What’s the (programmable) logic in that?

Welcome to the fourth article in the Supplyframe Commodities Analyzed series that explores programmable logic devices overall and field programmable gate arrays (FPGAs) in particular. With predictive insights about market demand , pricing, lead times and availability,  this article affords real-world analytics  and expert analysis to further understand the sourcing  of FPGAs in H2 2023 and beyond

Demand, pricing, and lead time outlook for H2 2023

The prospect of increasing demand for artificial intelligence (AI) inference servers, robust automotive and factory automation markets, expanding use cases for edge processing, and with FPGAs as key data center components – the programmable logic market is poised to prosper. 

Supplyframe projects the AI cloud and enterprise data center market alone for FPGAs could drive more than 50% year-on-year growth in 2024. Leading low-power FPGA manufacturer Lattice Semiconductor puts its serviceable addressable market at $10 billion in 2028, with half coming from automotive and industrial and 40% from communications and compute.

For the second half of 2023, the pace of programmable logic (including simple and complex programmable devices and FPGAs) sourcing activity growth is expected to be in the high single-digit percentage range, after rising by 8.6% sequentially in Q1 2023 to match the Q1 2022 demand at nearly one and half times the Commodity IQ Demand Index baseline. t. FPGA demand rose 11.3% from Q4 2022 through Q1 2023 to nearly twice the index baseline before retreating by nearly 16% from March through May. 

Through H1 2020, improved lead times were in stark contrast to the last few quarters in which more mature FPGA parts had lead times over 60 weeks – placing a significant burden on procurement teams to forecast demand and place non-cancellable, non-returnable (NCNR) orders with suppliers. FPGA lead times overall collapsed from about two and half times the Commodity IQ Lead Time Index baseline in Q4 2022 to just above one and half times the baseline through the first five months of the year. 

Supply and availability for programmable logic have improved markedly, but some bottlenecks remain. Programmable logic channel inventories, extraordinarily high for many active and passive devices, persisted at one-third of the Commodity IQ Inventory Index in May. Allocations and NCNR order terms will be lifted for most FPGA product families in Q3 2023 but a host of manufacturers are requesting one year of demand visibility. Generally, lead times will decline and in some case be halved through Q3 2023.

Supply and availability for programmable logic have improved markedly, but some bottlenecks remain, including:

• FPGAs belonging to the Intel Cyclone V and IV series are constrained, with market evidence of some order decommitments and deliveries into 2025, pointing to potential part family shortages in Q3 2023. 
• AMD-Xilinx Artix-7 FPGAs are challenged with substrate availability and its Spartan-6 family has improved to 35-week lead times since coming off allocation in March. 
• Microchip FPGA supply remains tight, including its SmartFusion2 devices which are at 60 weeks of lead time. 
• Lattice Semiconductor MachXO2, and MachXO3 FPGAs, and its ispMACH family of CPLDs are supply limited at 48-week lead times. 

What’s the (programmable) logic in that?
As logic density increases, FPGAs will be used in more designs and applications, driving prices down and adding new features, such as embedded processors and DSP blocks. AI, new data center applications, hardware security, and cloud-connected factories are driving FPGA expansions, augmenting existing 5G wireless infrastructure, server, client computing, industrial automation,  and automotive electronics demand.

Automotive

The automotive sector is a significant market for FPGAs due to the adoption and rapid growth of advanced driver assistance systems (ADAS) and in-vehicle infotainment (IVI) over the last decade. ADAS and IVI rely heavily on high-performance processors for complex algorithms and real-time data processing. FPGAs offer a flexible, high-performance solution for these requirements, making them an ideal choice for use in these applications. ADAS uses FPGAs to accelerate the processing of image and sensor data from cameras and radar systems. These systems use computer vision algorithms to analyze the data and detect objects on the road, such as other vehicles, pedestrians, and obstacles. Automotive silicon and IP solutions are used for driver assistance, comfort, convenience, and in-vehicle infotainment.

FPGAs in automobiles are extensively used in LiDAR to construct images from the laser beam. They are  employed in self-driving cars to instantly evaluate footage for impediments or the road’s edge for obstacle detection. Also, FPGAs are widely used in car-infotainment systems for reliable high-speed communications within the car. They enhance efficiency and conserve energy.

Consumer Electronics 

A global shift is taking place that is transforming how homeowners manage the appliances in their homes. It’s a long list that includes air conditioners, washing machines, clothes dryers, dishwashers, refrigerators, freezers, water heaters, kitchen stoves, microwave ovens, Wi-Fi coffee makers, and toaster ovens. Harnessing IoT, artificial intelligence, and apps developed by appliance manufacturers, households can use their mobile phones and voice assistants to access and control their smart home devices from anywhere, 24/7. The application for smart devices extends beyond appliances to home security systems, lighting systems and climate control thermostats.

Healthcare

FPGAs are used in the healthcare medical sector because they can process large amounts of data in real time with low latency and high accuracy. For example, they are used in a wide variety of medical systems, including magnetic resonance imaging machines, to process the signals from the scanner’s coils. They are also used in tomography scanners to perform real-time image reconstruction. A significant advantage of FPGAs in healthcare is their ability to integrate multiple functions into a single device. This capability is crucial in medical applications where space and power consumption are critical factors. For example, FPGAs can incorporate multiple sensors, signal processing units, and communication interfaces into a single device, reducing the overall size and power consumption of medical devices.

FPGAs also offer a high degree of flexibility, enabling developers to modify and optimize medical device functionality even after deployment. This is important for medical devices that must be updated with new features or modified to suit specific patient needs. In addition, FPGAs can be reprogrammed remotely, enabling developers to fix bugs or add new features without physically accessing the device.

Industrial 

AMD-Xilinx FPGAs and targeted design platforms for Industrial, Scientific and Medical (ISM) enable higher degrees of flexibility, faster time-to-market, and lower overall non-recurring engineering costs (NRE) for a wide range of applications such as industrial imaging and surveillance, industrial automation, and medical imaging equipment.

Military & Aviation

As the technologies evolve, government defense organizations are evolving their electronic warfare solutions to counteract the capabilities of their adversaries by leapfrogging them in an ongoing battle to maintain dominance. Electronic warfare systems’ essential functions include acquiring radio frequency signals and performing the required signal-processing tasks to respond effectively. The process involves constantly improving technologies and architectures that boost performance levels.

FPGAs have become commonplace on signal-processing boards for defense applications such as radar and sonar. These include military-qualified and radiation-tolerant FPGAs, intellectual property for image processing, waveform generation, and partial reconfiguration for SDRs. The future for the military is about radiation-tolerant FPGAs matched with intellectual property for image processing, waveform generation, and partial reconfiguration for SDRs.

Space 

Space agencies worldwide have used SRAM-based FPGAs in space for over a decade. They are attractive for space applications because of their in-flight reconfigurability, decreased development time and cost, and increased design and testing flexibility. According to a ten-year-old MIT report, the Xilinx Virtex-5QV was the first commercially available Radiation Hardened By Design SRAM-based FPGA to operate in space.  

However, until recently, few reprogrammable devices have been used in space. For example, the European Spacecraft Agency deemed them problematic because of their sensitivity to involuntary reconfiguration due to Single Event Upsets (SEU) induced by radiation. However, with the advent of reprogrammable devices featuring a million system gates or more, it is no longer feasible to disregard these technologies. 

Making this type of vision a reality is a seismic change that is taking place in the way satellites are designed and built. Today’s space designers seek low-cost plastic equivalents for standard space-grade components. But choosing a commercial off-the-shelf (COTS) product and running an electrical, mechanical and radiation test campaign is risky and expensive. Many consumer off-the-shelf (COTS) products are not sufficiently robust for use in space and therefore pose too much of a risk for satellite designers.

A new product class is emerging to bridge this gap: plastic radiation-tolerant space-level components at much lower costs than their traditional counterparts. Also necessary for the shift is introducing new FPGA suppliers to the space market. 

Telecommunications

The telecommunications sector had the highest revenue share of the industrial FPGA market in 2022 and is expected to maintain its dominant market share for the foreseeable future. FPGAs in the telecom market are forecast to reach $1.37 billion at a CAGR of 5.3% by 2030. The widespread use of FPGAs in the wireless and telecommunication sectors for various applications, such as data packet switching, packet processing, and optical transport networks, is propelling the FPGA market forward. For example, FPGAs provide bandwidth to telecom service providers so they can create compatible networks ranging from 3G to LTE and beyond. 

The proliferation of 5G networks is propelling FPGA growth because 5G provides configurability, flexible hardware accelerators, high-speed switching, and low-latency operation at low cost. FPGAs are replacing traditional circuit systems and are used in many applications for their reconfigurability. Because the telecom sector is focused on expanding network bandwidth, FPGAs are poised for a rapid rise in advanced integrated circuit systems. 

FPGAs are widely employed in communication systems to enhance connectivity and coverage and improve overall service quality. At the same time, they reduce delays and latency, particularly when the data is altered. Also, FPGAs are widely used in server and cloud applications by businesses.

The FPGA market will evolve rapidly as the demand for real-time adaptable silicon grows with next-generation technologies, including Machine Learning, Artificial Intelligence, Computer Vision, and other technologies. The importance of the FPGA is expanding due to its adaptive and programming capabilities, which make it an ideal semiconductor for training massive amounts of data on the fly. FPGA is also focused on speeding up AI workloads. The flexibility, bespoke parallelism, and ability to be reprogrammed for numerous applications are the key benefits of the FPGA to accelerate machine learning and deep learning processes.