AI accelerator cards.

PCIe x16 ports have 16 PCIe lanes and the most bandwidth available, and they are primarily used by your graphics card, which will need the fastest data transmissions possible. PCIe lanes are dictated by your processor, so you’ll be beholden to how many lanes can be used at a time. For ideal GPU performance, you’re going to need one of the best processors that has at least 16 lanes to allocate to the slot. That’s before allocating resources towards an NVMe SSD and other aforementioned components that share the pool.

Even entry-level graphics cards in 2025 have more than enough PCIe lanes to have your graphics card running at its best, with additional allocations for SSDs and other internal components on the board. For example, the Intel Core Ultra 5 245K (built on the LGA 1851 socket) and AM5 budget options like the AMD Ryzen 5 7600X have 24 usable lanes apiece. It’s similar to previous-generation LGA 1700 (Alder Lake and Raptor Lake) and AM4 processors as well, with anything older struggling.

Now that you understand what PCIe lanes are and the different slots you can use them on, we can detail the different bandwidths afforded by the two most recent generations. Starting with the previous-generation, PCIe 4.0, which debuted in 2017 and is widely in use today.

With the previous standard, a PCIe x1 port has a total bandwidth of 2GB/s, PCIe x4 has 8GB/s, PCIe x8 uses 16GB/s, and PCIe x16 tops out at 32GB/s. In brief, PCIe 5.0, first released in 2019 (but adopted in the early 2020s), doubles things. That means the PCIe 1x slot has a 4GB/s bandwidth, 4x has 16GB/s available, 8x is 32GB/s, and 16x utilizes 64GB/s for today’s graphics cards.

Because of the larger bandwidths available across the PCIe lanes, PCIe 5.0 components can be much faster than their PCIe 4.0 counterparts. We can see this evidenced easily by M.2 PCIe 5.0 NVMe SSDs (Gen 5), with many of the best M.2 SSDs now topping out at around 14,000 MB/s sequential read and write compared to the maximum of 8,000 MB/s sequential of PCIe 4.0 (or Gen 4) models. You can notice this in the jump between the Seagate FireCuda 540 and the Seagate FireCuda 530, though both have since been surpassed by later options.

It’s more apparent with the latest graphics cards. While PCIe 5.0 NVMe SSDs arrived as early as 2022, it wouldn’t be until January 30, 2025, that GPUs would begin to utilize the bandwidth available. The RTX 5090 and RTX 5080 lead the charge first, with the RTX 5070 Ti and RTX 5070 following in February and March, respectively. Similarly, AMD’s RDNA 4 generation also used PCIe 5.0 x16 ports with its RX 9070 and RX 9070 XT.

It makes the most sense to look at the two highest-end options for how they utilize the available bandwidth from 16 PCIe lanes. The RTX 5090 has 32GB GDDR7 memory with a speed of 28 Gbps and a total bandwidth of 1,792 GB/s on a 512-bit memory bus. In contrast, the RTX 4090 uses 24GB GDDR6X memory with 21 Gbps speed and a total bandwidth of 1,010 GB/s (or 1TB/s) with a 384-bit memory bus. That’s a difference of 77% going from PCIe 4.0 x16 to PCIe 5.0 x16, despite them both using 16 PCIe lanes. The data allocation isn’t the only defining factor in the bandwidth differences, but it shows how the two utilize the same slot, the same lanes and are similarly powered.

As touched upon above, a PCIe card (commonly called an expansion card) is any component that uses any of the four available PCIe slots on your motherboard. This (traditionally) now means network cards and wireless modules, such as Wi-Fi 6E and Wi-Fi 7 cards, as well as Bluetooth 5.4 adapters, which can give you the functionality that your motherboard may not have shipped with. They are usually small and inexpensive but can be useful and easy to install.

You can also get what’s known as PCIe add-in cards, which can expand your computer’s I/O options. Popular examples include the Asus Thunderbolt 4 (which is a PCIe x4 slot card), which includes 100W power delivery and charging with two USB-C and Mini DisplayPort options. Add-in cards also exist for additional M.2 2280 SSDs on the PCIe x8 port, which can give you additional storage if you’ve run out of room to directly connect them to your motherboard. There are a lot of things you can do with the additional PCIe lanes available on your computer, provided your CPU has the allocation; otherwise, things will not work as intended.

OIder processors may only have 20 PCIe lanes available in total, which is enough for your graphics card, but you will be unlikely to have anything else connected on the motherboard running at its intended speed. Most modern processors will have 24-28 usable lanes as standard, with higher-end models increasing this number.

Both PCIe 4.0 and PCIe 5.0 will have the same number of lanes available by the same conventional ports on the motherboard. However, PCIe 5.0 is twice as fast as PCIe 4.0, with double the bandwidth across the board.