ASIC vs FPGA Cost: When to Switch Production Volumes
Deciding between a Field-Programmable Gate Array (FPGA) and an Application-Specific Integrated Circuit (ASIC) is a critical financial milestone in electronics manufacturing. This article explores the exact production volumes where designing a custom ASIC becomes more economical than purchasing off-the-shelf FPGAs. By breaking down Non-Recurring Engineering (NRE) costs, per-unit pricing, and the mathematics of the break-even point, this guide provides a clear framework for hardware developers and decision-makers to determine when to make the transition from FPGA to ASIC.
Understanding the Cost Structures: NRE vs. Unit Cost
The economic comparison between FPGAs and ASICs comes down to upfront costs versus long-term per-unit pricing.
FPGAs have virtually zero Non-Recurring Engineering (NRE) costs regarding silicon manufacturing. You buy them off the shelf, program them, and place them on your circuit board. However, the per-unit cost of an FPGA is high because you are paying for extra silicon needed to make the chip reprogrammable. Depending on complexity, a single FPGA can cost anywhere from $10 to several thousand dollars.
ASICs have the exact opposite financial model. Designing a custom chip requires massive upfront NRE costs. These include buying Intellectual Property (IP) blocks, Electronic Design Automation (EDA) software licenses, engineering salaries, and the creation of lithography masks at a semiconductor foundry. Total NRE costs for an ASIC can range from $100,000 for simple, older-node designs to over $50 million for advanced nanometer nodes. However, once the NRE is paid, the per-unit cost of an ASIC is extremely low—often just a few dollars or even cents per chip.
The Break-Even Volume: By the Numbers
As a general rule in the electronics industry, an ASIC becomes more economical than an FPGA at production volumes between 50,000 and 100,000 units. However, this number shifts drastically based on the complexity of the chip and the technology node used.
Low Complexity (Older Nodes: 65nm to 180nm)
- Estimated NRE: $150,000 – $500,000
- Break-Even Volume: 10,000 to 50,000 units
- Use Case: IoT devices, simple sensors, and basic microcontroller replacements.
Medium Complexity (Mature Nodes: 28nm to 45nm)
- Estimated NRE: $1 million – $5 million
- Break-Even Volume: 50,000 to 250,000 units
- Use Case: Consumer electronics, automotive components, and medical devices.
High Complexity (Advanced Nodes: 7nm and below)
- Estimated NRE: $20 million – $100 million+
- Break-Even Volume: 500,000 to several million units
- Use Case: Smartphones, AI accelerators, and high-performance computing.
How to Calculate Your Exact Break-Even Point
To find the exact volume where an ASIC makes financial sense for your specific project, use the following formula:
Break-Even Volume = Total ASIC NRE Cost / (FPGA Unit Cost – ASIC Unit Cost)
For example, imagine you are building a product using an FPGA that costs $35 per unit. You get a quote from a design house stating that a custom ASIC will cost $1,000,000 in NRE, and the resulting ASIC will cost $5 per unit to manufacture.
- ASIC NRE: $1,000,000
- Unit Cost Savings: $35 (FPGA) - $5 (ASIC) = $30 per unit
- Calculation: $1,000,000 / $30 = 33,333 units.
In this scenario, your 33,334th unit is where the ASIC officially becomes more profitable than the FPGA.
Hidden Economic Factors Beyond Unit Volume
Sometimes, companies switch to ASICs at lower volumes (even under 10,000 units) due to secondary economic factors that impact the total Bill of Materials (BOM) and product viability:
- Power Consumption: ASICs are significantly more power-efficient than FPGAs. Lower power consumption means you can use smaller, cheaper batteries and eliminate expensive cooling solutions like heatsinks or fans, reducing the overall product cost.
- Size and Integration: An ASIC is much smaller than an FPGA. Furthermore, an ASIC allows you to integrate analog components, memory, and other peripheral chips onto a single die. This shrinks the Printed Circuit Board (PCB) size, saving money on enclosure materials and board manufacturing.
- Supply Chain Security: Relying on a single FPGA vendor leaves you vulnerable to price hikes and chip shortages. Owning your ASIC tooling gives you more control over long-term production pricing.
The Middle Ground: FPGA-to-ASIC Conversions
If your volume falls in the gray area—say, 20,000 to 50,000 units—you can utilize "Structured ASICs" or FPGA-to-ASIC conversion services ( offered by companies like Microchip or Intel). These services take your existing FPGA code and hardwire it into a partially pre-manufactured silicon wafer. This approach lowers the NRE cost to around $50,000 to $200,000 and cuts the per-unit price in half, shifting the economic break-even point to much lower production volumes.