The SSD that held everything is suddenly not there. The system firmware sees nothing at the port, or it shows a drive with the wrong name and a capacity of 0 bytes or 2 MB. Or the drive appears, Windows offers to initialize it, and every file is gone. Or someone reinstalled the operating system onto the wrong volume and the realization landed about ten minutes too late.

Solid-state drives fail differently from hard drives, and honest expectations matter more here than anywhere else in data recovery. Gillware has recovered data from SSDs for as long as SSDs have been in laptops and servers, and more than two decades in this business has taught us to say plainly when a case will be hard. This page explains what actually goes wrong inside an SSD, why modern encryption and firmware signing make some cases expensive or unrecoverable, what a reformatted SSD looks like once TRIM has run, and what to expect brand by brand from the drives we see most.

How SSDs really fail: firmware and the wear-leveling map

An SSD has no platters to scratch and no heads to crash, so people expect it to fail gracefully. It rarely does. The most common failure we see is not worn-out NAND. It is corruption of the firmware and the internal mapping tables that the controller needs in order to make sense of the NAND at all.

Every SSD keeps a translation layer between the logical block addresses your computer asks for and the physical pages where the data actually lives. That map changes constantly as the controller spreads writes across the flash to even out wear, retires blocks that are wearing out, and garbage-collects pages that have been superseded. The map itself, along with the wear-leveling tables, bad-block lists, and the controller’s own firmware, lives in reserved regions of the same NAND. When any of that structure is damaged by a power loss at the wrong instant, a firmware defect, or a physical fault in the NAND that holds it, the controller boots into a safe mode or does not boot at all. The user data is usually still physically present. The drive simply no longer knows where anything is.

The symptoms are characteristic: the drive disappears from the system entirely; it reports a generic controller name instead of its model; it shows a capacity of 0, 2 MB, or 8 MB; it stays “busy” indefinitely; or it mounts but throws read errors across whole regions. None of these mean the data is gone. All of them mean the drive should be powered off and not experimented on, because every failed boot attempt gives the controller another chance to write to the damaged region.

What recovery involves

Firmware-level recovery means getting the controller into a service mode, reading out the remaining mapping structures, reconstructing what was lost, and reading the user area through the drive’s own translation layer. When the controller cannot be brought up at all, the alternative is reading the NAND chips directly and rebuilding the translation layer from the metadata scattered across them, which requires knowing the controller’s page layout, its error-correction scheme, its scrambling, and how it interleaves data across dies and channels. That is painstaking work, and on modern drives it runs into the wall described next.

Encrypted NAND and signed firmware: why some SSD cases are hard

Almost every SSD sold since roughly 2015 encrypts user data before it is written to NAND. The controller generates a key, holds it in protected hardware, and encrypts and decrypts every page on the fly. This happens whether or not you ever set a password; it is how the drive can implement instant secure-erase and, on models that support it, hardware encryption standards such as TCG Opal.

The consequence for recovery is direct. If the controller cannot be made to work, reading the NAND chips yields ciphertext. Without the key, which exists only inside that specific controller, the raw flash is unrecoverable no matter how much of it is intact. Everything therefore depends on reviving the original controller or, in a narrow set of cases, transplanting the NAND onto a donor board of the identical controller and firmware revision and persuading it to use the original drive’s key material. That is delicate, drive-specific work.

Signed firmware raises the wall further. Manufacturers now cryptographically sign controller firmware so that only their own code will load. The older technique of pushing a diagnostic or loader firmware into a stuck drive to read it out is closed off on any drive that verifies signatures, unless a vendor-specific service pathway exists. Access to those pathways varies by manufacturer, by controller generation, and sometimes by production run.

We are not going to sugarcoat what this means. Some SSD failures that would have been routine recoveries in 2014 are today either expensive, because they require research and tooling specific to one controller family, or not recoverable at any price, because the key is gone with the controller. We tell you which situation you are in after a real evaluation of the drive, not from a phone description. The evaluation is free, the quote is fixed before any work begins, and you owe nothing if we cannot recover your data. What we will not do is quote a low number to win the case and then come back for more.

Reformatted or deleted SSDs: the TRIM problem

The second most common SSD case is not a failure at all. The drive works perfectly. Someone reformatted it, reinstalled the OS on it, deleted a partition, or emptied files they needed. On a hard drive this is often a good-odds recovery, because deleting a file only removes the directory entry and the data sits on the platters until it is overwritten.

SSDs behave differently because of TRIM. When the operating system deletes files or formats a volume, it tells the SSD which blocks no longer hold live data. The controller uses that information to erase those blocks in the background so future writes are fast. TRIM is what keeps an SSD quick over its life, and it is also what makes deleted data on an SSD disappear for real. Once the controller’s garbage collection has processed the trimmed blocks, they read back as zeros. There is no lab technique that recovers data from a block that has been physically erased.

The recovery odds in these cases come down to a race. Windows, macOS, and Linux all issue TRIM on a normal format and on file deletion, and most controllers act on it within seconds to minutes. Data that had not yet been trimmed and erased when the drive lost power is recoverable; data that was is not. Several factors shift the odds in your favor:

  • Power off immediately. The single most important thing. Pull the plug or hold the power button; do not shut down cleanly, because a clean shutdown gives the controller time to finish garbage collection.
  • Do not reinitialize, reformat again, or let the OS “fix” the drive. Each of those issues more TRIM commands.
  • External enclosures help. Many USB-to-SATA and USB-to-NVMe bridges do not pass TRIM through. An SSD that was formatted inside an external enclosure often has far more recoverable data than the same drive formatted while installed internally.
  • A quick format is better than a full one. A full format on modern operating systems writes zeros or issues TRIM across the whole volume; a quick format touches only the file-system structures.

We evaluate reformatted SSDs the same way we evaluate failed ones: read the drive, determine how much of the user area still holds data, reconstruct the file system where possible, and tell you exactly what came back before you commit to anything. If TRIM has already done its work, we say so, and you pay nothing.

The five SSD brands we see most

Between 2022 and 2024 the consumer and small-business SSD market consolidated around a handful of manufacturers, and those are the drives that arrive at our lab most often. None of these brands makes an unreliable product. Each one makes engineering choices that shape what recovery looks like when something goes wrong.

Samsung

Samsung designs its own controllers, its own V-NAND, and its own firmware, and it has implemented always-on hardware encryption across its consumer line for years. Drives such as the 860 and 870 EVO, the 970 EVO Plus, and the 980 and 990 PRO are among the most common we receive. The fully integrated design means Samsung recoveries are almost entirely firmware-level: when the controller can be brought into a service state, results are often excellent, because the drive itself handles the decryption; when it cannot, chip-level reads yield encrypted data. Samsung’s signed firmware limits third-party access to service modes, so the depth of what is possible depends heavily on the controller generation in the specific drive.

Western Digital and SanDisk

Since bringing SanDisk in-house, Western Digital has shipped its own controllers and firmware across the WD Blue, Black, Green, and SN-series NVMe lines, alongside SanDisk-branded SSDs and portable drives built on the same platforms. The WD Blue SA510 and 3D NAND SATA drives, and the SN550, SN570, SN770, and SN850 family of NVMe drives, make up a large share of our intake. Recovery follows the same pattern as any in-house platform: controller-level work first, with the drive doing its own decryption, and a firmware-specific approach for each controller family. Portable SanDisk and WD SSDs add a USB bridge layer that occasionally fails on its own, which is a simpler case than a failed drive inside.

Crucial (Micron)

Crucial is Micron’s consumer brand, so every Crucial SSD carries Micron NAND, paired with either Micron’s own controllers or third-party controllers from established vendors depending on the model and production period. The MX500 has been one of the best-selling SATA SSDs for years, and the P3, P3 Plus, P5, and T500 cover NVMe. Because the controller can differ within a product name across production runs, the first step in any Crucial case is identifying the exact controller and firmware on the board; the recovery path for one revision does not necessarily apply to another. Drives using widely deployed third-party controllers tend to have more mature recovery tooling than fully proprietary designs.

Kingston

Kingston is a large SSD vendor that sources controllers and NAND from multiple suppliers, so two drives with the same model name, such as the A400, NV2, KC3000, or Fury Renegade, can have different controllers and different flash inside. This is normal industry practice and does not make the drives less reliable, but it does mean a Kingston case always begins with opening the drive and reading the components. Many Kingston drives use controllers from vendors whose platforms are well understood, which can work in the customer’s favor; a firmware-corrupted drive on a common controller platform is frequently a recoverable case.

SK hynix and Solidigm

SK hynix became a top-tier consumer SSD brand with the Gold P31 and Platinum P41, and its Solidigm subsidiary, formed from Intel’s NAND and SSD business, sells drives such as the P44 Pro and the enterprise-oriented D-series. These are vertically integrated products: SK hynix NAND, in-house controllers, and signed proprietary firmware. Recovery is firmware-level or nothing, in the same sense as Samsung, and the newer the controller generation, the more the outcome depends on whether a service pathway exists for that specific silicon. Solidigm enterprise drives frequently arrive as part of a larger RAID or server case, where the recovery strategy is usually to work around the failed member rather than through it.

What to do right now

  • If the drive is not detected or shows the wrong capacity: power it down and leave it down. Do not run manufacturer utilities, firmware updates, or “fix” tools against it. Each boot attempt is a write opportunity in the region that is already damaged.
  • If the drive works but the data is gone: pull power immediately, do not reformat or reinitialize, and do not install anything on it.
  • If the SSD is inside a laptop or a Mac with soldered storage, the recovery path runs through the whole machine; send the device, not just the board.
  • Write down what happened and when, especially the moment the drive was last known good. Timing tells us a lot about what is likely still there.

What to expect from Gillware

Every SSD case starts with a free evaluation in our lab in Madison, Wisconsin. We identify the controller and firmware, determine whether the failure is firmware, NAND, controller hardware, or a TRIM race that has already been lost, and tell you what is recoverable and what it will cost as a fixed quote. You approve or decline before any billable work begins, and if we do not recover your data, you pay nothing. Inbound shipping for the evaluation is on us.

We will be direct about the hard cases. Some SSD recoveries take weeks of controller-specific research and cost accordingly. Some cannot be done because the encryption key died with the controller. You will hear that from us after we have examined the drive, in plain terms, with the option to have it returned at no charge. That candor is part of what more than two decades of recovering data has taught us to value, and it is the reason customers send us their hardest cases.

SSD recovery guides

Case studies and guides from our lab on specific SSD failure patterns:

Start your SSD recovery case

Call 877-624-7206 to speak with a recovery specialist, or start a case online and we will send prepaid shipping for a free evaluation.

Start a free SSD evaluation

Related services: hard drive data recovery · flash drive and memory card recovery · RAID and server recovery