AMD Ryzen AI Embedded X188 socket: FP11

The AMD Ryzen AI Embedded X188 uses the FP11 package, soldered with 2,077 balls. It is the 12-core Strix Halo embedded APU with 32 graphics compute units for industrial and robotics designs.

Socket
FP11
Launched
July 2026
Cores and threads
12 and 24
Default TDP
55 W
Codename
Strix Halo
Segment
Embedded
Clocks
3.2 / 5.0 GHz

AMD Ryzen AI Embedded X188 specifications

Cores12 (uniform Zen 5 cores)
Threads24
Core architectureZen 5, chiplet CCDs with large I/O die
Clocks3.2 GHz base, up to 5.0 GHz maximum boost
Cache64 MB L3 and 12 MB L2, plus 32 MB MALL cache for the GPU
Default TDP55 W
Configurable TDP45 to 120 W
Manufacturing processTSMC 4 nm
Memory supportSoldered LPDDR5X up to 8533 MT/s, 8 channels (256-bit), up to 128 GB unified, with Link-ECC
PCI Express16 PCIe Gen4 lanes (16L)
Integrated graphicsAMD Radeon Graphics, 32 compute units, up to 2800 MHz
NPUXDNA 2, up to 50 TOPS
MultiplierLocked
Junction temperature0 to 105 °C (commercial grade)
PackageFP11 (BGA2077, soldered)
Cooler in the boxNone (soldered, system thermal solution required)
Lifecycle10-year embedded availability, last time buy 2037

Motherboard compatibility

AMD Ryzen AI Embedded X188 needs a motherboard with a FP11 socket. Chipset by chipset data for this codename is not recorded yet. See the FP11 chipset list and the board maker's CPU support list.

FP11 at a glance

TypeBGA, 2,077 contacts
Package size37.5 × 45 mm
IntroducedJanuary 2025
MemorySoldered LPDDR5X only, 256-bit bus (8 channels), up to LPDDR5X-8533 and up to 128 GB of unified memory shared by CPU, GPU and NPU, with Link-ECC on embedded boards
Cooler mounting holesNot applicable (soldered)

Read the full FP11 socket guide

What the Ryzen AI Embedded X188 is

The Ryzen AI Embedded X188 is the 12-core member of the X100 series that AMD launched in July 2026, sitting between the 16-core X199 flagship and the 8-core X168. It carries 12 Zen 5 cores with 24 threads at 3.2 GHz base and up to 5.0 GHz boost, backed by 12 MB of L2 and the full 64 MB of L3. Graphics stay strong at 32 RDNA 3.5 compute units running to 2800 MHz, and the XDNA 2 NPU contributes up to 50 TOPS. Default TDP is 55 W within a 45 to 120 W configurable range.

The balance of this SKU is deliberate: it keeps two thirds of the flagship CPU while holding most of the graphics throughput, since the step from 40 to 32 compute units is smaller than the step from 16 to 12 cores. Vision and inference loads that lean on the GPU lose little, while control software still gets 12 full Zen 5 cores with AVX-512. The die is the same Strix Halo chiplet assembly built on TSMC 4 nm.

Embedded qualification adds continuous operation rating, a 10-year supply program ending with last time buy in 2037, and module support including the Kria AI system-on-module. Sampling opened in June 2026 with production in the fourth quarter. The standard grade spans junction temperatures from 0 to 105 degrees, and buyers who need frost tolerance take the AMD Ryzen AI Embedded X188i instead.

Socket and board compatibility

Like every Halo chip, the X188 solders to its board in the FP11 array, so compatibility means board and module choice rather than socket matching. The 37.5 × 45 mm package dwarfs the FP8 of smaller APUs because the 256-bit memory bus needs the ball count. Once soldered, core count, graphics size and memory are permanent for that unit.

Boards and modules built for one X100 SKU generally share power and memory design with the others, but each SKU needs its own validated build and firmware. COM-HPC carriers and the Kria AI robotics platform give developers a starting point before committing to a custom board. Production modules were announced from congatec, Arbor, iBase, IEI, Sapphire and Seavo.

Upward the series offers the 16-core AMD Ryzen AI Embedded X199, downward the 8-core X168 keeps the same 32 compute units. Smaller power budgets move to FP8 and the P100 series.

Memory, PCIe and cooling

Memory is eight channels of soldered LPDDR5X up to 8533 MT/s, a 256-bit unified pool to 128 GB with Link-ECC that the CPU, GPU and NPU share without copies. For a 12-core design the bandwidth matters as much as capacity, because 32 graphics compute units and 12 CPU cores contend for the same bus. Fixing the size at design time is mandatory since nothing can be added later.

The SoC contributes 16 PCIe Gen4 lanes with USB4, USB 3.2 and display outputs to 7680x4320 across four screens. Typical splits feed NVMe, networking and vision interfaces, while the integrated graphics remove the need for a discrete card. Board documentation defines the actual allocation.

Cooling covers the configured TDP between 45 and 120 W with a system heatsink or cold plate on the lidless package. The 55 W default suits compact enclosures, while sustained full load wants the upper end of the thermal design validated. No boxed cooler exists for soldered embedded parts.

Platform, software and lifecycle

The X188 fits automation, medical imaging and robotics designs that want near-flagship graphics with a smaller CPU than the X199. Software runs Linux with ROCm, PyTorch, ONNX and TensorFlow, and Xen virtualization separates real-time control from AI workloads. AMD's 10-year availability with last time buy in 2037 protects certified designs from forced requalification.

Against client silicon the argument is lifecycle and temperature range rather than peak clocks. Against smaller embedded SoCs it is headroom: 12 full Zen 5 cores plus 32 compute units in one power envelope that starts at 45 W configurable.

Designs watching the low end of the thermometer should compare the industrial AMD Ryzen AI Embedded X188i, which repeats this specification from minus 40 degrees.

Upgrade paths

No processor swap is possible on soldered FP11. Growth within the package means a new board build around the 16-core X199, and savings mean a build around the 8-core X168. Each move revalidates power, thermals and memory size.

Outside Halo, the FP8 P100 series addresses lower power embedded systems, and desktop Gorgon silicon sits on AM5. Plan memory capacity early: soldered LPDDR5X cannot be extended after manufacture.

Frequently asked questions

What socket does the Ryzen AI Embedded X188 use?

It uses FP11, a 2,077-ball grid array. The chip is soldered and cannot be moved to another board.

How large is the integrated GPU in the X188?

32 compute units at up to 2800 MHz, the same graphics size as the 8-core X168 and close to the 40 of the flagship X199.

What memory does the X188 use?

Soldered LPDDR5X up to 8533 MT/s on a 256-bit bus, up to 128 GB shared by the CPU, GPU and NPU.

What is the difference between the X188 and the X188i?

The X188i repeats the clocks and core count with industrial qualification from minus 40 to 105 degrees.

Sources and verification

  1. AMD Ryzen AI Embedded X188 (AMD)
  2. AMD Ryzen AI Embedded X100 for Robotics and Edge AI (Igor's Lab)

Facts last checked against these sources on . Page last edited on .