FPGA vs. ASIC: Similarities, Differences

Field programmable gate arrays (FPGAs) and application-specific integrated circuits (ASICs) are two distinct types of microchips used in a wide range of applications, from medical instruments to consumer electronics, satellites to cell phones.

Ubiquitous in electronic devices since the mid-1990s, these microchips are in the spotlight today as never before, thanks to the passage of the CHIPS and Science Act, which allocates $52.7 billion for America’s semiconductor industry — the single-largest investment in U.S. manufacturing in decades — and similar legislation in Europe.

In light of these recent developments, let’s review some of the key differences between FPGAs and ASICs, then turn to the implications of the CHIPS and Science Act not only for engineering talent, but for engineering talent strategies as well.

“There are big economic trade-offs between ASICs and FPGAs. If you're planning to build a million digital stereos to go in automobiles, for example, you probably want to go with ASICs, which would wind up costing $100 per chip, compared to $5,000 or $10,000 per chip for an FPGA, once you’ve made the initial investment.”

—Peter Parsons, FPGA/ASIC solutions principal at Randstad USA

FPGA vs. ASIC: the key differences

While FPGAs and ASICs are both microchips, there are some key differences between them as well. From a high-level, we might summarize those differences as follows:

An FPGA is a multipurpose microchip. As such, it can be reprogrammed in the field for multiple applications — and “field,” in this context, simply means outside of semiconductor foundries (or “fabs”).
By contrast, an ASIC is designed solely for a specific application; once it has been produced, you can’t reprogram or modify it.

Dialing up the granularity a little bit, below is a more nuanced overview of the differences between FPGAs and ASICs.

configurability
- configurability
- FPGA easily reconfigurable with different designs
- ASIC not reconfigurable — the circuitry is permanent
barrier for entry
- barrier for entry
- FPGA low barrier for entry — no costly production tooling is required
- ASIC high barrier for entry in terms of costs and learning curve
high-volume mass production
- high-volume mass production
- FPGA not suited for high-volume mass production
- ASIC suited for high-volume mass production
power consumption
- power consumption
- FPGA requires a lot of power in order to function
- ASIC far more power efficient — and power consumption can be closely controlled
design flexibility
- design flexibility
- FPGA limited design flexibility (analog design isn’t possible, for example)
- ASIC capable of implementing both analog and digital functionalities
update flexibility
- update flexibility
- FPGA ideal for applications in which occasional updates or upgrades are required
- ASIC not ideal for applications in which the design might need to be updated later on

	FPGA	ASIC
configurability	easily reconfigurable with different designs	not reconfigurable — the circuitry is permanent
barrier for entry	low barrier for entry — no costly production tooling is required	high barrier for entry in terms of costs and learning curve
high-volume mass production	not suited for high-volume mass production	suited for high-volume mass production
power consumption	requires a lot of power in order to function	far more power efficient — and power consumption can be closely controlled
design flexibility	limited design flexibility (analog design isn’t possible, for example)	capable of implementing both analog and digital functionalities
update flexibility	ideal for applications in which occasional updates or upgrades are required	not ideal for applications in which the design might need to be updated later on

“With an ASIC, it’s going to cost you somewhere in the ballpark of $20-50 million to build and develop it — and if you make a mistake, then you're going to spend another $20 million or so fixing it. With an FPGA, if you make a mistake, you can spend six or seven hours reprogramming it, and it doesn't cost you anything.”

—Peter Parsons, FPGA/ASIC solutions principal at Randstad USA

FPGA and ASIC: the talent outlook

While FPGAs and ASICs share a number of similarities, as we have seen, there are also salient differences to keep in mind. Hopefully, you have a clearer sense of both.

But for organizations looking to leverage the momentum and resources of the recently passed CHIPS and Science Act, there might be a more important takeaway: engineering talent challenges are right around the corner. The broad-based scarcity of process, verification and test engineers, in particular, could quickly turn into a major pain point for companies across the board.

Why? The answer, in part, has to do with the way professional development is structured in the field. As Peter Parsons, FPGA/ASIC solutions principal at Randstad USA, explained, “The skill set associated with, say, verification, which is an area where talent is probably the scarcest and most in demand right now, really isn’t taught in college. It’s mostly acquired through on-the-job training.”

That combination of demand and scarcity is changing the dynamics of the talent market, too.

“Right now, we’re seeing companies paying interns $80,000 or more simply to try to convince them to come back when they graduate — and even if they do, they’re going to need two years or so of training to really be able to do the job. But that’s how important these verification engineers are. You’re always going to need more people with verification expertise than design expertise on your team,” says Peter.

Along with the challenge of professional pathways, longer-term educational trends are part of the picture as well. For example, while the number of computer science majors has increased by more than 90 percent in the past 50 years, the number of electrical engineering majors has declined by roughly the same amount during that period. In other words, people with the right backgrounds to design, verify and validate FPGAs and ASICs are already in perilously short supply.

In that context, internal upskilling and reskilling initiatives may be the most effective workaround, particularly in light of the fact that many of the public-private partnerships in this space, like the one between Intel and the National Science Foundation (NSF), tend to be focused on talent development for the semiconductor manufacturing workforce — which might not include engineers. Hence the primacy of upskilling or reskilling.

Peter, for one, is optimistic about the potential for impact: “If somebody has a four-year degree in electrical and computer engineering and a background in C++ or object-oriented programming, we’re confident that we can put them through the equivalent of a master's degree in about 12 weeks, and that from there they would be ready to do the job. This kind of accelerated training is something we’re actively working on right now.”

next steps

With so much momentum in the semiconductor space right now, opportunities and challenges loom on the horizon. But that’s also where Randstad can step in and deliver value, from best-in-class talent development and upskilling programs to contingent, permanent and project-based solutions. Click here to learn more — or schedule a meeting with us today if you’re ready to get started.

Alternatively, if you’re interested in FPGA or ASIC engineering careers, start exploring opportunities with Randstad right now.

tags:

email
print