ASIC vs FPGA: How to Choose the Right Architecture for Real-World Products

The question of whether to design a product around an FPGA or commission an ASIC is one that comes up early in almost every serious electronics programme. It rarely has a simple answer. Both technologies have strengths and drawbacks. Both are widely used in demanding applications, and the choice between ASICs and FPGAs is not always an either/or decision. The decision often comes down to where you are in the product lifecycle, what your performance requirements are, and what the programme looks like over its full commercial lifecycle.

This guide works through the key differences, the situations in which each architecture makes sense, and how to think about the choice in the context of real product development (particularly in the automotive, industrial, aerospace, and defence markets, where the engineering requirements are most demanding).

What is an FPGA?

A Field Programmable Gate Array (FPGA) is an integrated circuit whose internal logic is configurable after manufacture. The device contains an array of programmable logic blocks and interconnects that can be wired together using a binary configuration file, allowing the designer to define the circuit’s behaviour in hardware without committing to a fixed design in silicon.

FPGAs are primarily digital devices. Their configurable fabric is built around logic gates, flip-flops, and memory blocks, and they are typically programmed using hardware description languages such as VHDL or Verilog. Some modern FPGAs include embedded hard processor cores, DSP (Digital Signal Processing) blocks, and high-speed transceivers, which extend their capability for complex signal processing applications.

Reconfigurability is the defining characteristic of an FPGA, and the source of both its primary strength and its primary limitation. The same physical device can be reprogrammed to implement different functions, which makes it extremely useful during development and prototyping. 

However, it is important to be clear about what an FPGA can and cannot do on its own. Because FPGAs are fundamentally digital devices, applications that require analogue signal conditioning, precision measurement, or integrated power management cannot be delivered by an FPGA alone. In practice, FPGA-based systems typically pair the FPGA with a collection of additional ICs to cover the functions the FPGA cannot handle. It is often this complete system – the FPGA plus its supporting components – that is later optimised and consolidated into a single ASIC. The choice between the two architectures is therefore not always a straightforward either/or decision.

What is an ASIC?

An Application-Specific Integrated Circuit (ASIC) is an integrated circuit designed from the ground up to perform a defined function in a defined application. Unlike an FPGA, which is a general-purpose programmable device adapted to a task, an ASIC is engineered precisely for the signal types, performance levels, environmental conditions, and power budget of the specific product it will be used in.

Unlike FPGAs, which are primarily digital devices, an ASIC can implement both digital and analogue functions in the same device. For sensor-based applications, this is the critical distinction. A mixed-signal ASIC integrates amplification, filtering, analogue-to-digital conversion, digital signal processing, and communication interfaces into a single piece of silicon, optimising the noise, losses, and component count that come from connecting discrete devices together. It is this mixed-signal capability that makes ASICs the natural solution for the signal chains found in automotive, industrial, and aerospace applications.

The trade-off is development investment upfront. An ASIC requires a design and verification programme before the first silicon is produced, and mask costs are committed at tape-out. For programmes with sufficient volumes and a clear performance requirement, that investment is recovered quickly and the long-term economics are usually better than an FPGA-based solution. It is also worth noting that FPGAs are subject to the same obsolescence risks as any catalogue component: a specific device can be discontinued by its manufacturer, potentially forcing a costly redesign mid-programme. An ASIC, with supply managed directly and process transfer options available, removes that dependency. The question is whether the programme justifies the upfront investment. As we will explore, the answer is often yes earlier in the product journey than engineers expect.

Key Differences: Cost, Performance, Flexibility, and Power

Cost

The cost comparison between ASICs and FPGAs is one of the most misunderstood areas in product engineering. This is partly because people sometimes focus on the wrong numbers and miss the economy of scale which is often required to justify an ASIC business case. At low volumes (say, hundreds or a few thousand units) FPGA unit costs may be the most competitive solution. At tens of thousands of units, however, the economics start to shift towards an ASIC.

At hundreds of thousands or even millions of units – numbers which are typical for automotive and industrial production – the difference in cost between an FPGA-based solution and an ASIC drives the business case for an ASIC. 

There is also the BOM (Bill of Materials) effect to consider. An ASIC that integrates the functions of several discrete components reduces not just the cost of those components but the assembly cost, PCB (Printed Circuit Board) area, and test complexity of the overall product.

This means that the upfront non-recurring engineering (NRE) investment required to develop an ASIC can often be recovered quickly, and the total cost of ownership for an ASIC solution can be substantially lower.

Performance

An optimised ASIC outperforms an FPGA in a directly comparable application. This is a consequence of how each solution is built, not a limitation of any particular FPGA device.

An FPGA’s logic is implemented using look-up tables and programmable routing. This introduces propagation delays, limits the precision of analogue functions, and means the device is consuming power to maintain its configuration even when not actively processing. An ASIC can implement the same function directly in optimised gates, on the CMOS (Complementary Metal-Oxide-Semiconductor) process node that matches the application’s speed and power requirements, and integrate analogue circuitry designed precisely for the signal levels and noise environment of the application.

The performance gap is most significant in three areas: analogue signal conditioning (where FPGAs are inherently limited), power efficiency (where ASICs can be dramatically more efficient), and latency-critical applications (where the deterministic timing of a purpose-built design is difficult to match with programmable logic).

Flexibility

FPGAs are more flexible than ASICs. They can be reconfigured, which has value in applications where functionality needs to evolve or where different configurations are required across product variants. For programmes where that flexibility is genuinely needed, it is a real consideration.

However, flexibility has a cost: the FPGA carries the overhead of reconfigurability in every unit, whether or not that capability is ever used. For products where the functionality is well-defined and stable, that overhead represents waste in power, cost, and size. An ASIC’s inflexibility is, in those contexts, a feature rather than a limitation. It reflects a design optimised for exactly what is needed.

It is also worth noting that ASICs are not as inflexible as they are sometimes portrayed. Metal layer changes can be used to implement modifications at relatively low cost, and designing for testability and future revision can be planned into the architecture from the outset. Some ASICs also incorporate embedded microcontrollers and non-volatile memory (such as flash) which means that firmware and configuration parameters can be updated in the field without any change to the underlying silicon.

Power

Power efficiency is one of the clearest advantages of the ASIC approach, and it matters increasingly across all three of the markets Swindon Silicon serves. In automotive applications, power consumption directly affects battery life of certain sensors, range in electric vehicles and thermal management in tightly packaged systems. In industrial applications, low-power ASICs enable battery-operated or energy-harvesting sensor nodes. In aerospace and defence, power efficiency translates to reduced thermal load in sealed enclosures and extended mission duration.

An ASIC is sized for the performance and functionality it needs. It does not draw the quiescent current of a general-purpose FPGA designed to cover a broad range of applications. At scale, this difference is measurable in system-level efficiency and, ultimately, in product competitiveness.

Real-World Application Scenarios

The ASIC vs FPGA decision looks different in practice depending on the application. Here is how the trade-offs play out across three different markets.

Automotive: Sensor Interfaces and TPMS

Automotive electronics are defined by a combination of requirements that FPGAs struggle to satisfy simultaneously: operation across a wide temperature range (-40°C to +150°C), the integration of analogue signal conditioning for sensor interfaces, and supply assurance over programme lifetimes of fifteen to twenty years.

Tyre Pressure Monitoring Systems (TPMS) are a good illustration. The sensor module must integrate pressure sensing, signal conditioning, RF communication, and power management into a package small enough to fit into a mechanical housing on the back side of the valve, within the tyre cavity. An FPGA-based implementation would be physically too large, too power-hungry for the battery constraints, and unable to integrate the analogue functions without additional external components. An ASIC solves all of these problems in a single device.

Industrial: Smart Sensors and Factory Automation

Industrial applications increasingly demand sensor intelligence at the node rather than centralised signal processing. This means the signal conditioning, calibration, and communication electronics need to be compact, low-power, and robust enough to operate in industrial environments. This often includes extended temperature ranges and significant electromagnetic interference from motors, drives, and switching power supplies.

FPGAs are used in industrial applications where digital flexibility is the primary requirement. For example, in industrial communication protocol bridging or in high-channel-count data acquisition systems where reconfigurability has value. But for sensor nodes and embedded measurement applications, the power, size, and analogue performance of an ASIC-based solution is usually the better fit.

The cost argument is also compelling in industrial production. A sensor product selling in the tens or hundreds of thousands of units per year will typically recover ASIC NRE within one to two years of production, after which the unit cost advantage of the ASIC directly improves margin.

Aerospace and Defence: Navigation, Ranging, and Long-Life Requirements

Aerospace and defence applications present a specific combination of challenges: radiation tolerance or radiation hardness requirements in some cases, long programme lifetimes (often twenty to thirty years for defence programmes), supply assurance, and the need for devices that can be qualified to relevant standards and then frozen in configuration for the life of the programme.

FPGAs are widely used in aerospace and defence, particularly where reconfigurability in the field is genuinely required (for instance, in software-defined radio or in systems that need to be updated post-deployment). But for sensor interface and signal conditioning applications in navigation systems and ranging equipment, the performance, size, and supply assurance characteristics of an ASIC solution are often more appropriate.

The obsolescence question is particularly acute in long-life aerospace and defence programmes. A specific FPGA device that a product was designed around may be discontinued well within the programme’s service life, triggering a costly redesign and requalification. An ASIC, with process transfer and wafer storage options planned from the outset, gives the programme a supply strategy that is not dependent on a component manufacturer’s portfolio decisions.

When to Choose an FPGA

An FPGA is the right choice (or at least the right starting point) in a number of well-defined situations:

1. The product is still in active development and the specification is likely to change. FPGA reconfigurability means that hardware changes can be made without new silicon, which reduces cost and schedule risk during the development phase.
2. Volumes are genuinely low and will remain so. For products selling in the hundreds or low thousands per year, the total system cost differential may not justify an ASIC development.
3. The application requires in-field reconfigurability as a feature (for example, where different firmware configurations need to be deployed after the product has shipped). An ASIC can handle a certain amount of programmability at the software level, but not at the architectural level.
4. The design is primarily digital and does not require integrated analogue functions that an FPGA cannot deliver.
5. Time to first silicon is the overriding constraint. An FPGA can be configured and tested quickly, while ASIC development takes longer.

What ‘still in development’ means here is worth defining precisely. If the core architecture and signal chain are well-defined even if peripheral features are still evolving, an ASIC development programme can run in parallel with product development rather than after it. Starting the ASIC conversation earlier than feels natural is typically the most efficient approach.

When to Choose an ASIC

The conditions that favour an ASIC become clearer when the full product lifecycle is taken into account rather than just the development phase.

1. Volume production is the target. Once a product is in production at meaningful scale, the ASIC unit cost advantage over an FPGA compounds with every unit shipped.
2. Analogue signal conditioning is required. Mixed-signal ASICs integrate analogue and digital functions in a single device in a way that FPGAs cannot match.
3. Power consumption is a constraint. For battery-powered, energy-harvesting, or thermally limited applications, the efficiency advantage of an ASIC is often decisive.
4. Physical size is limited. An ASIC is optimised for the specific functionality it needs, not for a general-purpose programmable platform. Where space is constrained (as in TPMS sensor modules or wearable devices) this matters significantly.
5. IP protection is important. The design is embedded in silicon and cannot be reverse-engineered from a configuration file.
6. Long-term supply assurance is required. ASIC supply can be managed directly, with wafer storage, process transfer, and second-source options available.
7. The specification is stable. If the product’s core function is well-defined and unlikely to require post-deployment hardware changes, the inflexibility of an ASIC is not a limitation.

How Products Often Transition from FPGA to ASIC

One of the most common routes to an ASIC programme is not a greenfield design decision but a transition from an existing FPGA-based product. This path comes up frequently enough in practice that it is worth understanding how it typically works and when it makes sense to start planning for it.

The typical journey looks like this. A product is developed around an FPGA because it is early-stage, the specification is still evolving, or the team needs to move quickly to market. The FPGA serves the product well during this phase. But as volumes grow, the total cost of the FPGA-based solution becomes a meaningful drag on margins. The power consumption becomes a design constraint. The size of the device limits product packaging options. And the supply risk of a single FPGA part becomes harder to ignore as the programme matures.

At this point, the ASIC transition starts to look attractive. The FPGA implementation has validated the architecture and the signal processing approach, which reduces the risk and timescale of the ASIC design programme. The product requirements are well-understood. And the commercial case of recovering NRE across the volumes already being shipped is straightforward to make.

The key is not to wait until the problems become acute before starting the conversation. ASIC development takes twelve to thirty-six months depending on complexity, so the decision to transition needs to be made when there is still runway to complete the development before the FPGA cost, power, or supply issues reach a critical point.

A good ASIC supplier will be familiar with supporting FPGA-to-ASIC transitions across automotive, industrial, and aerospace programmes. The process begins with a feasibility study that establishes the technical and commercial case: whether the performance requirements are achievable, what the NRE and unit cost economics look like, and what the supply strategy for the programme should be.

Choosing the Right Architecture for Long-Term Success

The ASIC vs FPGA decision is ultimately a product lifecycle question rather than a purely technical one. An FPGA is the right answer when flexibility, speed to first hardware, or genuinely low volumes are the governing factors. An ASIC is the right answer when the programme is heading to meaningful production, when analogue performance or power efficiency are real requirements, and when the product needs a supply and longevity strategy that a catalogue component cannot provide.

The most costly mistake in this decision is not choosing the wrong technology, but making the choice too late. Teams that start the ASIC conversation when the FPGA problems become obvious are already behind schedule for a transition. Teams that evaluate the ASIC case early, even before they are sure they need it, can make more informed decisions and enter production with better economics and a more robust supply position.

For a detailed look at how the trade-offs compare across the full product lifecycle, see our guide to ASIC and FPGA advantages and disadvantages. The Swindon Silicon Systems team can also help you evaluate whether an ASIC is the right step for your programme.

Get in touch today to discuss your programme with the Swindon Silicon Systems team.

Frequently Asked Questions