Silicon or SiC: Where the Break-Even Line for Power Devices Actually Sits

Explainers · Published 2026-10-10 · Updated 2026-10-10

The SiC case cannot be made on unit price alone; switching loss, thermal volume, whether the drive and protection chain must be redesigned, and second sources set where the line falls.

The decision to move to SiC usually gets compressed into "higher efficiency", then stalls in the BOM review, because the few dollars saved on the device do not cover the drive, protection, EMI and re-qualification work around it. The dividing line is not a voltage class. It is three questions: does your system actually need a higher switching frequency to buy volume, can the gate drive and protection chain keep up, and does a second source still exist after the switch. Answer those honestly and many projects find the right move is not to switch at all — or to switch only in selected positions.

What the material buys is frequency and volume, not efficiency

The SiC case has to be derived from the material. A higher breakdown field means the drift region can be thinner at the same blocking voltage, which changes the structure of both conduction and switching loss. The consequence is not a few points of efficiency — it is the ability to push switching frequency up, and only once frequency rises do magnetics and heatsinks get smaller.

That is the real selling point: same power, less volume and weight. For onboard chargers, PV inverters and energy-storage converters, where volume, weight and thermal headroom are binding constraints, that value converts into system cost. For an industrial power supply that already has room for a heatsink, the same device may buy nothing but a few watts of unused margin.

So the first question is not whether SiC is good. It is whether your system is constrained by volume or heat. If it is not, a material upgrade has nowhere to show up.

A cheaper device is not a cheaper system

The second layer of cost is the one most often missed: change the material and the surrounding circuit changes with it. SiC gate behaviour differs from silicon, so drive voltage, gate charge and switching speed all need re-matching. Short-circuit withstand time is shorter, so protection must be faster and tighter. Switching is faster, so dv/dt raises EMI, isolation and insulation questions that have to be re-answered.

Added together, these often cost more than the device. Novosense in the directory is positioned around exactly this link, with isolated gate drive as a product line — install a new power device but keep the old drive and it will run at the old rhythm, with the benefit unrealised and reliability degraded.

The third layer is qualification: system efficiency, temperature rise, EMC and lifetime all get re-tested. For anything soldered to a board and certified, this layer is almost always underestimated.

When it does not pay: low voltage and low frequency

Silicon remains the cost-optimal answer at low and medium voltage and low frequency. The reason is simple: those applications are not volume-sensitive but price-sensitive, and SiC wafer cost and yield are not yet at a point where they compete head-on with silicon. Putting SiC into consumer power supplies, low-voltage motor drives or ordinary adapters usually spends money on performance nobody needs.

The applications that do pay share three traits: medium voltage to high current, a hard constraint on volume or weight, and enough annual volume to amortise qualification. Remove any one and the arithmetic has to be redone.

Topology matters too. The switching-loss advantage of SiC shows up more clearly in hard-switched designs; in a converter that already uses soft switching to suppress loss, the gain is diluted. Look at the topology before the device table.

Who supplies: why IDM matters more on this line

Power devices couple process and device tightly, so the IDM model is worth more here than in logic. Device structure and process are tuned together, parameter distributions and yield stay under control, failure analysis closes the loop, and long-term supply is easier to commit to.

The directory splits cleanly. onsemi runs from MOSFETs positioned in the mid- and low-voltage mainstream, through IGBT and IPM for industrial and automotive, to SiC power devices positioned for automotive, and is recorded as a leader in power and sensing analog. Silan covers IGBT, MOSFET, IPM modules and LED drivers as a full power line, recorded as an IDM-mode power and sensor vendor. CR Micro pairs MOSFET, IGBT and power ICs with its own packaging and test service — keeping test in-house is itself a promise about lot consistency.

Above them sits Infineon, recorded as the automotive semiconductor leader for six consecutive years, supplying automotive power and MCU from the same house and therefore leaving the carmaker with fewer interfaces to manage. The lesson: SiC competition is not only about device parameters but about process control and supply structure.

Test	Lean silicon	Lean SiC
System constraint	Volume and thermal headroom remain	Volume or weight is binding
Operating frequency	Low frequency, or soft switching already applied	Higher frequency needed to shrink magnetics
Surrounding cost	Drive and protection should stay untouched	Willing to redo drive, protection and EMI
Supply structure	Needs very low unit price and many sources	Accepts fewer vendors but demands long commitments

What to do

Decide whether to switch based on volume and thermal constraints first, then rank by efficiency; the reverse order turns the project into a parameter exercise.
Budget the drive, protection, EMI and re-qualification cost alongside the device. Comparing unit prices alone badly understates the change.
When moving to SiC, evaluate isolated gate drive and protection in the same pass; leave them unchanged and the benefit never materialises.
Prefer IDMs with real process control, or channels tightly bound to them. Lot consistency matters more than peak parameters in power devices.
Confirm second sources and long-term supply commitments, particularly for automotive and storage applications with long qualification cycles.
For comparison, review the vendor structure in the discrete and power category and the analog and interface category.

Data and scope

Product lines and positioning come from ICNAV company records (generated 2026-10-28); statements such as "automotive semiconductor leader for six consecutive years" reflect publicly reported figures captured in the directory. No blocking voltage, on-resistance, efficiency or price figures appear here. The material and topology arguments are qualitative; for actual designs, use device datasheets, simulation and measurement. Nothing here is a purchasing recommendation.

Written by ICNAV 编辑部 · ICNAV editorial team. References: the ICNAV company directory and public industry sources.