Hubbry Logo
M.2M.2Main
Open search
M.2
Community hub
M.2
logo
8 pages, 0 posts
0 subscribers
Be the first to start a discussion here.
Be the first to start a discussion here.
M.2
M.2
M.2
View on Wikipedia
from Wikipedia

M.2
An M.2 2280 solid-state drive (SSD), 22 mm wide and 80 mm long with the key in the M position
Connects toMotherboard via one of:
Common manufacturers
Design firmPCI-SIG
IntroducedNovember 1, 2013; 11 years ago (2013-11-01)
Dimensions
  • Width: 22 mm (0.87 in)
  • Lengths: 30, 42, 60, 80 or 110 mm (1.2, 1.7, 2.4, 3.1 or 4.3 in)
A size comparison of an mSATA SSD (left) and an M.2 2242 SSD (right)

M.2 (pronounced "M-dot-2"),[1] formerly known as the Next Generation Form Factor (NGFF), is a specification for internally mounted computer expansion cards and connectors. It was developed to replace the older Mini SATA (mSATA) and Mini PCIe (mPCIe) standards.

M.2 supports a variety of module sizes and interface types, offering greater flexibility for modern devices. It is widely used in compact systems such as ultrabooks and tablet computers, particularly for solid-state drives (SSDs), due to its smaller size and higher performance compared to mSATA.[2][3][4]

The M.2 connector can provide multiple interface options, including up to four lanes of PCI Express, as well as Serial ATA 3.0 and USB 3.0. The supported interfaces vary depending on the device and host implementation. M.2 modules and slots use different "keying" notches to indicate supported interfaces and to prevent incompatible installations.[2][3][5]

For storage devices, M.2 supports both the older Advanced Host Controller Interface (AHCI) and the newer NVM Express (NVMe) protocols. AHCI provides compatibility with legacy SATA-based systems and operating systems, while NVMe is designed for high-speed SSDs and allows for much faster performance by supporting multiple simultaneous I/O operations.[2]: 14 [6]

Features

[edit]
High-level software architecture for SATA Express (also used by M.2),[2]: 14  supporting SATA and PCIe devices via AHCI or NVMe.[6]: 4 

M.2 modules can integrate multiple functions, including the following device classes: Wi-Fi, Bluetooth, satellite navigation, near-field communication (NFC), digital radio, WiGig, wireless WAN (WWAN), and solid-state drives (SSDs).[7] The SATA revision 3.2 specification, in its gold revision as of August 2013, standardizes M.2 as a new format for storage devices and specifies its hardware layout.[2]: 12 [8] Buses exposed through the M.2 connector include PCI Express (PCIe) 3.0 and newer, Serial ATA (SATA) 3.0 and USB 3.0; all these standards are backward compatible.

The M.2 specification provides up to four PCI Express lanes and one logical SATA 3.0 (6 Gbit/s) port, and exposes them through the same connector so both PCI Express and SATA storage devices may exist in the form of M.2 modules. Exposed PCI Express lanes provide a pure PCI Express connection between the host and storage device, with no additional layers of bus abstraction.[9] PCI-SIG M.2 specification, in its revision 1.0 as of December 2013, provides detailed M.2 specifications.[2]: 12 [10]

Storage interfaces

[edit]

Three options are available for the logical device interfaces and command sets used for interfacing with M.2 storage devices, which may be used depending on the type of M.2 storage device and available operating system support:[2]: 14 [6][9]

Legacy SATA
Used for SATA SSDs, and interfaced through the AHCI driver and legacy SATA 3.0 (6 Gbit/s) port exposed through the M.2 connector.
PCI Express using AHCI
Used for PCI Express SSDs and interfaced through the AHCI driver and provided PCI Express lanes, providing backward compatibility with widespread SATA support in operating systems at the cost of lower performance. AHCI was developed when the purpose of a host bus adapter (HBA) in a system was to connect the CPU/memory subsystem with a much slower storage subsystem based on rotating magnetic media; as a result, AHCI has some inherent inefficiencies when applied to SSD devices, which behave much more like RAM than like spinning media.
PCI Express using NVMe
Used for PCI Express SSDs and interfaced through the NVMe driver and provided PCI Express lanes, as a high-performance and scalable host controller interface designed and optimized especially for interfacing with PCI Express SSDs. NVMe has been designed from the ground up, capitalizing on the low latency and enhanced parallelism of PCI Express SSDs, and complementing the parallelism of contemporary CPUs, platforms and applications. At a high level, primary advantages of NVMe over AHCI relate to NVMe's ability to exploit parallelism in host hardware and software, based on its design advantages that include data transfers with fewer stages, greater depth of command queues, and more efficient interrupt processing.

Form factors and keying

[edit]
M.2 keying notches in B and M positions, with visible pin offset on each side of the module.[11]
Size examples of M.2 SSDs. The first two digits indicate width, the rest length in millimeters (e.g., 22110 = 22 mm wide × 110 mm long). Not all devices support every size.[12]

The M.2 standard is based on the mSATA standard, which uses the existing PCI Express Mini Card (Mini PCIe) form factor and connector. M.2 adds the possibility of larger printed circuit boards (PCBs), allowing longer modules and double-sided component population. Consequently, M.2 SSD modules can provide double the storage capacity within the footprint of an mSATA device.[2]: 20, 22–23 [4][13]

M.2 modules are rectangular, with an edge connector on one side and a semicircular mounting hole at the center of the opposite edge. The edge connector has 75 positions with up to 67 pins, employing a 0.5 mm pitch and offsetting the pins on opposing sides of the PCB from each other. Each pin on the connector is rated for up to 50 V and 0.5 A, while the connector itself is specified to endure 60 mating cycles.[14]: 6  However, many M.2 slots (Socket 1, 2 and 3) found on motherboards only provide up to 3.3 V power.[15][16][17]

The M.2 standard allows module widths of 12, 16, 22 and 30 mm, and lengths of 16, 26, 30, 38, 42, 60, 80 and 110 mm. Initial line-up of the commercially available M.2 expansion cards is 22 mm wide, with varying lengths of 30, 42, 60, 80 and 110 mm.[3][5][14][18] The codes for the M.2 module sizes contain both the width and length of a particular module; for example, "2242" as a module code means that the module is 22 mm wide and 42 mm long, while "2280" denotes a module 22 mm wide and 80 mm long.

An M.2 module is installed into a mating connector provided by the host's circuit board, and a single mounting screw secures the module into place. Components may be mounted on either side of the module, with the actual module type limiting how thick the components can be; the maximum allowable thickness of components is 1.5 mm per side, and the thickness of the PCB is 0.8 mm ± 10%.[10] Different host-side connectors are used for single- and double-sided M.2 modules, providing different amounts of space between the M.2 expansion card and the host's PCB.[4][5][14] Circuit boards on the hosts are usually designed to accept multiple lengths of M.2 modules, which means that the sockets capable of accepting longer M.2 modules usually also accept shorter ones by providing different positions for the mounting screw.[19][20]

M.2 module keying and provided interfaces[5]: 8 [14]: 3 [21][22][23]
Key
ID 		Notched
pins 		Provided interfaces Dimensions Uses
A
(Socket 1) 		8–15 2 × PCIe ×1, USB 2.0, I2C and DP ×4 1630, 2230, 3030 Wi-Fi, WWAN, GPS, Bluetooth, NFC
B
(Socket 2) 		12–19 SATA, PCIe ×2, USB 2.0 and 3.0, audio, UIM, HSIC, SSIC, I2C and SMBus 2230, 2242, 2260, 2280, 22110 SSD
C 16–23 Reserved for future use
D 20–27
E
(Socket 1) 		24–31 2 × PCIe ×1, USB 2.0, I2C, SDIO, UART, PCM and CNVi 1630, 2230, 3030 Wi-Fi, WWAN, GPS, Bluetooth, NFC
A+E
(Socket 1) 		8–15 and 24–31 2 × PCIe ×1, USB 2.0 and CNVi 1630, 2230, 3030 Wi-Fi, WWAN, GPS, Bluetooth, NFC
F 28–35 Future Memory Interface (FMI)
G 39–46 Reserved for custom use (unused in the M.2 specification)
H 43–50 Reserved for future use
J 47–54
K 51–58
L 55–62
M
(Socket 3) 		59–66 SATA, PCIe ×4, and SMBus 2230, 2242, 2260, 2280, 22110 SSD
B+M
(Socket 2) 		12–19 and 59–66 SATA, PCIe ×2, and SMBus 2230, 2242, 2260, 2280, 22110 SSD
Maximum component thickness on M.2 modules[5]: 8 [14]: 3 
Type
ID 		Top
side 		Bottom
side
S1 1.20 mm —
S2 1.35 mm —
S3 1.50 mm —
D1 1.20 mm 1.35 mm
D2 1.35 mm 1.35 mm
D3 1.50 mm 1.35 mm
D4 1.50 mm 0.70 mm
D5 1.50 mm 1.50 mm
An M.2 socket on a motherboard, visible in the upper-left portion of the picture. The socket is keyed in the M position and provides two positions for the mounting screw, accepting 2260 and 2280 sizes of M.2 modules.

The PCB of an M.2 module provides a 75-position edge connector; depending on the type of module, certain pin positions are removed to present one or more keying notches. Host-side M.2 connectors (sockets) may populate one or more mating key positions, determining the type of modules accepted by the host; as of April 2014, host-side connectors are available with only one mating key position populated (either B or M).[5][14][11] Furthermore, M.2 sockets keyed for SATA or two PCI Express lanes (PCIe ×2) are referred to as "socket 2 configuration" or "socket 2", while the sockets keyed for four PCI Express lanes (PCIe ×4) are referred to as "socket 3 configuration" or "socket 3".[2]: 15 [24]

For example, M.2 modules with two notches in B and M positions use up to two PCI Express lanes and provide broader compatibility at the same time, while the M.2 modules with only one notch in the M position use up to four PCI Express lanes; both examples may also provide SATA storage devices. Similar keying applies to M.2 modules that utilize provided USB 3.0 connectivity.[5][11][25]

Various types of M.2 modules are denoted using the "WWLL-HH-K-K" or "WWLL-HH-K" naming schemes, in which "WW" and "LL" specify the module width and length in millimeters, respectively. The "HH" part specifies, in an encoded form, whether a module is single- or double-sided, and the maximum allowed thickness of mounted components; possible values are listed in the right table above. Module keying is specified by the "K-K" part, in an encoded form using the key IDs from the left table above; it can also be specified as "K" only, if a module has only one keying notch.[5][14]

Beside socketed modules, the M.2 standard also includes the option for having permanently soldered single-sided modules.[14]

Alternative standards

[edit]

NGSFF

[edit]

In 2017, Samsung introduced a new form factor called Next Generation Small Form Factor (NGSFF), also known as NF1 or M.3, which may replace U.2 in server applications.[26] The NGSFF connector is electrically and dimensionally compatible with M.2 (revision 1.1)'s connector; new functionality is achieved through previously unused (N/C) pins.[27] The main changes compared to M.2 are:

  • The width (or "height") of the SSD is increased from 22 mm to 30.5 mm; the thickness is increased from 3.88 mm to 4.38 mm. These changes allow more NAND chips to be fitted onto an SSD while still fitting inside a rack unit.[28]
  • New pins for 12 V power. Devices are supposed to mainly use 12 V power instead of the old 3.3 V, which has been made optional.[28]
  • Ability to run two PCIe ports (each with two lanes) on one NGSFF port.[27]
  • Features for rackmount servers: hotswap support, indicator LEDs, SSD tray (with new screw holes).[27]

In 2018, the PCI-SIG issued a warning that NGSFF's new pin usage clashes with the pin usage in the upcoming 1.2 revision of the M.2 standard. The new revision uses some of the previously non-connected (N/C) pins to deliver 1.8 V power and USB 2.0 data on the "M" socket. Samsung has sought to standardize its NGSFF/NF1 through JEDEC, but the process appears to have stalled.[29]

XFM

[edit]

JEDEC JESD233 is another specification called Crossover Flash Memory (XFM) for XFM Embedded and Removable Memory Devices (XFMD). It targets to replace the M.2 form factor with a significantly smaller one (also called XT2), so that it can also be designed as an alternative to soldered memory. XFM Express utilizes a NVMe logical interface over a PCI Express physical interface.[30][31]

[edit]
  • An M.2 2242 SSD connected into a USB 3.0 adapter and connected to a computer
    An M.2 2242 SSD connected into a USB 3.0 adapter and connected to a computer
  • A docking station for M.2 modules
    A docking station for M.2 modules
  • The connection slot of the docking station
    The connection slot of the docking station
  • M.2 2280 Samsung 980 PRO PCIe 4.0 NVMe SSD with 1 TB of storage capacity
    M.2 2280 Samsung 980 PRO PCIe 4.0 NVMe SSD with 1 TB of storage capacity
  • M.2 2230 (left) and M.2 1630 WiFi cards
    M.2 2230 (left) and M.2 1630 WiFi cards

See also

[edit]

Notes

[edit]

References

[edit]
[edit]
Revisions and contributorsEdit on WikipediaRead on Wikipedia

M.2

View on Grokipedia
from Grokipedia
M.2 is a compact form factor specification for internally mounted computer expansion cards and associated edge connectors, designed primarily for mobile, ultrathin, and embedded computing platforms. Developed and maintained by the PCI Special Interest Group (PCI-SIG), it enables the integration of functions such as solid-state drives (SSDs), wireless modules, and other peripherals onto small modules that support high-speed interfaces like PCI Express (PCIe). Originally developed starting in and released in as a successor to the Mini PCI Express and Half-Mini Card form factors, M.2—formerly known as the Next Generation Form Factor (NGFF)—provides a versatile, scalable design with the smallest footprint among PCIe connectors. The specification supports module widths of 12 mm, 16 mm, 22 mm, and 30 mm, with lengths ranging from 16 mm to 110 mm (common designations include 2230, 2242, 2260, 2280, and 22110), accommodating single-sided or double-sided configurations for varying power and thermal requirements. Power delivery options include 3.3 V and 1.8 V via dedicated pins, with 14 vendor-defined pins available for customization. The M.2 connector features keying notches (such as A, B, E, and M keys) on both the module and socket to prevent incompatible pairings and ensure proper signal routing, supporting up to four PCIe lanes for premium applications like SSDs, alongside compatibility for USB, serial ATA (SATA), and other protocols depending on the key type and host implementation. Applications span consumer electronics like laptops, tablets, and smartphones, to industrial and enterprise systems for wireless connectivity (Wi-Fi, Bluetooth, NFC, WWAN), storage, and I/O expansion. The latest revision, PCI Express M.2 Specification Revision 5.1 (as of May 2024), emphasizes interoperability, low power consumption, and forward compatibility with evolving PCIe generations including up to 6.0.

Overview and History

Definition and Purpose

M.2 is a registered trademark of PCI-SIG and refers to a compact form factor standard for expansion cards, as defined in the PCI Express M.2 Specification Revision 5.1 (with errata dated November 5, 2024). This specification outlines a versatile module design intended primarily for mobile adapters, enabling the integration of multiple functions such as storage and connectivity into slim computing platforms like laptops, tablets, desktops, and embedded systems. The primary purpose of M.2 is to provide a unified edge connector that accommodates diverse interfaces, including (PCIe), Serial ATA (SATA), and USB, thereby supporting devices like solid-state drives (SSDs), modules, and other peripherals without requiring separate connectors. Originally developed as the Next Generation Form Factor (NGFF) to succeed earlier standards like mSATA and Mini Card, M.2 offers a smaller physical footprint and greater flexibility for high-density integration in space-constrained environments. Key benefits include its reduced size compared to mSATA, which allows for thinner device profiles while supporting higher bandwidth capabilities—up to four PCIe lanes operating at 32 GT/s each under PCIe 5.0 for aggregate signaling rates of 128 GT/s. M.2 modules are not designed for hot-plugging and require the system to be powered off for safe insertion and removal to avoid potential damage. This design promotes scalability and efficiency, making it a foundational standard for modern high-performance, compact computing.

Development Timeline

The M.2 specification originated in 2012 as the Next Generation Form Factor (NGFF), developed by the to provide a compact replacement for mSATA and Mini expansion cards, enabling greater integration in mobile and embedded systems. Early drafts, such as Revision 0.3 dated May 16, 2012, outlined the basic electro-mechanical requirements for smaller form factors supporting , USB, and other interfaces. This initiative addressed the need for thinner profiles in ultrabooks and tablets, with collaborating closely on the standard's evolution. The M.2 Specification Revision 1.0 was formally released on November 1, 2013, officially adopting the M.2 name and establishing the core pinout, keying, and socket configurations for mobile adapters. This version targeted applications like wireless modules and storage, supporting up to PCIe 3.0 and while emphasizing low power and small footprints. Industry accelerated with integration into Intel's 4th-generation Core processors (Haswell) and 8-series chipsets in 2013, enabling M.2 slots in laptops and desktops for SSDs and WWAN cards. By 2015, M.2 had seen widespread adoption in solid-state drives, with manufacturers like and releasing consumer NVMe SSDs in the form factor, driven by falling prices and performance gains over interfaces. Revision 4.0 Version 1.0 was released on November 17, 2020, optimizing for higher integration in thin clients and supporting PCIe 4.0 compatibility. Revision 5.0 Version 1.0 was released on April 29, 2023, supporting PCIe 5.0 for increased bandwidth in storage and networking modules, aligning with broader ecosystem shifts toward faster I/O. An associated Engineering Change Notice (ECN) streamlined the specification by removing legacy interfaces like High-Speed Inter-Chip (HSIC), SuperSpeed Inter-Chip (SSIC), and Mini-PCIe (M-PCIe), focusing on modern PCIe and USB standards to reduce complexity. Consumer devices began supporting PCIe 5.0 via M.2 by 2022, with announcements at CES for Gen5 NVMe SSDs reaching up to 14 GB/s reads, marking a milestone in mainstream high-speed storage. The latest update, Revision 5.1 errata released on November 5, 2024, introduced (UFS) support for Socket 3 configurations and 1.2V I/O signaling for WWAN modules, enhancing compatibility with emerging mobile technologies. As of November 2025, ongoing development work on M.2 and related form factors continues, with updates discussed at the Developers Conference in June 2025 to support evolving PCIe generations. These changes reflect ongoing refinements to accommodate PCIe evolution and diverse applications without altering the core form factor.

Technical Specifications

Supported Interfaces

The M.2 form factor supports a range of communication interfaces designed for storage, networking, and expansion applications, with PCIe serving as the primary high-speed pathway. It accommodates up to four PCIe lanes, compatible with generations from 1.0 to 6.0 and forward-compatible with future generations, enabling configurations such as x1, x2, or x4. For PCIe 5.0 x4, this provides a maximum bidirectional bandwidth of approximately 32 GB/s (16 GB/s per direction), leveraging 32 GT/s per lane with 128b/130b encoding efficiency. In addition to PCIe, M.2 includes a single 3.0 port, delivering 6 Gbit/s bandwidth for legacy-compatible storage. Optional USB support extends to versions 2.0, 3.0, and 3.1, facilitating connectivity for peripherals like wireless modules, though typically limited to lower-speed implementations on the shared connector. For storage-specific protocols, M.2 leverages NVMe over PCIe to enable high-performance solid-state drives, offering low-latency access and parallel command queuing for demanding workloads. In contrast, SATA-based devices utilize the AHCI protocol to maintain compatibility with traditional hard drives and older SSDs, ensuring seamless integration in mixed environments. Beyond core storage, M.2 supports SDIO for wireless communication cards, such as Wi-Fi or Bluetooth modules, providing a standardized interface for card-like expansions. Revision 5.1 of the specification (as of November 2024) introduced optional UFS (Universal Flash Storage) interface support via an Engineering Change Notice (ECN) to Socket 3, targeting mobile and embedded storage with high sequential throughput suitable for smartphones and tablets. Interface configurations on M.2 modules are determined by keying and pin assignments on the 75-pin , allowing flexible to share lanes among protocols. Common modes include PCIe x4 for maximum throughput, PCIe x2 for balanced performance, or -only for simpler setups; hybrid options, such as PCIe x2 combined with , enable dual-protocol operation on compatible hosts by dynamically allocating resources. This optimizes the connector's limited pins, supporting across generations while accommodating diverse device types without requiring separate slots.

Electrical and Power Characteristics

M.2 modules primarily rely on a 3.3 V power rail as the main voltage supply, tolerant to ±5% variation and capable of delivering up to 3 A of current, which supports a maximum power budget of approximately 9.9 W for high-performance devices. Optional auxiliary rails include 1.8 V (±8% tolerance, up to 1 A) for signaling and low-power operations in interfaces like SDIO or USB, and a 1.2 V rail introduced via a 2021 Engineering Change Notice (ECN) specifically for wireless wide-area network (WWAN) modules to enable efficient power delivery in mobile applications. These rails ensure stable operation across diverse host environments, with power-up sequencing requiring the 3.3 V rail to settle within 100 ms before auxiliary supplies. Power consumption profiles differ significantly by module type and workload. For PCIe x4 solid-state drives (SSDs), active operation can reach up to 9.5 W, reflecting the demands of high-throughput data transfers, while idle or low-activity states drop below 1 W. In contrast, modules exhibit lower demands, typically averaging 2-3 W during transmission and reception, with peaks around 5 W for dual-band 802.11ac configurations. To promote energy efficiency, M.2 interfaces incorporate PCIe (ASPM) features, including L0s (link partial power-down) and L1 (link clock power-down) states, which reduce power draw by gating the reference clock and suspending idle lanes without . Electrical signaling in M.2 utilizes differential pairs for PCIe lanes, accommodating up to four lanes with data rates scaling from 2.5 GT/s (Gen 1) to 64 GT/s (Gen 6) as of 2025, enabling bidirectional throughput of approximately 252 Gbit/s (126 Gbit/s per direction) in x4 configurations for PCIe 5.0. A 100 MHz reference clock (±300 ppm accuracy) synchronizes operations, distributed via dedicated pins to maintain over short traces. Hot-plug functionality is supported through key signals like PERST# (fundamental reset) for device initialization and CLKREQ# (clock request) for dynamic , allowing modules to enter and exit low-power modes seamlessly during connection events. Thermal management is integral to reliable M.2 operation, with commercial-grade modules specified for temperatures from 0°C to 70°C to prevent performance degradation or failure under typical workloads. Standard configurations rely on via the host system's or thermal pads, without necessitating active fans or heatsinks, as the form factor's compact design and power limits facilitate natural convection in most consumer and embedded applications.

Physical Design

Form Factors and Dimensions

The M.2 form factor encompasses a range of standardized physical dimensions designed to accommodate diverse applications, from mobile devices to desktops. The notation for these sizes follows a "widthlength" convention in millimeters, where the first two digits represent the width and the latter two the length. Standard widths are 12 mm, 16 mm, 22 mm (the most prevalent for general use), and 30 mm, with lengths ranging from 16 mm to 110 mm to suit space constraints and performance needs. For instance, the 2230 variant measures 22 mm wide by 30 mm long, ideal for compact wireless modules and small SSDs in mobile devices, while the 2280 is 22 mm by 80 mm, widely used in storage drives. Shorter form factors like 2230 can also be used in desktop PCs if the motherboard's M.2 slot supports the required interface (typically PCIe NVMe M-key), though additional mounting solutions such as length extender adapters are often required to secure them properly in slots designed for longer sizes.
Form FactorWidth (mm)Length (mm)Typical Use Case
22302230Wireless cards, small SSDs in mobile devices; adaptable for desktops with mounting solutions
22422242Entry-level storage
22602260Balanced mobile storage
22802280High-capacity SSDs in laptops/desktops
2211022110Extended-length modules
30303030Wider connectivity options
30423042Industrial or legacy applications
M.2 cards are constructed as either single-sided, with components on one PCB face for thinner profiles, or double-sided, allowing higher density but increasing thickness. These modules primarily use an edge-card design with a 75-pin gold-finger connector for socketed insertion, enabling easy upgrades. Alternatively, (BGA) packaging supports direct onto the host board, common in embedded systems for permanence and . M.2 sockets are categorized into three types to ensure compatibility with specific module functions, each featuring 75 pins at a 0.5 mm pitch. Socket 1 employs key E for peripheral connectivity, such as or adapters. Socket 2 uses key B+M, supporting storage or wireless wide-area network (WWAN) modules with dual-notch keying for broader compatibility. Socket 3 utilizes key M, optimized for high-speed solid-state drives (SSDs). Keying prevents incorrect insertions by aligning notches on the card edge with socket protrusions. A 2016 Engineering Change Notice (ECN) to the M.2 specification introduced an ultra-compact 11.5 mm by 13 mm PCIe BGA SSD form factor, targeted at space-limited devices like wearables and IoT hardware, expanding options beyond traditional edge-card designs.

Keying and Pinout

The M.2 interface employs mechanical keying notches on the edge connector to ensure proper alignment and prevent insertion of incompatible modules into sockets, thereby safeguarding electrical and mechanical integrity. The connector features 75 pins in total, arranged in an edge-card configuration with signals, power, and ground distributed across both sides. Keying types are defined by the position of removed pins (notches), which correspond to specific supported interfaces and prevent cross-compatibility errors. There are four primary key types standardized in the M.2 specification. Key A removes pins 8 through 15 and is designated for CNVi and Wi-Fi modules, supporting interfaces such as PCIe x2, USB, I²C, and DisplayPort. Key B notches pins 12 through 19, targeting applications like SATA storage and USB devices, with support for PCIe x2, SATA, USB 2.0/3.0, and additional signals like SSIC or audio. Key E eliminates pins 24 through 31, optimized for PCIe x2 connectivity in wireless scenarios, accommodating SDIO, UART, PCM, and USB. Key M removes pins 59 through 66, primarily for high-performance storage with PCIe x4 and SATA capabilities. The pinout mapping allocates specific positions for critical signals to maintain interface consistency across keys. For PCIe lanes in Key M, differential pairs include Lane 0: TX+ on pin 49, TX- on 47, RX+ on 43, RX- on 41; Lane 1: TX+ on 37, TX- on 35, RX+ on 31, RX- on 29; Lane 2: TX+ on 25, TX- on 23, RX+ on 19, RX- on 17; Lane 3: TX+ on 13, TX- on 11, RX+ on 7, RX- on 5, enabling configurations from x1 to x4. SATA signals use pins 49/47 (TX) and 43/41 (RX). Power delivery includes +3.3V rails and grounds on various pins, such as pin 3 (+3.3V) and multiple grounds (e.g., pin 1, 75), with total power budgets varying by key (e.g., up to 5.0 W for Key M). These assignments ensure robust signaling while reserving pins for keying and optional functions like configuration detection. Compatibility rules rely on the keying to enforce module-socket matching; for instance, a Key E module cannot physically insert into a Key M socket due to offset notches, avoiding potential damage from mismatched signals. The dual-key variant, with notches at both Key B (pins 12-19) and Key M (pins 59-66) positions, allows a single module to function in either socket type, supporting both and PCIe x4 operations for versatile storage applications. Configuration pins (e.g., CONFIG_0 to CONFIG_3 on Key B) further guide host detection of the active interface. Specification updates have expanded Key B functionality through engineering change notices (ECNs). The Revision 5.1 ECN, effective March 17, , modifies Key B to enable PCIe x2 and USB 3.1 Gen1 signaling alongside existing and USB options, broadening its use beyond legacy WWAN modules. A separate ECN adds a second PCIe lane to Type 1216 SDIO-based LGA modules, enhancing connectivity for compact wireless form factors while maintaining .

Compatibility and Usage

Host Platform Support

M.2 modules have been supported on Intel platforms since the introduction of 4th-generation Core processors (codenamed Haswell) in 2013, paired with 8-series chipsets like Z87 that provided initial PCIe-based M.2 slots. Subsequent generations expanded this, with modern Intel 700- and 800-series chipsets (launched in 2023 and 2024 for 14th- and 15th-generation Core processors) offering multiple M.2 slots—typically up to four on high-end motherboards—capable of PCIe 5.0 x4 connectivity and bifurcation options to allocate lanes dynamically for storage or other peripherals. For AMD systems, M.2 support began with FM2+ socket motherboards in 2014, utilizing A88X and later chipsets to enable PCIe 3.0 x4 interfaces for compatible modules. Modern AM5 platforms with X870-series chipsets (launched 2024 for Ryzen 9000 series) provide similar PCIe 5.0 x4 M.2 support on high-end boards. Integration with host platforms often requires configuration through or firmware, where users select operational modes for M.2 slots such as PCIe (for NVMe devices), , or hybrid configurations to match the module type and avoid conflicts. functionality is enabled via vendor-specific technologies, including (RST) for NVMe arrays on supported chipsets starting from 100-series and later, or / and Windows Storage Spaces for platforms. These settings ensure optimal performance but may necessitate disabling legacy modes or adjusting boot priorities for compatibility. Operating system support for M.2, particularly NVMe variants, is native in and later versions through built-in drivers that handle PCIe-attached storage without additional software. Linux kernels from version 3.3 (released in 2012) onward include core NVMe drivers, enabling seamless detection and utilization of M.2 modules as block devices. macOS provides limited support, with native NVMe recognition starting in High Sierra (10.13, ) for specific PCIe configurations, though booting from third-party M.2 NVMe drives requires compatible hardware and may not work on all pre-2018 Macs without adapters. Despite broad adoption, M.2 implementation faces limitations due to shared PCIe lanes on motherboards, where populating an M.2 slot can reduce bandwidth to the primary GPU slot (e.g., from x16 to x8) or disable ports in lane-constrained designs. Thermal throttling becomes prominent in dense configurations with multiple high-power modules, as PCIe 5.0 operation generates significant heat without adequate cooling, potentially capping speeds below rated levels. Desktop motherboards typically support 2 to 4 M.2 slots, varying by and board layout, with higher counts reserved for enthusiast models to balance storage expansion against overall system I/O demands. Desktop motherboards' M.2 slots are commonly designed for longer form factors such as 2280, with corresponding mounting standoffs and screw holes for secure installation. Shorter modules such as 2230 can electrically connect and function if the slot supports the required interface (typically PCIe NVMe with M-key). However, due to the lack of appropriate mounting points for shorter lengths, these modules often require an adapter (such as a 2230 to 2280 extender) or custom securing methods (e.g., tape or pressure from a heatsink) to remain properly mounted and avoid potential issues.

Common Applications

M.2 modules are widely used for in consumer and enterprise devices, particularly NVMe SSDs in the 2280 form factor, which serve as boot drives in laptops and desktops. These drives offer capacities up to 16 TB or more, as seen in models like the WD Black SN850X (up to 8 TB) and the Sabrent Rocket 5 (16 TB). In laptops, M.2 NVMe SSDs significantly outperform traditional 2.5-inch SSDs, delivering sequential read speeds over 7,000 MB/s compared to SATA's maximum of around 560 MB/s, resulting in faster boot times and application loading. For wireless connectivity, M.2 modules in the 2230 size and Key E configuration are common in ultrabooks, supporting 6E and 7 standards alongside combo functionality. Examples include Intel's AX211 module, which provides 6E and 5.2 in compact form for seamless integration into thin-and-light laptops. More advanced options like MediaTek's MT7925 enable 7 with 5.3, enhancing multi-gigabit wireless performance and low-latency connections in mobile devices. Other applications include WWAN cellular modems using the Key B interface. Modules from Quectel, such as the BG95 series, fit this form factor for 4G/5G connectivity in laptops and embedded systems. Additionally, M.2 GPU accelerators are emerging in edge computing setups, with devices like the MemryX MX3 providing 24 TOPS of AI inference performance in a compact slot, facilitating real-time processing for IoT and industrial applications without dedicated GPU cards. In gaming PCs, PCIe 4.0 x4 M.2 SSDs like the 990 Pro achieve up to 7,450 MB/s sequential reads, reducing game load times by leveraging high bandwidth for quick asset streaming and improving overall responsiveness compared to slower storage options.

Alternatives and Future Developments

Competing Form Factors

One prominent predecessor to the M.2 form factor is mSATA, introduced in 2009 as a compact alternative to the 2.5-inch drive for ultrabooks and thin laptops. mSATA maintained a smaller , measuring approximately 50.8 mm by 29.85 mm, but was limited to the 3.0 interface with a maximum bandwidth of 6 Gbit/s. This constraint became a key factor in its decline, as M.2's support for PCIe interfaces enabled significantly higher speeds, leading to mSATA's phase-out by the mid-2010s in favor of the more versatile M.2 standard. In enterprise environments, the form factor—previously known as SFF-8639—serves as a direct alternative to M.2, particularly for 2.5-inch SSDs in servers and data centers. supports PCIe x4 and SAS interfaces, allowing for high-performance NVMe storage up to 32 Gbit/s, and accommodates a taller 15 mm height profile that facilitates better heat dissipation compared to M.2's slimmer design. Unlike standard M.2 sockets, U.2 connectors enable hot-swapping, making them suitable for mission-critical systems requiring minimal downtime. For high-density server applications, the Enterprise and Data Center SSD Form Factor (EDSFF) standards, including E1.S and E1.L, offer specialized alternatives to M.2 by prioritizing storage density and thermal management. E1.S, slightly longer and wider than M.2 at 32 mm by 110.15 mm, is designed for 1U compute-optimized servers and replaces M.2 in data center use cases due to its doubled power budget (up to 25 W) and improved airflow for PCIe Gen5 saturation. E1.L extends this for 1U storage servers with even greater capacity per drive at approximately 38.4 mm by 318.75 mm, emphasizing hot-plug functionality and enhanced cooling to address M.2's limitations in sustained high-load environments where overheating can throttle performance. While M.2 excels in client devices, its cooling constraints make it less viable for dense data center deployments compared to these EDSFF variants. M.2's dominance in consumer and mobile markets stems from its compact size, which integrates seamlessly into slim laptops and desktops without occupying drive bays, and its cost-effectiveness to simplified and compatibility. However, it lacks native hot-swap support in most sockets, requiring system shutdowns for module replacement, unlike enterprise-oriented competitors. Keying differences between M.2 and these alternatives, such as U.2's SFF-8639 connector, ensure backward incompatibility but allow for targeted interface support.

Emerging Standards

The development of emerging standards for storage form factors is driven by the limitations of M.2 in handling the thermal and power demands of generative AI (GenAI) workloads, where high-performance SSDs require sustained power above 25W and efficient cooling to maintain performance without throttling. M.2's maximum module height of 3.5mm and power envelope of approximately 9W restrict its scalability in dense server environments for AI accelerators and large-scale data processing. By 2024, adoption of advanced form factors in enterprise servers had accelerated, with hyperscalers integrating them to support PCIe 5.0 SSDs and beyond, enabling up to 45% more drives per rack under power density limits around 15kW. The Enterprise and Data Center Standard Form Factor (EDSFF), standardized by the SNIA SFF Technology Affiliate in 2020, serves as a key post-M.2 development, with the E1.S variant acting as a direct successor to the M.2 2280 in enterprise scenarios. E1.S supports PCIe 5.0 x4 interfaces in a compact footprint of approximately 110mm x 32mm, allowing for denser SSD deployments with up to 16 NAND dies per module—doubling or quadrupling capacity compared to equivalent M.2 drives—while overcoming height constraints through thicknesses of 9.5mm or 15mm. This enables hot-plug functionality and power budgets up to 25W, facilitating seamless upgrades in 1U servers without system downtime. A variant within the EDSFF 1.0 family, the E3.S form factor, targets hyperscale data centers with dimensions of approximately 7.5 mm width x 112.75 mm length x 76 mm height (single or double width), supporting hot-plug operations and power levels exceeding 25W—up to 70W in high-performance configurations. Designed for 2U servers, E3.S replaces traditional 2.5-inch drives, offering improved and for AI and tasks that demand sustained high throughput. Further advancements include proposed extensions for Compute Express Link (CXL) integration over EDSFF form factors, such as the E3.S 2T variant introduced in 2023, which enables coherent memory pooling for AI accelerators by leveraging dual-port PCIe connectivity. Additionally, the PCI-SIG continues to update the M.2 specification to support evolving PCIe generations up to 6.0 at 64 GT/s, with enhanced signal integrity for low-power edge devices while maintaining backward compatibility.

References

  1. https://pcisig.com/specifications?speclib=M.2%2BSpecification
  2. https://pcisig.com/introduction-form-factors-pci-express%C2%AE
  3. https://pcisig.com/pci-express-m2-specification-revision-11
  4. https://www.te.com/en/products/connectors/pcb-connectors/card-edge-connectors/m-2-connectors.html
  5. https://www.atpinc.com/blog/what-is-m.2-M-B-BM-key-socket-3
  6. https://pcisig.com/pci-express-m2-specification-revision-51
  7. https://pcisig.com/specifications
  8. https://sata-io.org/developers/sata-ecosystem/sata-m2-card
  9. https://www.crucial.com/articles/about-ssd/m2-with-pcie-or-sata
  10. https://www.highpoint-tech.com/post/mastering-nvme-hot-plug-navigating-challenges-and-ensuring-safe-removal
  11. https://picture.iczhiku.com/resource/eetop/sHksKPigIJigRbbx.pdf
  12. https://www.delock.com/infothek/M.2_2022/M.2_e.html
  13. https://dokumen.pub/pci-express-m2-specification-revision-10.html
  14. https://pcisig.com/specifications/pciexpress/
  15. https://www.intel.com/content/dam/www/public/us/en/documents/design-guides/4th-gen-core-family-desktop-tmsdg.pdf
  16. https://insights.samsung.com/2015/12/29/a-look-back-at-the-2015-storage-trends-m-2-ssds-and-3d-v-nand-technology/
  17. https://www.scribd.com/document/790812147/PCI-Express-M-2-Spec-Rev5-0-Ver1-0-0429202-NCB-2
  18. https://nascompares.com/2022/01/10/ces-2022-data-round-up-new-pcie-5-ssds-wifi-6e-releases-usb-4-phison-e26-speeds-new-qnaps-and-more/
  19. https://pcisig.com/pci-sig-developers-conference-2025-agenda
  20. https://pcisig.com/specifications/pciexpress
  21. https://forums.tomshardware.com/threads/using-m-2-2230-ssd-on-my-pc-motherboard-adaptor.3840244/
  22. https://www.congatec.com/fileadmin/user_upload/Documents/Application_Notes/AN43_M.2_Pinout_Descriptions_and_Reference_Designs.pdf
  23. https://pinoutguide.com/HD/M.2_NGFF_connector_pinout.shtml
  24. https://www.nitroware.net/reviews/281-haswell1
  25. https://cdrdv2-public.intel.com/743835/743835-004.pdf
  26. https://www.intel.com/content/www/us/en/products/sku/241152/intel-z890-chipset/specifications.html
  27. https://www.guru3d.com/story/ecs-launches-new-amd-fm2-platform-motherboards/
  28. https://www.amd.com/en/products/chipsets-am5.html
  29. https://www.crucial.com/support/articles-faq-ssd/bios-uefi-configuration-optimizing-m2
  30. https://nvmexpress.org/drivers/linux-driver-information/
  31. https://forums.anandtech.com/threads/pciex16-drops-to-x-8-when-m2-ssd-is-installed.2611401/
  32. https://www.tomshardware.com/pc-components/motherboards/if-you-think-pcie-50-runs-hot-wait-till-you-see-pcie-60s-new-thermal-throttling-technique
  33. https://www.tomshardware.com/news/b760-motherboard-leaks-online-
  34. https://www.reddit.com/r/buildapc/comments/18oizbh/can_i_put_an_m2_2230_ssd_in_a_motherboard_with_an/
  35. https://www.tomshardware.com/reviews/wd-black-sn850x-ssd-review-back-in-black
  36. https://www.sabrent.com/product/RKT5-16TB/
  37. https://www.tomshardware.com/raspberry-pi/lattepanda-mu-review
  38. https://www.notebookcheck.net/Asus-ProArt-PX13-review-The-world-s-fastest-13-3-inch-2-in-1-thanks-to-AMD-Zen-5-and-RTX-4070-laptop.868429.0.html
  39. https://images.quectel.com/python/2023/04/Quectel_BG95_Series_Hardware_Design_V1.5.pdf
  40. https://www.tomshardware.com/tech-industry/artificial-intelligence/memryx-launches-usd149-mx3-m-2-ai-accelerator-module-capable-of-24-tops-compute-power
  41. https://www.tomshardware.com/reviews/samsung-990-pro-ssd-review
  42. https://news.solidigm.com/en-WW/228148-tracing-the-evolution-of-data-center-ssd-form-factors/
  43. https://www.oscoo.com/news/what-are-the-differences-between-m-2-and-msata-ssd-form-factors/
  44. https://www.cdw.com/content/cdw/en/articles/hardware/ssd-types-m2-sata-nvme-u2.html
  45. https://www.techtarget.com/searchstorage/definition/U2-SSD-formerly-SFF-8639
  46. https://www.ssstc.com/knowledge-detail/guide-to-ssd-form-factors/
  47. https://americas.kioxia.com/en-us/business/ssd/solution/edsff/e1.html
  48. https://documents.westerndigital.com/content/dam/doc-library/en_us/assets/public/western-digital/collateral/white-paper/white-paper-top-considerations-for-enterprise-ssds.pdf
  49. https://www.sabrepc.com/blog/computer-hardware/explaining-edsff-ssds
  50. https://www.ssstc.com/knowledge-detail/ssd-m2-edsff/
  51. https://www.pcgamer.com/you-know-m2-ssds-suck-right/
  52. https://rog.asus.com/articles/hands-on/easy-guide-to-ssds-sata-msata-m-2-and-u-2/
  53. https://www.servethehome.com/e1-and-e3-edsff-to-take-over-from-m-2-and-2-5-in-ssds-kioxia/
  54. https://www.snia.org/sites/default/files/SSSI/CMSS24/CMSS24-Hands-The-Latest-on-Form-Factors.pdf
  55. https://www.snia.org/educational-library/edsff-dynamic-family-form-factors-data-center-ssds-2020
  56. https://www.snia.org/blog/2020/answering-your-questions-edsff
Add your contribution
Related Hubs
User Avatar
No comments yet.