SSD Not Detected in BIOS: Data Recovery

The first 60 seconds: stop write activity

The most damaging thing you can do to a failing SSD is keep applying power to it. The dead state is safe. Re-detection is the hazard. If the controller comes back to the host even for a moment, the operating system can issue TRIM and the controller can resume garbage collection, and either one finalizes the mapping data a recovery depends on.

Most people respond to a missing drive by rebooting again and again, hoping it shows up. On a healthy drive that's harmless. On a drive that is intermittently failing, every reboot is a chance for the controller to re-enumerate, and that brief window is exactly when TRIM commands and background garbage collection can run.

The result is the same whether it happens once or ten times: the logical-to-physical mappings that point at your files get unmapped or rewritten, and what was recoverable an hour ago no longer is.

The right move takes four steps. Power down the machine fully. Disconnect the SSD so it draws no power. Do not run recovery software against it, do not plug it into a USB enclosure to "check if it still works," and do not let Windows or macOS mount it. Then send it in for evaluation.

A drive that sat untouched after the first failure is the best version of itself a lab will ever see.

Why re-detection, not the dead state, destroys SSD data

TRIM is a host command issued at the ATA or NVMe layer that tells the controller which logical blocks the file system no longer needs. The controller unmaps those blocks from its translation table and returns zeros when they're read, then garbage collection erases the physical cells afterward. On an intermittently failing drive, the dangerous moment is when it re-detects, because that is when the OS can issue TRIM against the very blocks holding your files.

TRIM is a logical deallocate, not an instant physical wipe. When the operating system decides a block is free (after a delete, a format, or a move), it sends a TRIM command. The controller marks that logical block as unmapped in its Flash Translation Layer. From that point on, reading the block returns deterministic zeros, because the controller no longer points it at any physical page. Garbage collection then erases the underlying cells in the background to make them ready for new writes.

The unmap is what ends recoverability. Once the controller stops pointing a logical block at its data, no lab can ask it to point back.

This is why the power-cycling habit is so destructive. A drive sitting unpowered cannot run TRIM or garbage collection. A drive that re-enumerates can.

Every reboot that briefly brings the controller back to the host is an opportunity for the OS to flush queued TRIM commands and for the controller to resume the housekeeping it paused when it failed. The drive does not have to stay alive long. A few seconds of re-detection is enough to finalize unmaps that were pending.

On modern Windows (7 and later) and macOS (10.6.8 and later), TRIM is on by default, so this is the normal behavior, not an edge case. Recovery of deleted or moved files from an SSD is only possible when TRIM never executed against those blocks: the drive was pulled immediately, TRIM was disabled, or the file system doesn't support it. For a drive that won't enumerate at all, keeping it powered off protects whatever unmaps were still pending when it died.

Why a dead NVMe SSD often needs board repair before it can be read

An NVMe SSD that draws no power, or that the motherboard never lists at all, frequently has an electrical fault rather than a firmware one. NVMe drives rely on PCIe Gen3, Gen4, or Gen5 lanes, and the controller cannot complete PCIe link training until its power rails are clean. A dead PMIC or voltage regulator stops the controller before it ever gets to enumerate.

An NVMe controller has to negotiate the PCIe link with the host before it presents itself as a storage device. That negotiation, link training, needs the controller fully powered. The board steps the incoming rail down to the controller's core voltages through a power-management IC and discrete regulators.

If one of those is shorted, often from an over-voltage event or a failed coupling capacitor, the controller never powers up and link training never starts. To the motherboard, the slot is empty.

The fix is board-level microsoldering, not software. We locate the fault with FLIR thermal imaging: a shorted PMIC or regulator shows up as a hot spot within seconds of applying bench power. The failed component is replaced with a Hakko FM-2032 on an FX-951 base station, and where a controller or other BGA package needs reflow, a Zhuo Mao precision BGA rework station handles the thermal profile. Once the rails are clean, the controller completes link training, enumerates, and PC-3000 SSD can interface with it.

Reviving the original controller matters for one more reason. On modern always-on AES NVMe drives, the decryption key lives inside the original controller. Board repair that brings that controller back is not a side service; for an encrypted drive it is the recovery. NVMe board-level repair is $600–$900. NVMe firmware recovery, when the board is healthy but the controller refuses to mount its translator, is $900–$1,200.

Failure modes mapped to the right intervention

A drive that won't enumerate falls into one of four states, and each maps to a specific intervention. The triage at intake is reading what the bus does when power is applied, then choosing loader injection, board repair, or declining a case that can't be recovered.

ROM mode: factory string at 0GB / 1GB / 2MB

The controller failed to validate its on-NAND firmware and fell back to its bootloader. Intervention: PC-3000 SSD loader injection into the controller's volatile memory, then FTL reconstruction. A Phison PS3111-S11 reports SATAFIRM S11 here. Priced in the $600–$900 SATA firmware tier.

ATA BSY hang: BIOS freezes mid-POST

The controller completed the handshake, then hit an unhandled exception parsing corrupted system blocks and is holding the busy bit high. A Silicon Motion SM2259XT is a common offender. Intervention: a sustained test-pad short through initialization so the controller never loads the bad firmware, then loader injection.

No power, no enumeration: dead PMIC or voltage rail

The board never brings the controller up, so nothing answers the host. Intervention: board-level microsoldering. FLIR thermal imaging locates the short, and the failed PMIC, regulator, or capacitor is replaced. Priced in the $450–$600 SATA or $600–$900 NVMe board-repair tier.

Encrypted controller, dead board: chip-off limits

On a modern always-on AES NVMe controller, the AES-256 Media Encryption Key is bound to the original controller silicon, so desoldering the NAND yields ciphertext. Chip-off can't recover it. The only viable path is board-level microsoldering to revive the original controller so it decrypts its own NAND. We do not claim to defeat manufacturer encryption.

When desoldering the NAND is on the table at all, the limits are covered in detail on our chip-off NAND recovery page.

Frequently asked questions

Why does my SSD not appear in BIOS anymore?

An SSD vanishes from BIOS or UEFI for one of three hardware reasons, not a software one. The controller firmware panicked and the chip dropped into ROM mode (often with a factory string and a 0GB capacity), the controller hit an unhandled exception and is holding the bus busy so the BIOS hangs, or a voltage rail on the board is dead so the controller never powers up to answer the host. In every case the data is still on the NAND. What failed is the controller's ability to present itself to the host. Software can't see a drive that never enumerates at the protocol layer.

Can I recover data from an SSD that won't show up in BIOS at all?

Yes, when the drive is a board-repair or firmware-recovery candidate. A drive that draws power but never answers the host usually has a dead PMIC or voltage regulator (board-level microsoldering territory) or a corrupted Flash Translation Layer the controller refuses to mount (PC-3000 SSD loader-injection territory). Both paths revive the original controller, or interface with it below its dead firmware, so the NAND can be read. Recovery is not possible if TRIM already ran on the deleted blocks, if the NAND itself is worn past the ECC correction threshold across the service area, or if the controller binds an always-on AES key that can't be revived.

Should I keep power-cycling an SSD that disappeared from BIOS?

No. Repeated power cycling is the single most damaging thing you can do to an already-failing SSD. The danger is not the dead state. The danger is the moment the drive momentarily re-detects. If the controller comes back to the host even briefly, the operating system can issue TRIM and the controller can resume garbage collection, both of which finalize the logical-to-physical mappings the recovery depends on. Pull the drive, stop applying power, and send it in.

Why does my SSD show up as 0GB or with a strange factory name?

A 0GB, 1GB, or 2MB capacity paired with a generic name is ROM mode. The controller failed to validate its on-NAND firmware during boot, fell back to its read-only bootloader, and reported a diagnostic identity instead of the real model. A Phison PS3111-S11 reports "SATAFIRM S11." A Silicon Motion controller reports a raw silicon alias such as SM2258AB or holds the bus busy with no string at all. The full-capacity model name is gone because the firmware that knew it is the thing that failed.

Can Disk Drill, EaseUS, or R-Studio recover an SSD missing from BIOS?

No. Recovery software runs above the operating system storage driver and needs the controller to present a valid block device with a real capacity. A drive that never enumerates in BIOS exposes no addressable logical blocks, so there is nothing for the software to scan. Those tools work well for logical failures on a physically healthy drive (deleted files where TRIM never ran, a corrupted partition table, a formatted volume). They cannot interface with a controller that won't answer the host at the protocol layer.

How does PC-3000 SSD read a drive the BIOS can't see?

PC-3000 SSD does not go through the operating system storage driver. It connects to the drive over a native SATA or NVMe port with vendor-command access, coordinates entry into the controller's diagnostic mode (a test-pad short on the ROM_CS or R/B line during power-on), and injects a volatile microcode loader matched to the silicon revision and NAND geometry into the controller's volatile memory. That loader bypasses the dead on-NAND firmware and gives the workstation raw NAND read access. The Flash Translation Layer is then reconstructed in workstation RAM and the data is imaged read-only, sector by sector, to a destination drive.

My SSD made the BIOS freeze at POST. What does that mean?

A BIOS that hangs mid-POST when the drive is connected, then boots normally once the drive is unplugged, is the classic ATA BSY signature. A controller such as the SM2258XT completed the SATA handshake, then hit an unhandled exception parsing corrupted system blocks and is holding the ATA busy bit high forever. The host waits on a command that never returns. This is a firmware panic, not a dead board. Diagnostic access requires shorting the controller's test pad through initialization so it never tries to load the corrupted firmware.

How much does it cost to recover an SSD not detected in BIOS?

It depends on which of the three failure families the drive falls into. Firmware panic and ROM-mode recovery is $600–$900 for SATA and $900–$1,200 for NVMe. Board-level repair on a dead PMIC or voltage rail is $450–$600 for SATA and $600–$900 for NVMe. Every case starts with a free evaluation and a firm quote before any paid work. No diagnostic fees. If we recover nothing, you pay nothing. Rush service is +$100 rush fee to move to the front of the queue.

Why can't chip-off recover a modern NVMe SSD that won't power on?

Modern always-on AES NVMe controllers generate an AES-256 Media Encryption Key on the controller and bind it to a hardware-unique root inside the controller silicon. The key never leaves the original controller and is never written to the NAND in plaintext. Desoldering the memory chips and reading them in a NAND reader yields ciphertext that no donor controller can decrypt. For these drives the only viable path is board-level microsoldering to revive the original controller so it can decrypt its own NAND. We do not claim to crack or defeat manufacturer encryption.