The "10 times faster" claim is a peak sequential number
The jump comes from comparing maximum sequential throughput. That figure measures how fast a drive streams large, contiguous chunks of data under favorable conditions. Khurshid's example is a SATA SSD topping out around 550MB/s against a PCIe 5.0 NVMe drive at 14,000MB/s or more.
Those are class-level illustrations, not universal specs. Vendor datasheets list "up to" figures that vary by model and capacity. One current PCIe 5.0 x4 datasheet lists reads up to 14,900MB/s. SanDisk's SATA specs list up to 550/520MB/s read/write. The ratio depends on which two drives you compare. It also only describes a workload that moves very large files without anything else getting in the way.
Everyday Windows use rarely looks like that. Launching an app means reading executables, DLLs, config files and other resources, many of them small. As the article puts it, NVMe SSDs generally perform better than SATA SSDs here as well, but the difference is often much smaller than their sequential speed figures suggest.
Why the gain shrinks: the storage wait is only part of the job
Khurshid's best argument is simple arithmetic, in the spirit of Amdahl's law. Suppose an application takes two seconds to launch from a SATA SSD, and only 200 milliseconds of that time is spent waiting for storage. Remove the storage wait entirely and the launch still takes 1.8 seconds, a 10% improvement. The example is hypothetical, not a measurement of any real app.
The rest of the launch is work storage can't speed up:
- The CPU executes code and initializes processes.
- The app restores sessions and loads extensions.
- Browsers like Chrome also fetch data over the network.
Even if your NVMe SSD finishes its portion of the work much faster, all those other operations still take time.
What other testers report
Independent comparisons point the same way, though they vary in quality. Treat the numbers below as illustrative, not authoritative.
- One comparison site (TechCompare) lists a Photoshop launch of about 3 seconds on SATA against about 2 on NVMe. It lists Windows boot at about 10-12 seconds against 8-10, and a game load at about 12 seconds against 10. Its conclusion is that for OS boot, app launch, and game load, the difference is within a couple seconds.
- A separate NVMe-versus-SATA explainer says that at queue depth 1, NVMe's advantage is measurable but often falls below human perception thresholds for tasks like launching applications. It adds that the gap in app launch times on a fast system is often under half a second. These are secondary sources with no disclosed test setups, so they show the pattern, not a precise result.
- A retail-oriented explainer claims that on random workloads, both drives perform within 10–20% of each other. That is a looser claim, and it likely depends on which drives are compared.
Queue depth is the term to know. It is how many I/O requests are in flight at once. NVMe's design allows 65,535 queues of up to 65,535 commands each, versus SATA's single 32-deep queue. That is protocol capacity, not what a typical desktop task uses. A person opening a few apps rarely generates enough parallel requests to use a fraction of it. A database, a virtual-machine host or a video editor does.
The hard-drive comparison is the real baseline
Moving from a hard drive to any SSD is a much bigger leap than SATA to NVMe, because it removes mechanical delay. A spinning disk must move its head and wait for the right sector to rotate under it. A SATA SSD has none of that. By the time you add NVMe, the largest bottleneck is already gone. Khurshid compares it to going from 5Mbps to 100Mbps internet, which transforms browsing, then from 100Mbps to 1Gbps, which most people barely notice outside big downloads.
Windows already avoids the disk when it can
Windows doesn't read from the drive every time. Microsoft's Learn documentation on the Cache Manager says that, by default, Windows caches file data read from and written to disks. Reads can then come from system memory instead of the physical drive. Frequently read data stays in the cache the longest.
You can see this yourself:
- Restart the PC.
- Launch a large application and note roughly how long it takes.
- Close it and launch it again.
The second launch will often be quicker because the operating system can reuse cached data. This is a general tendency, not a guarantee for every app. When a read is served from RAM, the difference between SATA and NVMe doesn't matter for that read.
This is also why RAM can matter more than a faster SSD. If the system keeps running short of memory, it has to reclaim cached pages or push memory contents to storage, and that is where a faster drive helps. If your working set fits comfortably in RAM, more memory won't help, and neither will a faster SSD. Check Task Manager's Memory and Performance tabs before buying anything.
Where NVMe does pay off
The article doesn't dismiss NVMe. It names the workloads where the difference is real:
- Large file copies between two fast drives. Copying a 50GB video file between two sufficiently fast NVMe drives can be much quicker than the same copy between SATA SSDs.
- Video editing with large or HDR footage, big datasets and several virtual machines.
- Database-style random I/O at higher queue depths. One homelab test using fio at queue depth 32 showed roughly 2.7 times the random read IOPS for an NVMe drive over a SATA drive. That is a synthetic test and not typical desktop use.
Transfers are also limited by the slowest part of the path:
- Copying from a SATA SSD to an NVMe drive is capped near the SATA drive's roughly 550MB/s.
- A 1Gbps network typically tops out around 110 to 120MB/s under favorable conditions, whatever your local drive can do.
- Real throughput also depends on protocol overhead and equipment.
Gaming, DirectStorage and BypassIO
Game loading is a mixed bag. Many games spend load time decompressing assets and compiling shaders, so a much faster drive may shave only a few seconds off a loading screen.
Microsoft's DirectStorage is meant to change that for games that support it. Microsoft's GDK documentation, which covers Xbox consoles, says the goal is to cut operating-system overhead on high volumes of small NVMe requests. The design target is up to 50,000 requests per second using at most 10% of a single CPU core. The game has to be written for it, so it isn't an automatic system-wide speedup.
A related Windows piece is BypassIO. Microsoft's driver documentation describes it as an optimized read path meant to reduce CPU overhead for next-generation game I/O. It is part of the DirectStorage infrastructure and is available starting in Windows 11. Microsoft lists these limits:
- Windows client systems only.
- NVMe storage only.
- The NTFS file system only.
- Non-cached reads only.
Storage drivers must also opt in, or Windows falls back to the traditional I/O path. So the benefit depends on the whole stack, not just the drive you bought.
Should you upgrade?
Buying NVMe for a new build: If your motherboard has an M.2 slot, NVMe is generally the sensible choice. M.2 slots usually connect directly to the board with no extra data or power cables. M.2 is a form factor, not a protocol, so not every M.2 drive is NVMe. Check the listing.
Replacing a healthy SATA SSD: Don't do it just to make Windows boot or Chrome open faster. You probably won't see the improvement you expect. Before buying, work through this list:
- Memory pressure: if Windows regularly runs low on RAM, more memory may help more.
- CPU limits: if your apps are mostly processor-bound, a CPU upgrade may matter more.
- Workload: if you move huge files, edit large video projects, run VMs or work with big datasets, NVMe is more likely to justify the money.
- Spec sheets: note whether a quoted figure is sequential or random, and the queue depth and test conditions. "Up to" numbers describe favorable benchmark conditions, not app-loading speed.
The bottom line
Storage speed only matters when storage is the bottleneck. For most desktop work on a machine with enough RAM, the SATA SSD already cleared the biggest hurdle, and Windows caching clears more of what's left. Independent benchmark roundups put the gap at a second or two for launches and boots, though those are secondary sources, not controlled lab data. NVMe is the right default for new builds and heavy workloads. For a healthy SATA drive in an ordinary desktop, a faster SSD is probably an upgrade you'll see on a benchmark chart more than on the screen.
References
- Your NVMe SSD is 10 times faster than SATA, so why does your PC feel almost the same? How-To Geek · 2026-11-16T22:30:15+00:00
- SATA SSD vs NVMe SSD: Sequential Speed - TechCompare techcompare.app
- NVMe vs SATA SSD: Which Is Faster and Worth It? — Jim Bouton jimbouton.com