For about five years the answer to this question was so obvious it barely needed asking. Buy an SSD. They were cheap, they were fast, and mechanical drives were for archives and nostalgia.
Then 2026 happened. Datacenter demand squeezed both mechanical drives and flash memory, prices climbed across the board, and a component that used to be a rounding error in a build became a line you have to think about.
So the SSD vs HDD gaming question is genuinely live again — not because mechanical drives got better, but because the price gap that made the decision automatic stopped being automatic.
What each type is actually good at
Stripped of marketing, here's the honest technical picture.
A mechanical drive has a physical arm moving across spinning platters. That makes it decent at reading a long continuous file and poor at reading thousands of small scattered ones, because the arm has to physically travel. Access times are measured in milliseconds.
Flash storage has no moving parts, so it doesn't care where data sits. Access times are measured in microseconds — three orders of magnitude faster.
That gap is why the transition happened. But it matters far more for some workloads than others, and games are not one uniform workload.
The tiers, and where the marketing oversells
| Type | Real-world role in gaming |
|---|---|
| NVMe Gen5 | Enormous sequential numbers. In games, essentially no benefit over Gen4. |
| NVMe Gen4 | The sweet spot. Fast enough for anything a game asks. |
| SATA SSD | Slower on paper, hard to distinguish in play. Excellent value. |
| Mechanical HDD | Cheapest per gigabyte. Fine for games that load once. |
The Gen5 row is the one I want to dwell on, because it's where most storage money gets wasted.
Gen5 drives post spectacular sequential read figures. Games do almost no sequential reading — they do enormous numbers of small random reads, where the difference between a good Gen4 drive and a flagship Gen5 is close to invisible. You're buying a benchmark number, and paying for extra cooling to sustain it.
Storage marketing sells sequential throughput because it's the biggest number available. Games are bound by random access latency, which is the number nobody puts on the box.
The genuinely surprising one is SATA. A SATA SSD is several times slower than NVMe on paper and in actual game loading the difference is frequently a second or two. Both eliminate the mechanical bottleneck; past that, you're into diminishing returns almost immediately.
Where the difference genuinely shows up
Three places, and only three.
1. Initial load times
The obvious one. Going from mechanical to any SSD dramatically cuts the wait to get into a game. Going from SATA to NVMe cuts it a bit more. Going from Gen4 to Gen5 cuts it by an amount you would struggle to notice without a stopwatch.
Diminishing returns arrive early and hit hard.
2. Asset streaming in open worlds
This is the important one and it's underdiscussed.
Modern open-world games load terrain and textures continuously as you move. On a mechanical drive, moving fast can outrun the drive — you get texture pop-in, low-detail geometry, and occasional hitching as the game waits.
This isn't a loading screen you can wait through. It's a visible degradation of the game while you play. For big open worlds and modern engine titles, flash storage isn't a convenience, it's a requirement.
The UE5 Halo remake is a good example — its draw distance is the technical showpiece, and it's exactly the thing a mechanical drive ruins.
3. Fast travel and level transitions
Same mechanism as the first point but more frequent. If a game makes you sit through a transition every few minutes, storage speed changes the texture of the whole session.
Where it makes no difference at all
Frame rate. Storage does not affect frames per second once a game is running. If someone tells you a faster drive will improve your performance in a competitive shooter, they're wrong — that's your CPU and GPU, and I cover which one matters per genre in the build guide.
Also no difference: turn-based games, strategy games, most 2D games, older titles, and anything that loads once and then runs from memory. Fogpiercer plays identically from a mechanical drive. So does most of my strategy library.
The price situation, and why it changed
Quick context, because it drives every recommendation below.
The AI datacenter buildout needs storage at a scale that dwarfs consumer demand. High-capacity mechanical drives got allocated to enterprise buyers on long contracts. Buyers substituted toward flash, pushing demand into NAND, which had its own datacenter pull. And manufacturers, burned by past overcapacity, haven't rushed to expand.
Result: both technologies got more expensive at once. The full background is in the storage shortage piece, but the consequence for this decision is specific — mechanical drives lost most of their price advantage.
That's the thing that changes the recommendation. The old logic was "mechanical is a fraction of the cost per gigabyte, so use it for bulk." When that fraction narrows, the calculation shifts toward flash for more of your library — not because flash got cheap, but because the alternative stopped being cheap enough to justify the compromise.
What I'd actually buy, by budget
Tight budget
One 1TB SATA SSD. Nothing else.
Not 2TB, not NVMe, not a second drive. One terabyte of decent SATA flash holds your system and six to eight modern games, eliminates every mechanical bottleneck, and costs meaningfully less than the NVMe equivalent for a difference you won't perceive.
People overspend here more than anywhere else in a build. The money saved belongs in your processor.
Mid budget
1TB NVMe Gen4, add capacity later.
Gen4 is the sweet spot — fast enough for anything, widely available, sensibly priced relative to Gen5. Start at 1TB and expand when you actually hit the limit rather than buying capacity for a need you're guessing at.
In a market where prices move month to month, buying twelve months of capacity beats buying five years of it. If prices normalise, you didn't overpay for space you weren't using.
Higher budget, or a large library
2TB NVMe Gen4 plus a mechanical drive for bulk.
This is the tiered setup that fell out of fashion when flash got cheap, and it's correct again. Fast storage for the three or four open-world games that genuinely benefit; cheap bulk storage for everything that doesn't.
It takes five minutes of thought when you install a game and saves real money.
Rough test: if you can outrun the asset streaming by moving fast through the world, it needs flash. If the game shows a loading screen and then never loads again, mechanical is fine. Open worlds and modern engine titles go on the fast drive. Strategy, turn-based, 2D and older games do not.
What to look for beyond the headline speed
Four things that matter more than the number on the box.
- DRAM cache. DRAM-less drives are cheaper and slow down noticeably during sustained writes — which is what a large game install is. Worth the small premium.
- Endurance rating (TBW). Any reputable drive far outlasts gaming use, so don't overthink it — but a suspiciously low figure signals corners cut elsewhere.
- Warranty. Five years is standard from serious manufacturers. Less than that tells you something.
- Thermals. Gen5 drives run hot enough to need real cooling. Another reason the sensible answer is Gen4.
And the thing that doesn't matter: sequential read benchmarks. I've said it twice because it's where the money goes.
Moving games between drives without re-downloading
A practical note that saves people a lot of time and bandwidth, because most don't realise it's possible.
If you add a drive later, you don't have to re-download your library. Steam supports multiple library folders and can move an installed game between them from its properties menu — it copies the files locally and updates its records. On a large install that's minutes instead of hours, and it matters even more if your connection is metered or slow.
Most other launchers have some version of this, though the implementations vary in quality. Where one doesn't, the fallback is to copy the game folder to the new drive manually, then point the launcher at it by installing to the same path and letting it verify the existing files rather than re-fetching them.
Two habits worth adopting when you're running a tiered setup:
- Move games as your rotation changes. The open-world game you're currently playing belongs on the fast drive. When you finish it, move it off and bring the next one across.
- Keep saves out of the equation. Cloud saves are independent of where the game is installed, so moving an install doesn't touch your progress. Confirm cloud saves are actually enabled for that title first — several launchers default them off for specific games.
Console storage
Briefly, since it works differently.
Current consoles require fast internal storage for their own titles and won't let you run them from an external mechanical drive. What you can do is keep games on a cheap external drive and copy them across when you want to play — the copy takes a few minutes and saves re-downloading a large install.
Console expansion cards use proprietary or semi-proprietary standards with less competition, so prices move less freely and the shortage hits harder there. If you're planning for a big release like GTA 6 in November, sort storage in advance. Buying in the launch week is how you pay the worst possible price.
Mike's take
Storage still matters, but not in the way the marketing implies. The big jump is mechanical to flash, and it's transformative. Every step after that — SATA to NVMe, Gen4 to Gen5 — delivers rapidly shrinking returns for rapidly growing prices.
The 2026 price squeeze changed one thing: mechanical drives lost most of their cost advantage, which weakens the case for using them as bulk storage. It didn't make Gen5 worth buying.
Where I land: get everything onto flash if you can, at whichever tier your budget reaches — a 1TB SATA drive beats a half-empty Gen5 every time. Keep a mechanical drive only for a genuinely large library of games that don't stream assets. And don't buy sequential benchmark numbers; games never read that way.
The bigger picture
One closing thought about where this goes.
Expensive storage is, perversely, good news for software. Game install sizes ballooned during the years when storage was almost free, and a lot of that bloat is genuinely avoidable — uncompressed audio for a dozen languages you'll never select, duplicated assets kept to reduce seek times on drives most people no longer use.
When storage costs real money again, shipping a 200GB install becomes a competitive disadvantage. Better compression and smarter asset streaming are rational responses, and there are already signs of developers taking install size seriously in a way they hadn't for years.
Constraints have historically been good for software efficiency. We got sloppy when storage was cheap. This particular squeeze is painful for buyers right now and may leave us with leaner games on the other side of it.
In the meantime: don't panic-buy, don't overbuy, and remember that some of the best games of this year run comfortably on a machine you already own.