SD Card Speed Classes Exposed: Why Your “Fast” Card Isn’t

That gap is what SD speed ratings exist to close. Each symbol on the label, whether C10, U3, V30 or A2, is a promise about the slowest a card will perform under a particular kind of workload. Class 10 guarantees 10 MB/s, U3 guarantees 30 MB/s, and V30, V60 and V90 guarantee 30, 60 and 90 MB/s. A1 and A2 are a different measure, aimed at running apps. What follows is how to read them, and how to choose a card without paying for speed you will never use.

Thank you for reading this post, don't forget to subscribe!

What the wrong speed class actually costs you

SD card speed classes are the reason a drone hovering over a coastline can suddenly stop recording with a card error mid-flight. The card wasn’t broken. It simply couldn’t write data as fast as the camera was sending it.

In plain terms: an undersized card doesn’t just run a little slower, it fails at the exact moment you need it most, and by then the moment is usually gone for good.

• Dropped frames and corrupted files. When sustained write speed falls behind the incoming data, the buffer fills and the camera drops frames, stutters, or stops the recording outright. The clip up to that point is often unrecoverable.
• A recording that never starts. Many cameras check the card’s rating before allowing a high-bitrate mode at all, and simply refuse to record 4K60 or RAW on a card that doesn’t meet the minimum, with no warning until you’re already on location.
• Gaps in footage you needed to keep. A dashcam or security camera with a standard card, rather than a high-endurance one, can wear out early from constant overwriting and silently stop recording, which only becomes obvious after the one incident you needed it for.
• A device that won’t work at all. Some newer hardware makes a class mandatory rather than advisory. The Nintendo Switch 2 is the clearest case: insert a standard microSD card and it will not run games from it, full stop.
• Money spent on the wrong metric. Overpaying is the quieter version of the same mistake. A high V-rated card adds nothing to a phone that only benefits from A2, and a high-A2 card adds nothing to a camera that only benefits from V60. Matching the rating to the workload, not just buying the biggest number, is what actually protects both your footage and your budget.

The number on the front isn’t the one that matters

Packaging loves a big figure, something like 170 MB/s. That is a maximum read speed, and it describes the best case when you copy files from the card to a computer using a fast reader. It says very little about recording, since flash memory reads faster than it writes.

Recording video means writing a constant stream of data without stalling. What decides whether that works is the guaranteed minimum sustained write speed, and every speed class is a statement about that floor. A card rated at 30 MB/s may often write faster, but it promises never to fall below 30 when the workload gets heavy. When choosing a card for recording, ignore the large read-speed figure and look only for the C, U or V symbol; save the big number for comparing how fast finished files transfer to a computer.

Three rating systems, one idea

Over the years the SD Association has introduced three overlapping scales for that minimum write speed, and together they make up what people mean by SD card speed classes. Older cameras understand the older symbols, newer cameras want the newer ones, and many cards carry more than one.

Symbol	Minimum write	What it is usually good for
C2, C4, C6	2, 4, 6 MB/s	Obsolete for modern devices
C10	10 MB/s	Photos and standard HD video
U1	10 MB/s	Full HD video, everyday use
U3	30 MB/s	Typical 4K video
V6 / V10	6 / 10 MB/s	HD video
V30	30 MB/s	4K video
V60	60 MB/s	High-bitrate 4K, 6K and 8K
V90	90 MB/s	Professional 8K, RAW and cinema footage

Note the overlaps: C10, U1 and V10 all promise 10 MB/s, and U3 and V30 both promise 30. The Video Speed Class, introduced in 2016, is the one that stretches the scale far enough for today’s high-resolution cameras.

These labels tell you the floor, not what your camera needs. Bitrates climb quickly with higher frame rates, greater bit depth and RAW formats, and some cameras refuse to record certain modes unless a V60 or V90 card is inserted. The camera manual’s recommendation beats any general rule of thumb.

The bus sets the ceiling

The Roman numeral you sometimes see beside the U symbol, UHS-I, UHS-II or UHS-III, describes the interface between card and device. It caps how fast data can move at all. UHS-I tops out at 104 MB/s in theory, UHS-II at 312 MB/s (half-duplex; full-duplex is 156 MB/s) and UHS-III at 624 MB/s.

UHS-II cards are easy to spot because they have a second row of pins on the back. They only run at full speed in a device and reader that support UHS-II. In a UHS-I slot they still work, but they fall back to UHS-I speeds, which makes an expensive card pointless in the wrong gear.

How to calculate the card speed your footage actually needs

In plain terms: multiply your video’s bitrate by a small safety margin, convert it to megabytes per second, and that is the minimum class to buy. Every camera spec sheet lists a bitrate in megabits per second (Mbps), and the formula to turn that into what a card label promises is simple:

Required write speed (MB/s) = bitrate (Mbps) ÷ 8, then add about 20% headroom.

Dividing by 8 converts bits to bytes, since card ratings are given in megabytes per second while cameras quote megabits. The headroom covers the moments when a scene has more motion or detail and the encoder briefly pushes past its average bitrate.

Recording mode	Typical bitrate	Card speed needed
1080p30, standard	~20 Mbps	Class 10 / U1 comfortably covers it
4K30, H.264	~100 Mbps	U3 / V30 (about 15 MB/s needed, V30 gives headroom)
4K60, 10-bit H.265	~150-200 Mbps	V30 minimum, V60 safer
4K120 or 6K, high bitrate	~400 Mbps	V60
8K or RAW video	600 Mbps and up	V90, UHS-II

Worked example: a camera recording 4K at 60fps with a 200 Mbps bitrate needs 200 ÷ 8 = 25 MB/s, plus headroom, which lands just under the 30 MB/s that a V30 card guarantees. That is why V30 is marketed as the 4K standard, and why a demanding 4K60 mode in a 10-bit codec can tip a V30 card over its limit and push you toward V60. Check your specific camera’s bitrate in its manual rather than assuming from resolution alone, since two cameras shooting the same resolution can differ by a factor of four depending on the codec.

A1 and A2: built for apps, not for cameras

Running an app from a card is a different job from filming. Apps read and write lots of small pieces of data at random, so what counts is IOPS, the number of input/output operations per second. A1 guarantees at least 1,500 random reads and 500 random writes per second. A2 raises that to 4,000 and 2,000. Both also guarantee 10 MB/s of sustained sequential writing.

That makes A-rated cards a good fit for Android phones that use the card as adoptable storage, handheld gaming devices and single-board computers such as the Raspberry Pi. A2 relies on a feature called command queuing, so in an older device that lacks support it may perform no better than A1.

Camera owners: cameras write sequentially, so A1 and A2 make no difference to them. Spend on a better U or V rating instead.

What popular hardware actually requires

In plain terms: match the card to the device you own, not to the highest number on the shelf. Manufacturers publish minimums, and going well beyond them rarely improves anything.

Device	Minimum to look for	Why
Action cameras (GoPro-type)	U3 / V30, higher for high-bitrate 4K+ modes	High frame-rate and slow-motion modes push bitrate up sharply
Consumer drones	U3 / V30	Matches typical 4K flight footage bitrates
Mirrorless cameras, 4K/RAW video	V60 or V90, UHS-II	RAW and high-bitrate 10-bit modes exceed V30’s floor
Nintendo Switch 2	microSD Express only	The console will not run games from a standard UHS card, only transfer media with one
Steam Deck and other handhelds	A2, U3	Game loading is random small-file access, which is what A2 measures
Raspberry Pi (OS and storage)	A1 or A2, U1/U3	Improves boot time and app responsiveness, not just file transfers
Dashcams and security cameras	High-endurance card, U1/U3	Constant loop-recording overwrites wear a standard card out early

The Switch 2 case is worth calling out on its own: Nintendo built the console around microSD Express and dropped support for running games off older UHS-I or UHS-II cards. A standard microSD card still works for copying screenshots or video, but the console will not load a game from it. It is the clearest example yet of a device making a speed class mandatory rather than optional.

SD Express is a different animal

SD Express drops the UHS bus and uses PCIe and NVMe, the same technology as modern SSDs. Cards built to the SD 7.0 specification reach up to about 985 MB/s. SD 8.0 goes further, at roughly 2 GB/s on single-lane cards and up to about 3.9 GB/s on dual-lane cards, which carry three rows of pins. They are usually marked with an EX logo.

The catch is support. Express cards stay backward compatible, so they work in older slots, but only at slower speeds. To get the headline numbers you need a camera or reader designed for them, so check a device’s compatibility list before paying the premium.

Speed and size are separate labels

Capacity has its own naming scheme, and it is easy to mix up with speed. Standard SD cards go up to 2 GB, SDHC covers 2 GB to 32 GB, SDXC covers 32 GB to 2 TB and SDUC stretches from 2 TB to 128 TB. SDHC cards use FAT32, while SDXC and SDUC use exFAT. A device has to support the card type, so an older camera that only knows SDHC will not read an SDXC card.

microSD cards follow exactly the same rules: the same C, U, V and A classes, the same buses and the same capacity types. Only the size changes.

Why real-world speed often disappoints

Even a correctly rated card can underperform for reasons that have nothing to do with its class. Cards write in quick bursts and then slow once their buffer fills, which is exactly why classes measure the sustained floor rather than a peak. A card that is nearly full or heavily fragmented tends to slow down, and reformatting it in the camera usually helps. Long 4K and 8K recording sessions also generate heat that can throttle the card and the camera alike.

What actually determines the ideal speed to choose

In plain terms: the right card is set by four things working together, not by picking the biggest number you can afford. Get any one of these wrong and you either overpay or bottleneck your device.

1. What your device can actually use

A card only performs as fast as the slowest link between it and the device. Check what bus your camera, phone or reader supports (UHS-I, UHS-II or SD Express) before buying, since a V90 card dropped into a UHS-I slot runs at UHS-I speed and none of the extra money is recovered.

2. What workload you are actually feeding it

Video recording is a sustained, sequential write, so it is governed by the C, U or V rating and the bitrate calculation above. Running apps, games or an operating system is random, small-file access, so it is governed by A1 or A2 instead. Buying a high V rating for a Raspberry Pi, or a high A rating for a cinema camera, spends money on the wrong metric.

3. How much headroom the format needs above its stated minimum

A rating is a floor, not a target, and some formats leave very little margin above it. High frame rate, 10-bit color and RAW modes are the ones most likely to spike above a card’s guaranteed minimum for a moment, causing a dropped frame even though the card is technically rated high enough on average. When a mode sits close to a class boundary, it is worth buying one class higher.

4. How long you plan to keep the device

Speed classes exist because past hardware kept demanding more from cards; the next camera or phone generation will likely do the same. Buying one step above today’s stated minimum costs little extra and avoids replacing the card again in a year, whereas buying a device-appropriate card for a device you will replace within months is money better saved.

Rule of thumb: confirm the bus your device supports first, calculate the bitrate or IOPS your workload actually needs second, then buy one class above that minimum if the budget allows it.

Which brands independent testing says actually match their spec sheet

In plain terms: the class symbol tells you what a card should do, but only a track record tells you whether it actually will. Several independent testers have run large batches of cards against their advertised numbers, and a consistent pattern shows up across them.

• SanDisk, particularly the Extreme and Extreme PRO lines, has repeatedly measured closest to its advertised speeds in side-by-side testing, and is the brand most often named by working photographers and videographers as their default choice.
• Kingston has performed consistently well across large-scale independent batch testing covering hundreds of cards from many brands.
• ProGrade Digital, founded by former Lexar and SanDisk engineers, individually tests every card down to the memory chip rather than sampling a batch, which is why it has built a following among working professionals despite being a newer name.
• Lexar had a rocky stretch after its 2017 sale by Micron, but community reports and independent testers now place it back among the trusted picks.
• Also noted for reliable, spec-matching performance in large-scale testing: Samsung, PNY, Delkin, HP and Kioxia.

The pattern on the other side is just as consistent: most spec mismatches, understated capacities and early failures trace back to counterfeit cards carrying a real brand’s logo, sold through unmoderated marketplaces, rather than to genuine cards from an established manufacturer. Buying the right brand only helps if the card is authentic, which is what the next section covers.

A word about fakes

Counterfeit cards are common, especially at prices that look too good. Many report a larger capacity than they really hold, or miss the speeds printed on them. Buy from reputable sellers with a warranty, look closely at print quality and packaging, and test a new card with a free tool such as H2testw on Windows or F3 on Mac and Linux, which check real capacity and speed.

The short version

Read the symbols as three layers. The C, U and V marks are the recording floor, the UHS or Express bus is the ceiling, and A1 or A2 is how well the card handles apps. Once you know how to read SD card speed classes, matching one to your device and workload takes a minute, and it means avoiding both overpaying and dropped footage. When you are ready, you can browse memory cards and card readers at DigitalChoiceHub.