The Hard Ceiling on Read Speeds
The most persistent myth in handheld gaming circles is that swapping a slow SD card for a high-speed UHS-II model will make games load faster on the Steam Deck. This belief persists because it mirrors the logic of desktop PC storage, where a faster drive directly reduces asset fetch times. However, the Steam Deck’s hardware architecture imposes a strict limit that renders this upgrade pointless for the specific task of loading levels or textures. Valve’s official technical specifications explicitly state that the microSD slot supports UHS-I speeds. This is a physical and protocol-level constraint, not a software limitation that can be patched away.
When you insert a UHS-II card, which theoretically offers double the bandwidth of UHS-I, the Deck’s controller simply cannot speak to it at that higher rate. The interface negotiates down to the highest common denominator, which is UHS-I. Consequently, the theoretical maximum sequential read speed is capped at around 104 MB/s. Once your card exceeds this threshold, the additional silicon and controller logic inside the more expensive card sit idle. You are paying a premium for performance that the host device physically cannot access. For the specific workload of loading a game, where the system reads large chunks of data sequentially, the bottleneck is the bus interface, not the NAND flash inside the card.
This distinction is critical because it separates "speed" from "capacity." Many buyers conflate the two, assuming that a faster card is inherently better for all tasks. In reality, the read speed during a load screen is determined by the slowest link in the chain: the SD reader. Since the reader is fixed at UHS-I, upgrading the card beyond that standard yields zero improvement in load times. The only scenario where a faster card might help is if the internal SSD is completely full and the system is forced to use the SD card for virtual memory or temporary files, but even then, the UHS-I ceiling applies. Therefore, any marketing material suggesting that a UHS-II card will make your Deck "snappier" during gameplay is technically incorrect regarding the hardware constraints.
Where Card Speed Actually Matters
If read speed is capped, why do SD cards have different speed ratings at all? The answer lies in the direction of data flow. While the Deck cannot read faster than UHS-I allows, it also cannot write faster than that same interface permits. However, the nature of the writing workload is different from reading. When you install a game from Steam onto an SD card, the system must write hundreds of gigabytes of data in a sustained, sequential manner. This is where the "sustained write speed" of the card becomes the limiting factor, not the peak read speed.
Most budget SD cards, particularly those marketed for general use or video recording at lower bitrates, struggle with sustained writes. They may advertise a peak speed of 100 MB/s but drop to 20 or 30 MB/s after a few seconds as their internal cache fills up and they begin writing directly to the slower NAND chips. This is known as "cache throttling." For a Steam Deck user, this means installing a 100GB game could take significantly longer if the card’s sustained write performance is poor. A high-quality UHS-I card with a robust cache and efficient controller can maintain speeds closer to the theoretical maximum for the duration of the install, reducing the total time you spend waiting for downloads to complete.
Furthermore, the Steam Deck relies on shader pre-caching and asset optimization. When you launch a game for the first time, or after a significant update, the system may need to write compiled shaders and optimized assets to the storage device. If these files are being written to an SD card, a slow sustained write speed will cause stutters or long freeze times during this process. While this is a one-time or infrequent cost, it is a tangible performance hit that a cheap card exacerbates. The key takeaway is that you are not buying a card for how fast it can give data to the CPU, but how fast it can accept data from the CPU. This shifts the purchasing criteria away from peak read benchmarks and toward sustained write consistency.
The Trap of Peak Speed Benchmarks
Consumer electronics marketing often relies on "peak" performance numbers that are achievable only under ideal, short-duration conditions. An SD card might advertise 150 MB/s reads, but this figure is often obtained during a brief burst before the card’s internal buffer is exhausted. For a Steam Deck user, this number is irrelevant for two reasons. First, the Deck caps reads at UHS-I speeds, so any number above 104 MB/s is unattainable. Second, even if the Deck could read faster, the burst speed would only apply to the first few seconds of a load, after which the speed would drop.
This creates a disconnect between what the box says and what the user experiences. A reader who sees "150 MB/s" on a package and compares it to a "100 MB/s" card assumes a 50% speed increase. In reality, on a Steam Deck, both cards will perform identically at the UHS-I ceiling. The difference between them lies in their reliability, endurance, and sustained write performance. The more expensive card may have a better controller that manages heat and data flow more efficiently, but it will not make a level load 50% faster. It might make an install 20% faster if the cheaper card throttles aggressively, but it will not change the load time.
This is why benchmark videos that show dramatic differences in load times between different SD cards on a Steam Deck are often misleading or, in some cases, fabricated. If the video shows a significant difference in load times, it is likely due to other variables, such as the state of the internal SSD, the specific game’s optimization, or the use of a different device entirely. On a standard Steam Deck setup, the load time from an SD card is determined by the UHS-I interface. Once you have a card that meets the minimum UHS-I standard, you have reached the performance ceiling for reads. Spending more money for a faster read speed is a sunk cost that provides no tangible benefit for the primary use case of playing games.
What to Buy for Which Use Case
Given that read speed is capped and write speed is the variable factor, how should a buyer approach purchasing storage? The answer depends on whether the SD card is your primary storage or a supplementary one. If you are using the SD card as your main library because you have the base 64GB or 256GB model, you need a card that can handle sustained writes without throttling. Look for cards that explicitly advertise "sustained" write speeds or those with high A2 ratings, which are designed for app storage and offer better random I/O performance, though sequential writes are still the primary concern for game installs.
For most users, a high-capacity UHS-I card (1TB or 2TB) is the optimal choice. These cards are widely available, reasonably priced, and meet the UHS-I standard that the Deck can fully utilize. Brands like SanDisk, Samsung, and PNY offer reliable options in this tier. You should avoid "Ultra" or "Extreme" branded cards that push UHS-II speeds, as you are paying for features you cannot use. Instead, focus on capacity. A 1TB UHS-I card will cost less than a 512GB UHS-II card and provide double the space, which is far more valuable for a library of modern games.
If you are using the SD card for backups or transferring files, the sustained write speed becomes even more important. In this scenario, a card with a larger internal cache can significantly reduce the time it takes to copy large files. However, for the average gamer who is simply playing games and installing them, the difference between a mid-range UHS-I card and a high-end UHS-I card is minimal. The real bottleneck is the interface, not the card. Therefore, the best value is found in the most reliable, highest-capacity UHS-I card you can afford. Do not pay extra for speed you cannot access; pay for the space you need.
When the Rule Breaks Down
There are edge cases where the general rule about UHS-I caps does not apply, or where the nuance of the situation changes the recommendation. The first exception is the use of external USB storage. If you connect an external NVMe SSD via the Deck’s USB-C port, you are bypassing the SD slot entirely. In this case, the speed of the external drive is not capped by the UHS-I interface. A high-speed NVMe drive can offer significantly faster load times and install speeds than any SD card. This is the only way to truly "upgrade" the storage speed on a Steam Deck. If load times are a critical concern for you, an external SSD is the superior solution, not a faster SD card.
The second exception is the use of the Deck in desktop mode or for non-gaming tasks. If you are using the Deck as a Linux laptop for development or video editing, the sustained write speed of the SD card becomes much more relevant for tasks like compiling code or rendering video. In this context, a high-performance UHS-I card with excellent sustained writes is beneficial, but a UHS-II card still offers no advantage due to the interface cap. The third exception is future-proofing. If you anticipate Valve releasing a future model of the Deck with a UHS-II slot, buying a UHS-II card now could be a wise investment. However, given the current hardware landscape, this is speculative. For the devices on the market today, the UHS-I ceiling is absolute.
Finally, it is worth noting that the internal SSD on the Steam Deck is always faster than any SD card. The NVMe SSD can reach speeds of several hundred MB/s, far exceeding the UHS-I limit. Therefore, the best practice is to keep your most frequently played games on the internal SSD and use the SD card for your library or backups. This hybrid approach ensures that you get the fastest possible load times for your main games while leveraging the SD card’s capacity for the rest of your collection. The SD card is a capacity solution, not a speed solution. Understanding this distinction saves money and sets realistic expectations for performance.