The company's portfolio expanded across desktop and mobile computing, gaming, graphics, embedded systems, servers, data centers, and high-performance computing. More recently, artificial intelligence has become another central area of AMD development, with AI capabilities now spanning data-center accelerators, server CPUs, client processors, embedded computing, and adaptive devices.

Xilinx History

Xilinx pioneered the commercial FPGA and helped establish programmable logic as an alternative to fixed-function semiconductor designs.

Xilinx was founded in 1984 and played a foundational role in programmable logic. Xilinx co-founder Ross Freeman invented the field-programmable gate array, or FPGA, as a semiconductor device that could be configured after manufacturing rather than being limited to a fixed hardware function. The XC2064, shipped in 1985, became the world's first commercial FPGA. That innovation created a new approach to digital hardware development. Rather than committing design to an application-specific integrated circuit, engineers could implement and modify logic in a programmable semiconductor device. Over the following decades, Xilinx introduced major product families including Spartan, Virtex, Artix, Kintex, Zynq, and Versal. The company also helped pioneer the fabless semiconductor business model and progressively moved beyond conventional programmable logic into heterogeneous computing platforms combining processors, programmable logic, specialized processing resources, and software tools.

The Zynq family, introduced in the 2010s, combined Arm processors with programmable logic. Versal later extended this concept with additional processing resources, a programmable Network on Chip, and specialized engines designed for different workloads.

AMD Acquires Xilinx

AMD completed its acquisition of Xilinx on February 14, 2022, adding FPGA, adaptive SoC, and embedded computing technologies to its processor and graphics portfolio.

The acquisition brought Xilinx's adaptive computing portfolio into AMD and expanded the company's reach across embedded, communications, industrial, data center, automotive, aerospace, and other markets. The combination also created a significantly broader technology portfolio. AMD could now address workloads using general-purpose CPUs, GPUs, FPGA fabric, adaptive SoCs, embedded processors, and specialized acceleration technologies. The acquisition also means that many products historically known as Xilinx FPGAs or Xilinx SoCs are now part of AMD's adaptive computing portfolio. Legacy Xilinx part numbers remain important, particularly when maintaining existing systems or sourcing discontinued components.

AMD/Xilinx Today: High-Performance and Adaptive Computing

AMD/Xilinx today spans multiple computing architectures, enabling organizations to match CPUs, GPUs, FPGAs, adaptive SoCs, and dedicated accelerators to the performance requirements of specific workloads.

AMD's current portfolio is built around the idea that different workloads benefit from different computing architectures. CPUs provide flexible general-purpose processing. GPUs provide highly parallel computing resources for graphics, artificial intelligence, simulation, and other computational workloads. FPGAs allow hardware functionality to be configured for a specific application. Adaptive SoCs combine programmable logic with processors and other hardened resources. Dedicated accelerators deliver additional performance for demanding data center, AI, and high-performance computing workloads. This portfolio diversity has become increasingly important as computing requirements grow more specialized. A factory automation system, an AI server, a wireless base station, a gaming PC, an autonomous robot, and a supercomputer all require very different combinations of processing, latency, memory bandwidth, I/O, power consumption, and programmability.

AMD now addresses these requirements across data centers, PCs, embedded computing, AI, communications, graphics, physical AI, and adaptive computing.

AMD FPGA Portfolio

AMD’s FPGA portfolio spans cost-optimized Spartan and Artix devices, mid-range Kintex products, and high-performance Virtex FPGAs, which were inherited from Xilinx.

Field-programmable gate arrays remain one of the most important technologies inherited from Xilinx. An FPGA contains configurable logic resources, memory, interconnects, I/O, and other hardware blocks that engineers can program to implement custom digital circuitry. Unlike a CPU, which executes instructions using a fixed processor architecture, an FPGA can configure the underlying hardware to match the application. That flexibility makes FPGAs particularly useful when systems require deterministic processing, very low latency, custom interfaces, parallel data processing, hardware acceleration, or the ability to modify functionality after deployment. AMD currently maintains FPGA families across Spartan, Artix, Kintex, and Virtex product lines.

Spartan FPGAs

AMD Spartan FPGAs are designed primarily for cost-sensitive, high-volume applications that require programmable logic, flexible I/O, and efficient power consumption.

Spartan UltraScale+: Spartan UltraScale+ represents AMD's newest generation of Spartan devices. Built on AMD's 16 nm technology, the family emphasizes high I/O density, security, power efficiency, and cost-sensitive programmable logic applications. AMD moved the larger SU200P device into volume production in July 2026. Potential applications include industrial systems, healthcare equipment, wired and wireless infrastructure, edge computing, professional AV, embedded control, and data-center management. For clients familiar with older Spartan technologies, Spartan UltraScale+ also represents an important example of how the family has evolved beyond traditional low-end FPGA requirements.

Spartan 7: Spartan 7 devices are part of AMD's 7 Series portfolio and continue to provide programmable logic for cost, size, and power-sensitive systems. Spartan 7 FPGAs can be used in industrial control, connectivity, embedded systems, peripheral management, and other applications where high-end FPGA resources may not be necessary.

Spartan 6: Spartan 6 remains widely recognized because of its long use in embedded and industrial designs. However, it represents a legacy generation rather than AMD's newest Spartan architecture. For procurement teams, this distinction is important. Older FPGA families may remain installed in equipment for many years after newer technologies become available, creating sourcing requirements that differ significantly from those of engineers beginning new designs.

Artix FPGAs

AMD Artix FPGAs provide cost- and power-optimized programmable logic for applications that still require capable DSP resources, connectivity, and embedded processing performance.

Artix 7: Artix 7 FPGAs are part of AMD's 28 nm 7 Series and are used across embedded, industrial, communications, and other applications requiring relatively low-cost programmable logic with transceiver capability. AMD has extended the lifecycle of its 7 Series portfolio through 2040, helping support long-life designs that continue to depend on devices such as Artix 7.

Artix UltraScale+: Artix UltraScale+ moves the Artix family onto AMD's 16 nm UltraScale+ architecture. The devices are designed for cost-sensitive edge and networking applications and include compact packaging options that support designs with constrained board space. Applications include industrial networking, machine vision, medical equipment, broadcast systems, edge computing, and aerospace and defense systems.

Kintex FPGAs

AMD Kintex FPGAs occupy the mid-range of the portfolio, balancing performance, power, connectivity, logic density, and cost.

Kintex 7: Kintex 7 devices have been used for applications such as communications, signal processing, industrial equipment, and embedded computing.

Kintex UltraScale: Kintex UltraScale moved the family to AMD's 20 nm UltraScale architecture, bringing higher levels of integration and connectivity for applications requiring more performance than earlier generations.

Kintex UltraScale+: Kintex UltraScale+ further increased processing and connectivity capabilities using AMD's 16 nm architecture. These FPGAs can support applications including networking, signal processing, data-center acceleration, video processing, and embedded computing.

Kintex UltraScale+ Gen 2: AMD introduced Kintex UltraScale+ Gen 2 as the next evolution of its mid-range FPGA portfolio. The family is designed around high-bandwidth connectivity, deterministic processing, increased memory capability, security, and long-term design continuity. AMD identifies applications including broadcast and professional AV, healthcare, industrial systems, machine vision, robotics, and test and measurement. Because product availability can differ across individual Gen 2 devices, designers should verify the current production and sampling status of the exact device required for a new design.

Virtex FPGAs

AMD Virtex FPGAs represent the high-performance end of the company’s programmable logic portfolio, offering substantial logic density, bandwidth, DSP capability, and connectivity.

Virtex 7: Virtex 7 devices use the 28 nm 7 Series architecture and provide high logic capacity, DSP resources, and I/O bandwidth for demanding applications.

Virtex UltraScale: Virtex UltraScale advanced the family to the 20 nm node and targeted high-performance applications including networking, large-scale prototyping, and emulation.

Virtex UltraScale+: Virtex UltraScale+ devices use AMD's 16 nm architecture and provide high levels of logic capacity, serial connectivity, signal-processing resources, and system integration. Applications include networking, communications, data-center infrastructure, aerospace and defense, high-performance signal processing, and ASIC or SoC prototyping.

AMD Processors and Computing Platforms

AMD’s processor portfolio covers PCs, workstations, servers, AI systems, and data-center infrastructure through Ryzen, Ryzen AI, and EPYC platforms.

AMD Ryzen Processors

AMD Ryzen processors power consumer, commercial, gaming, workstation, mobile, and AI-focused PCs across a wide range of performance requirements.

The Ryzen portfolio includes conventional desktop and mobile processors as well as increasingly AI-focused platforms. During 2026, AMD expanded its Ryzen AI portfolio with Ryzen AI 400 Series and Ryzen AI PRO 400 Series products, including desktop options. AMD has also continued developing Ryzen AI Max-class products for systems requiring substantial CPU, GPU, and AI processing resources. Instead of relying entirely on cloud infrastructure, modern AI processing systems can use dedicated neural processing resources alongside CPU and GPU compute. For organizations deploying PCs, workstations, or intelligent edge systems, processor selection therefore increasingly depends on more than traditional CPU performance. Graphics performance, AI acceleration, power efficiency, memory architecture, and platform longevity can all affect the appropriate configuration.

AMD Socket AM5 Platform

AMD Socket AM5 is the long-term desktop platform for modern Ryzen processors, supporting DDR5, PCIe 5.0, and planned socket support through 2029.

AM5 introduced technologies, including DDR5 memory and PCIe 5.0, while providing compatibility across multiple generations of AMD desktop processors and chipsets. The platform is especially significant because AMD has placed a strong emphasis on upgradeability and socket longevity. At Computex 2026, AMD extended its commitment to Socket AM5 support through 2029.

For system builders and organizations managing hardware over longer periods, a sustained socket platform can reduce the need to replace an entire motherboard and surrounding system every time a CPU upgrade is required. Compatibility still depends on the individual processor, motherboard, chipset, and BIOS implementation, but AMD's extended roadmap gives AM5 a comparatively long planned lifespan.

AMD EPYC Processors

AMD EPYC processors are server CPUs designed for cloud infrastructure, enterprise workloads, high-performance computing, databases, and AI systems.

The EPYC portfolio has progressed through several generations as data centers have placed increasing demands on core density, memory bandwidth, I/O performance, energy efficiency, and accelerator connectivity. AMD's 5th Generation EPYC 9005 Series remains part of its server portfolio, while the company launched the 6th Generation EPYC 9006 Series, formerly code-named Venice, in July 2026. The new generation uses AMD's Zen 6 and Zen 6c architectures and is positioned for cloud, enterprise, HPC, database, and AI infrastructure. EPYC processors also play a critical supporting role in GPU-accelerated systems. In an AI server or supercomputer, GPUs may perform much of the parallel computation, but CPUs continue to handle operating systems, data preparation, orchestration, host processing, and other workloads necessary to keep accelerators utilized.

AMD Graphics and GPU Technologies

AMD graphics technologies serve gaming, professional visualization, content creation, local AI, and GPU-accelerated computing through Radeon and Radeon PRO products.

AMD Radeon Graphics

AMD Radeon graphics provide GPU performance for gaming, content creation, visualization, and accelerated computing on personal computers.

The current Radeon RX 9000 Series is built on the AMD RDNA 4 architecture. AMD positions RDNA 4 around improvements in graphics performance, ray tracing, and AI processing compared with previous Radeon architectures. Radeon products are closely connected to the broader Ryzen ecosystem, particularly in desktop and gaming systems where AMD CPUs and GPUs may be deployed together.

AMD Radeon PRO Graphics

AMD Radeon PRO and Radeon AI PRO graphics are designed for professional workloads, including CAD, engineering, visualization, media creation, and local AI processing. As of 2026, AMD's professional graphics portfolio includes Radeon PRO W-series products as well as Radeon AI PRO R9000 Series graphics based on RDNA 4.

AMD Adaptive SoCs

AMD adaptive SoCs combine programmable logic, processors, and specialized hardware resources in a single device, enabling workloads to be distributed across the architecture according to performance needs.

Instead of relying exclusively on FPGA fabric, the system can distribute workloads among CPUs, programmable logic, DSP resources, AI Engines, integrated interfaces, and other specialized blocks. AMD's adaptive SoC portfolio includes the established Zynq family and the more advanced Versal architecture.

Zynq Adaptive SoCs

AMD Zynq adaptive SoCs combine Arm processors with programmable logic to support systems that require both embedded software and customizable hardware acceleration.

Zynq-7000 SoC: Zynq-7000 combines Arm processor cores with FPGA programmable logic, allowing software and custom hardware to operate within one device. This architecture made Zynq suitable for applications requiring both embedded software and specialized hardware processing, including industrial automation, machine vision, motor control, communications, and embedded systems.

Zynq UltraScale+ MPSoC: Zynq UltraScale+ MPSoC expanded this architecture with a multiprocessor subsystem and UltraScale+ programmable logic. The platform supports more demanding embedded applications spanning industrial, automotive, communications, video, networking, aerospace, and edge computing. AMD describes the broader Zynq portfolio as combining the software programmability of Arm processors with FPGA hardware programmability.

Zynq UltraScale+ RFSoC: The Zynq UltraScale+ RFSoC family integrates RF data converters directly with programmable logic and processing resources. By bringing RF analog-to-digital and digital-to-analog conversion into the SoC, RFSoCs can reduce the number of separate components required in radio and signal-processing systems. AMD identifies applications including wireless infrastructure, radar, test and measurement, and other direct RF-sampling systems.

AMD Versal Adaptive SoCs

AMD Versal adaptive SoCs combine programmable logic with scalar processors, DSP resources, AI Engines, hardened interfaces, and a programmable Network-on-Chip for heterogeneous computing workloads. Rather than simply combining processors with FPGA fabric, Versal devices can integrate scalar processing, programmable logic, DSP resources, AI Engines, hardened interfaces, and a programmable Network on Chip. Different Versal families emphasize different combinations of these capabilities.

Versal AI Edge Series: Versal AI Edge devices are designed for embedded and edge systems that require real-time AI inference close to sensors and other data sources. Potential uses include intelligent vision, automotive systems, industrial AI, robotics, medical imaging, and other workloads requiring real-time processing.

Versal AI Edge Series Gen 2: Versal AI Edge Series Gen 2 expands AMD’s edge AI architecture with newer AI Engines, application and real-time processors, programmable logic, and integrated image and video processing. AMD describes the devices as combining programmable logic, next-generation AI Engines, application and real-time processors, and integrated image/video processing resources. This makes the family especially relevant as edge systems become responsible for more complete AI pipelines rather than merely forwarding sensor data to a remote server.

Versal AI Core Series: Versal AI Core devices prioritize AI Engine compute and acceleration for workloads including data centers, wireless processing, imaging, and test equipment. AMD identifies applications including data-center computing, wireless beamforming, video and image processing, and test equipment.

Versal Prime Series: Versal Prime provides a general-purpose balance of processing, programmable logic, connectivity, and system resources. The platform addresses applications across embedded computing, communications, networking, and industrial systems.

Versal Prime Series Gen 2: Versal Prime Series Gen 2 extends the Prime platform with newer processing and connectivity capabilities for next-generation embedded systems. AMD expanded the family during 2026 with additional Gen 2 devices.

Versal Premium Series: Versal Premium targets high-end systems that require substantial bandwidth, connectivity, programmable logic, and processing resources. The family is suitable for networking, communications, acceleration, data processing, and other high-end workloads.

Versal Premium Series Gen 2: Versal Premium Series Gen 2 adds technologies including PCIe Gen6, CXL 3.1, newer memory interfaces, upgraded transceivers, and enhanced security for data-intensive computing systems. These capabilities are particularly relevant to data-intensive systems where moving information between processors, memory, storage, and accelerators can become as important as raw computational performance.

Versal HBM Series: Versal HBM integrates high-bandwidth memory directly with adaptive computing resources to address workloads that are memory-bandwidth-limited. Potential applications include data-center acceleration, networking, memory-intensive compute, and analytics.

Versal RF Series: Versal RF integrates programmable logic, DSP resources, and high-speed RF data converters for demanding RF and signal-processing systems. AMD positions the family for applications including test and measurement, aerospace and defense, communications, and emerging wireless infrastructure.

Versal Premium VP1902 Adaptive SoC: The Versal Premium VP1902 is a high-capacity adaptive SoC designed primarily for large-scale ASIC and SoC emulation and prototyping.

The VP1902 belongs specifically to the Versal Premium family rather than representing a separate FPGA product line. AMD designed the VP1902 for large-scale emulation and prototyping, with 18.5 million logic cells and extensive connectivity resources. Its processing subsystem and very high programmable logic capacity allow developers to prototype complex ASICs and SoCs, perform hardware/software development, and validate large systems before committing them to fixed silicon. The device demonstrates how far programmable technology has progressed. AMD notes that the first XC2064 commercial FPGA contained 64 configurable logic blocks, while the VP1902 represents a modern device with billions of transistors and millions of logic cells.

AMD System-on-Modules

AMD Kria System-on-Modules provide production-ready embedded computing platforms that integrate processing hardware, memory, power management, and other core components into a compact module.

Not every embedded system designer wants to create a complete custom board around an FPGA or adaptive SoC. AMD's Kria System-on-Modules offer another approach by integrating processing hardware, memory, power management resources, and other components into a production-ready module.

AMD Kria System-on-Modules

AMD Kria K24 SOM: The AMD Kria K24 SOM targets industrial and embedded applications such as motor control, DSP, robotics, and factory automation. Using a SOM can simplify development by allowing engineers to focus more effort on carrier-board design and application-specific hardware rather than rebuilding the entire processing subsystem.

AMD Kria K26 SOM: The AMD Kria K26 SOM is designed for vision AI, robotics, industrial automation, and intelligent video processing applications. AMD also offers development platforms, such as the KV260 Vision AI and KR260 Robotics starter kits, based on K26 technology.

AMD Kria AI SOM and Robotics Platform: AMD expanded the Kria portfolio in July 2026 with the Kria AI SOM and Robotics Developer Platform for physical AI and autonomous robotics.

Physical AI places unique demands on computing hardware. Robots and autonomous systems must process sensor inputs, execute AI models, make decisions, control motion, and interact with the physical world in real time. AMD's expanded Kria strategy reflects the growing importance of combining AI processing with embedded, deterministic, and real-time computing.

AMD Accelerators

AMD accelerators provide dedicated compute for artificial intelligence, high-performance computing, scientific workloads, and specialized data-center applications through Instinct GPUs and Alveo adaptable accelerator cards.

As AI, scientific computing, and data-center workloads have grown, dedicated accelerators have become an increasingly important part of AMD's portfolio. AMD offers GPU-based Instinct accelerators as well as adaptable Alveo accelerator cards based on FPGA and adaptive SoC technology.

AMD Instinct Accelerators

AMD Instinct accelerators are data-center GPUs and APUs designed for large-scale AI, high-performance computing, and scientific workloads.

AMD Instinct MI300 Series: The MI300 generation includes the MI300X GPU accelerator and MI300A accelerated processing unit. MI300X is designed for GPU-accelerated AI and HPC workloads, while MI300A combines CPU and GPU resources in a highly integrated architecture. The MI300A has become particularly prominent through its use in exascale computing systems such as El Capitan.

AMD Instinct MI325X: The MI325X expanded the MI300-generation portfolio for memory-intensive AI workloads and represented another step in AMD's effort to compete for large-scale training and inference deployments.

AMD Instinct MI350 Series: AMD's MI350 generation includes the MI350X and MI355X and is based on the company's newer accelerator architecture for generative AI, training, inference, and high-performance computing. The MI350 generation also serves as the predecessor to AMD's 2026 rack-scale platform strategy.

AMD Instinct MI400 Series: At Advancing AI in July 2026, AMD introduced the Instinct MI400 Series as its next-generation AI and HPC accelerator family. The MI400 portfolio spans different workload targets. The MI455X is central to AMD's Helios rack-scale AI platform, while the MI430X is aimed at sovereign AI and HPC applications. AMD states that MI430X availability is expected in 2027, making it important to distinguish between products AMD has introduced as part of its roadmap and individual accelerators that are already generally available.

AMD Helios Rack-Scale AI Platform

AMD Helios is an integrated rack-scale AI platform that combines Instinct GPUs, EPYC CPUs, Pensando networking, and ROCm software for large-scale deployments.

The launch of AMD Helios marks a broader shift from selling individual processors and accelerators toward designing complete rack-scale AI infrastructure. AMD Helios combines 72 Instinct MI455X GPUs, 18 6th Gen EPYC Venice CPUs, Pensando networking, and ROCm software within an integrated rack-scale architecture. AMD announced in July 2026 that Helios systems were entering production for large-scale AI deployments. This type of architecture reflects how AI infrastructure is evolving. Performance increasingly depends on the interaction among GPUs, CPUs, memory, interconnects, networking, and software rather than any single component in isolation.

AMD Alveo Adaptable Accelerator Cards

AMD Alveo accelerator cards use programmable FPGAs or adaptive SoCs to accelerate specialized workloads that benefit from deterministic latency and customizable data processing. This can be valuable for applications requiring deterministic latency, custom data movement, nonstandard data types, or specialized acceleration.

AMD Alveo V80: The Alveo V80 is based on the Versal HBM architecture and combines programmable FPGA fabric with HBM2e memory, high-speed networking, and PCIe connectivity. AMD positions the V80 for memory-intensive applications, including HPC, genomic processing, data analytics, network security, computational storage, sensor processing, and FinTech.

AMD Embedded and Soft-Core Processors

AMD MicroBlaze and MicroBlaze V provide configurable soft-core processing for embedded systems implemented within AMD programmable logic.

AMD MicroBlaze Processor

AMD MicroBlaze is a configurable soft processor that enables embedded processing to be implemented directly within an FPGA or an adaptive device.

Unlike a discrete physical processor, a soft processor is implemented using programmable resources within an FPGA or an adaptive device. Designers can therefore combine custom logic and processor functionality within the same programmable architecture. MicroBlaze has been widely used for embedded control, system management, communications, industrial applications, and custom FPGA-based systems.

AMD MicroBlaze V Processor

AMD MicroBlaze V brings the open RISC-V instruction-set architecture to AMD’s configurable soft-processor ecosystem.

AMD MicroBlaze V is available in 32-bit and 64-bit implementations and is integrated into the Vivado and Vitis development flow. AMD also describes it as hardware-compatible with classic MicroBlaze, which can help developers transition existing programmable-logic designs to the RISC-V ecosystem. The use of RISC-V also opens access to a broader open-source software ecosystem while retaining the configurability of a soft processor implemented within AMD programmable hardware.

AMD/Xilinx Development Tools and Software

AMD supports its FPGA, adaptive SoC, embedded, AI, and GPU platforms with software tools covering hardware design, software development, AI inference, Linux, simulation, debugging, and GPU computing.

AMD Vivado Design Suite

AMD Vivado Design Suite is the primary hardware development environment for designing, implementing, verifying, and configuring AMD FPGAs and adaptive SoCs.

Vivado supports design entry, synthesis, place-and-route, simulation and verification, IP integration, timing analysis, hardware debugging, device configuration, and other stages of programmable logic development.

Vivado 2026.1: Vivado 2026.1 is particularly notable because AMD introduced a new tier-based licensing structure beginning with this release. The model ranges from free entry-level access through higher tiers supporting additional FPGA families, Versal adaptive SoCs, advanced features, and enterprise requirements. Existing perpetual Vivado Enterprise licensing remains supported.

Vivado 2026.1 also added or expanded support for newer devices including Spartan UltraScale+, Versal RF, Versal AI Edge Series Gen 2, Versal Prime Series Gen 2, and Versal Premium Series Gen 2. For engineering teams, keeping development tools aligned with device-generation support is important. A newer FPGA or adaptive SoC may require a newer Vivado release even when an organization has extensive experience with previous Xilinx tools.

AMD Vitis Unified Software Platform

AMD Vitis provides software development and debugging tools for embedded processors and heterogeneous applications running across AMD platforms.

While Vivado is focused heavily on the hardware implementation of programmable devices, Vitis provides an environment for developing software and heterogeneous applications targeting AMD platforms. AMD describes Vitis as a unified environment for software development, including applications for embedded processors and heterogeneous systems. It works alongside Vivado and includes development, debugging, libraries, compilers, analysis tools, and other resources. This hardware/software relationship is particularly important in Zynq and Versal systems, where developers may be working simultaneously with processor software, programmable logic, and specialized processing resources.

Zephyr RTOS Support for AMD MicroBlaze V

AMD’s 2026.1 Vitis workflow supports developing and debugging Zephyr RTOS applications on compatible MicroBlaze V configurations.

MicroBlaze V's integration with the RISC-V ecosystem has also created new options for real-time operating systems. AMD's current 2026.1 Vitis documentation includes workflows for debugging Zephyr RTOS applications running on MicroBlaze V processors. The workflow combines a hardware design created in Vivado with embedded software development and debugging through Vitis. AMD's documented Zephyr support currently focuses on the real-time MicroBlaze V configuration, so engineers should verify support for their operating system and processor configuration in their individual design. Zephyr support can be useful for applications such as industrial control, embedded IoT, robotics, edge systems, and other designs requiring deterministic or real-time software behavior.

PetaLinux and Embedded Linux

AMD PetaLinux provides embedded Linux development workflows for platforms including Zynq and Versal adaptive SoCs. These tools help developers configure boot components, Linux kernels, device trees, root filesystems, and software environments for AMD adaptive computing devices.

AMD Vitis AI

AMD Vitis AI supports AI inference development across compatible AMD adaptive and embedded platforms. As AI expands into machine vision, robotics, industrial inspection, medical imaging, and edge analytics, software frameworks that bridge trained neural-network models and embedded hardware are becoming an increasingly important part of the device ecosystem.

AMD ROCm

AMD ROCm is the company’s open GPU-computing software platform and a core component of its Instinct accelerator and rack-scale AI strategy. The platform supports AI and high-performance computing workloads and forms part of AMD's broader effort to provide an open software environment around large-scale GPU infrastructure. ROCm is also a core component of the Helios rack-scale AI platform introduced in 2026.

AMD/Xilinx Milestones and Industry Impact

AMD and Xilinx have influenced the semiconductor industry through FPGA innovation, heterogeneous computing, AI acceleration, and some of the world’s most powerful high-performance computing systems.

Four Decades of FPGA Innovation

FPGA technology has evolved from small arrays of configurable logic into highly integrated devices containing processors, high-speed connectivity, AI Engines, RF converters, and high-bandwidth memory.

Since the XC2064 shipped in 1985, programmable logic has evolved from relatively small arrays of configurable logic into devices containing processors, high-speed transceivers, hardened memory interfaces, AI Engines, RF converters, millions of logic cells, and high-bandwidth memory.

This development has allowed FPGA and adaptive computing technologies to move from relatively specialized logic applications into communications infrastructure, automobiles, factories, medical systems, satellites, data centers, AI systems, robotics, and high-performance computing.

Artificial Intelligence

AMD addresses artificial intelligence across data centers, PCs, embedded systems, edge devices, robotics, and adaptive computing through several complementary processor and accelerator architectures.

Instinct GPUs address large-scale training and inference in the data center. EPYC CPUs provide host and general-purpose compute. Ryzen AI processors bring dedicated AI processing into PCs. Versal AI devices combine AI Engines with adaptive logic and embedded processing. Kria targets edge and physical AI. Radeon and Radeon AI PRO products provide additional GPU-based AI capabilities. This heterogeneous portfolio enables AMD technologies to address AI workloads across multiple levels, from embedded sensors and robots to rack-scale clusters.

High-Performance and Exascale Computing

AMD EPYC processors and Instinct accelerators have become major technologies in high-performance and exascale computing systems.

HPC systems require extreme processing capability alongside memory bandwidth, power efficiency, interconnect performance, and software scalability. AMD EPYC processors and Instinct accelerators have been deployed in multiple major supercomputing systems, including exascale machines.

AMD and the El Capitan Supercomputer

El Capitan is an exascale-class HPE Cray supercomputer at Lawrence Livermore National Laboratory, powered by AMD MI300A accelerated processing units and 4th Generation EPYC processors.

LLNL lists more than 11,000 batch nodes and over one million CPU cores across the system. El Capitan is designed to support demanding scientific and national-security workloads, particularly the computational requirements of the U.S. stockpile stewardship program. El Capitan previously held the No. 1 position on the TOP500 list, but the June 2026 TOP500 ranking moved El Capitan to No. 2 following the debut of China's LineShine system. El Capitan nevertheless remains an exascale-class supercomputer and recorded 1.809 exaflop/s on the HPL benchmark in the June ranking. Its importance to AMD extends beyond any temporary ranking. El Capitan demonstrates how AMD processor and accelerator technologies can be combined at an enormous scale for some of the world's most computationally intensive workloads.

AMD/Xilinx Applications

AMD/Xilinx technologies are used across AI, data centers, communications, industrial automation, robotics, automotive, aerospace, healthcare, media, test equipment, storage, and low-latency computing.

Artificial Intelligence and Machine Learning: AMD technologies support AI training, inference, generative AI, computer vision, edge AI, AI PCs, robotics, and physical AI. Different workloads may use different combinations of EPYC CPUs, Instinct GPUs, Ryzen AI processors, Versal adaptive SoCs, Radeon GPUs, or Kria platforms.

Data Centers and Cloud Computing: AMD technologies support server processing, cloud computing, virtualization, AI infrastructure, networking, storage, HPC, and acceleration in modern data centers. EPYC CPUs and Instinct accelerators address conventional and AI-intensive compute, while Alveo and Versal devices can support specialized networking, storage, and acceleration workloads.

High-Performance Computing: AMD CPUs, GPUs, and adaptable accelerators offer a range of architectures for computationally intensive scientific, engineering, research, and simulation workloads. Scientific simulation, research, engineering, energy, genomics, weather modeling, and other HPC applications can require massive computational throughput. AMD CPUs, GPUs, and adaptable accelerators provide several architectural approaches depending on whether a workload benefits most from general-purpose computation, GPU parallelism, or customized hardware pipelines.

Networking and Communications: AMD programmable devices support networking and telecommunications through high-speed interfaces, packet processing, signal processing, acceleration, and reconfigurable data paths. Kintex, Virtex, Zynq, Versal, and Alveo technologies can provide high-speed interfacing, packet processing, signal processing, hardware acceleration, and reconfigurable data paths.

5G and Wireless Infrastructure: AMD Zynq RFSoCs and Versal RF devices combine programmable computing with high-speed RF capabilities for wireless and direct RF-sampling systems. Wireless systems require intensive signal processing and increasingly wide RF bandwidth. AMD Zynq RFSoCs and Versal RF devices integrate programmable compute with high-speed RF capabilities, helping reduce the number of separate processing and conversion components required in certain radio architectures.

Industrial Automation: AMD FPGAs, adaptive SoCs, Kria modules, and embedded processors support deterministic control, flexible I/O, robotics, machine vision, networking, and industrial IoT applications. Industrial systems require long product lifecycles, deterministic operation, flexible I/O, and support for specialized control interfaces. FPGAs, Zynq adaptive SoCs, Kria modules, and embedded processors can be used across machine control, factory networking, robotics, motor control, inspection, data acquisition, and industrial IoT.

Robotics and Physical AI: AMD technologies support robotics and physical AI systems that must combine sensor processing, AI inference, motion planning, control, and real-time response. Real-world systems may need to process camera and other sensor data, perform inference, plan movements, control actuators, and respond within strict latency requirements. AMD's expansion of its Kria AI and robotics platform in 2026 reflects the importance of this emerging physical-AI market.

Automotive and ADAS: AMD adaptive computing technologies support automotive and ADAS systems requiring sensor processing, AI acceleration, embedded control, and high-performance data processing. Versal AI Edge devices in particular target increasingly complex edge systems that combine processing, AI acceleration, and programmable hardware.

Aerospace and Defense: AMD FPGAs and adaptive SoCs support aerospace and defense applications that require configurable hardware, high-performance signal processing, and deterministic operation. Applications include radar, electronic warfare, communications, avionics, sensor processing, mission computing, and space systems.

Medical and Healthcare: AMD programmable computing technologies support medical imaging, diagnostics, laboratory equipment, ultrasound, endoscopy, and other healthcare systems requiring real-time processing. High-speed signal processing and real-time image processing can make FPGA and adaptive computing architectures particularly useful in imaging systems.

Image and Video Processing: FPGA parallelism allows AMD programmable devices to process image and video streams with high throughput and deterministic latency. Applications include machine vision, broadcast systems, inspection, video analytics, image enhancement, sensor processing, and professional AV.

Broadcast and Professional AV: AMD FPGAs support real-time broadcast and professional AV applications, including video capture, switching, AV-over-IP, displays, and professional cameras. Newer devices, such as the Kintex UltraScale+ Gen 2, continue to target high-bandwidth video and professional AV applications.

Test and Measurement: AMD programmable technologies support test and measurement systems that must interface with evolving protocols and process high-speed digital or RF data. Versal RF, Zynq RFSoC, Kintex, Virtex, and other programmable technologies can support data acquisition, RF measurement, semiconductor test, protocol analysis, and instrumentation.

Storage Acceleration: AMD programmable accelerators can offload specialized storage functions such as encryption, compression, data movement, computational storage, and memory acceleration from host CPUs. Programmable accelerators and newer Versal technologies provide options for implementing specialized data paths without relying entirely on host CPUs.

FinTech and Low-Latency Computing: FPGA-based acceleration is valuable in electronic trading and other financial applications where predictable, extremely low processing latency is critical. Because FPGA logic can implement highly deterministic data-processing pipelines in hardware, FPGA-based acceleration has long been relevant to low-latency financial systems. AMD also identifies FinTech as an application for the Alveo V80 accelerator.

AMD/Xilinx Product Lifecycle, Obsolescence, and Discontinued Components

AMD/Xilinx lifecycle management requires monitoring exact part numbers because legacy products may remain in service for decades, even as individual devices, packages, or grades are discontinued.

As AMD continues introducing new generations of processors and programmable devices, older Xilinx technologies remain installed in a vast amount of existing equipment. For engineers and procurement professionals, product lifecycle management can be just as important as understanding current devices. A product discontinuation can affect far more than one component. Changing an FPGA may require a PCB redesign, firmware changes, new qualification testing, software updates, regulatory recertification, or complete system revalidation. As a result, some organizations continue to source legacy components long after manufacturers have shifted their development resources toward newer generations.

AMD/Xilinx Product Discontinuations

AMD/Xilinx discontinuation notices should be evaluated at the exact part-number level because an end-of-life decision may affect only specific devices, packages, grades, or configurations rather than an entire family.

Semiconductor manufacturers may discontinue products due to declining demand, supplier changes, aging fabrication technology, legacy test equipment, packaging constraints, or difficulty maintaining older manufacturing infrastructure. When a manufacturer announces an end-of-life decision, the notice normally establishes dates for actions such as last-time buys and final shipments. These notices must be reviewed carefully. A discontinuation may apply to an entire family, a particular temperature grade, one package, or only selected device-package combinations. That distinction is especially important for Xilinx products because a statement such as "Virtex-7 is discontinued" can be inaccurate even when specific Virtex-7 part numbers have reached end of life.

Discontinuation of Legacy CPLD and Spartan Families

AMD discontinued multiple legacy CPLD and Spartan families through Product Discontinuation Notice XCN23009 because of declining run rates and challenges in maintaining legacy test infrastructure.

AMD Product Discontinuation Notice XCN23009, revised January 29, 2024, discontinued commercial/industrial XC and automotive XA products.

AMD attributed the decision to declining run rates and challenges in maintaining legacy test infrastructure. Many of the affected part numbers were listed with no direct replacement. These products had been used across generations of industrial, communications, automotive, medical, and embedded equipment. Their discontinuation does not necessarily eliminate demand. Instead, users of equipment designed around these devices may have to locate remaining inventory or evaluate the cost of redesigning the system.

Selective Virtex FPGA Discontinuations

AMD has discontinued selected Virtex-4, Virtex-5, Virtex-6, and Virtex-7 device-package combinations, but these notices do not mean every product in those families is discontinued. AMD's XCN23005 provides an important example of why lifecycle status should be checked at the exact part-number level. The notice does not state that every Virtex-4, Virtex-5, Virtex-6, and Virtex-7 FPGA has been universally discontinued. Instead, it identifies specific device-package codes from those families that were discontinued because of declining or zero run rates.

AMD listed no direct replacement for the affected part numbers included in the notice. For buyers, this means that lifecycle research should begin with the complete manufacturer part number, including package, speed, and temperature grade information, where applicable.

Legacy Xilinx Platform Flash Configuration PROMs

Legacy Xilinx systems may depend on discontinued Platform Flash configuration PROMs in addition to the FPGA itself, creating another potential source of obsolescence risk.

Xilinx Platform Flash PROM products were designed as in-system programmable configuration memories for loading configuration data into programmable devices. The family included devices such as XCF01S, XCF02S, XCF04S, XCF08P, XCF16P, and XCF32P. Various Platform Flash products have been discontinued over time, including selected products covered by XCN20013 due to supplier line discontinuations. This creates an additional challenge for legacy-system maintenance. An organization may still have access to the FPGA required for a design, but struggle to locate the configuration memory needed alongside it.

These components, therefore, belong within an obsolescence discussion rather than AMD's current product portfolio.

AMD/Xilinx Product Longevity

AMD has extended planned availability for 7 Series FPGAs and adaptive SoCs through 2040 and UltraScale+ FPGAs and adaptive SoCs through 2045.

Legacy product discontinuations should not be interpreted as AMD abandoning long-lifecycle programmable products. AMD has extended the planned availability of 7 Series FPGAs and adaptive SoCs through 2040, and of UltraScale+ FPGAs and adaptive SoCs through 2045. Long lifecycle commitments are especially important for industries such as industrial automation, medical equipment, transportation, aerospace and defense, and infrastructure, where a system may remain in production or service for a decade or more.

Even with extended manufacturer support, supply-chain planning remains important. Individual packages, grades, or products may have different lifecycle circumstances, and clients should monitor manufacturer notices throughout a program's lifecycle.

Sourcing AMD/Xilinx Components

Sourcing AMD/Xilinx components may require access to both current production technologies and legacy parts that have been installed in systems for decades. Some clients need current-production parts for new builds. Others need to maintain existing equipment built around obsolete Xilinx CPLDs, Spartan FPGAs, Virtex devices, configuration memories, or other legacy parts.

Source AMD/Xilinx FPGAs, Adaptive SoCs, Processors, and Components

Microchip USA sources current, obsolete, end-of-life, and hard-to-find AMD/Xilinx components for production, maintenance, and long-term supply requirements.

The appropriate sourcing strategy depends on the component's lifecycle status, availability, required quantity, target price, production schedule, and long-term demand.

Managing AMD/Xilinx Obsolescence

AMD/Xilinx obsolescence management can help organizations maintain legacy systems when replacing an FPGA or programmable device would require significant redesign, qualification, or certification work. Obsolescence management is especially important when a component cannot simply be replaced with a newer device. Changing a legacy Xilinx FPGA may affect board layout, pin assignments, voltage requirements, firmware, toolchains, timing, qualification, and software. For regulated or high-reliability products, redesigning can also trigger additional validation or certification requirements.

When a redesign is difficult, sourcing the original manufacturer part number may provide organizations with additional time to continue production, maintain existing equipment, or plan a controlled transition. We can source obsolete and end-of-life AMD/Xilinx components, evaluate inventory opportunities, and support long-term requirements.

Cost Savings and Hard-to-Find Components

Microchip USA sources hard-to-find AMD/Xilinx components and identifies potential cost-saving opportunities when allocation, shortages, lead times, or regional inventory differences disrupt traditional supply channels. Current components can also become difficult to procure due to allocation, extended lead times, market shortages, changing demand, minimum order requirements, or regional inventory differences.

As a full-line independent distributor, we search beyond traditional supply channels to identify availability and cost-saving opportunities.

Building a More Resilient AMD/Xilinx Supply Chain

A resilient AMD/Xilinx supply chain requires planning for component availability across the full product lifecycle rather than reacting only after shortages or discontinuations occur.

The AMD/Xilinx portfolio demonstrates why supply-chain planning must account for the entire component lifecycle. A newly introduced FPGA may remain available for many years, while a legacy product could become difficult to obtain long before the equipment using it is retired. Building resilience requires understanding both current inventory and future requirements. Organizations can strengthen their sourcing strategies by monitoring product-discontinuation notices, identifying single-source dependencies, forecasting long-term requirements, securing inventory before shortages develop, and evaluating multiple sourcing options.

Source AMD/Xilinx Components With Microchip USA

Microchip USA sources current, obsolete, and hard-to-find AMD/Xilinx components to help clients respond to shortages, support legacy equipment, secure long-term inventory, and identify potential cost savings.

From the earliest Xilinx programmable logic devices to today's AMD adaptive SoCs, processors, and accelerators, the AMD/Xilinx portfolio spans decades of semiconductor innovation. That history also creates a complex supply chain. Clients may simultaneously be designing with the newest AMD technology while maintaining products that depend on FPGA or CPLD families introduced many years ago.

As a full-line independent distributor of electronic components, we source current, obsolete, and hard-to-find AMD/Xilinx components. Whether you need to locate an end-of-life Xilinx FPGA, secure long-term inventory, respond to a component shortage, or identify potential cost-saving opportunities, we can help you build a more resilient sourcing strategy. Search our AMD/Xilinx inventory or request a quote below.