Palo Alto – June 22, 2025 – Imagine stepping onto the polished concrete floor of a brand-new Silicon Carbide (SiC) wafer fabrication plant on the outskirts of Shanghai in early 2024. Crated equipment still retains its shrink-wrap, but the furnaces are already glowing, and crates of pristine 6-inch and 8-inch wafers are stacked nearly to the ceiling. The factory’s owners are proud and even confident. They understand the same scene is unfolding in half a dozen other fabs across China, Europe, and the U.S., and they are determined to succeed.
They did not realize that a capacity boom was already underway. It is driving an equally dramatic price decline, reshaping the entire power semiconductor narrative.
A flood of crystal and a plunge in price
At the start of 2023, a 6-inch SiC epitaxial wafer could fetch well over $600 on the open market. By the end of that year – and just as those shiny new fabs were ramping up – prices had already tumbled to $400–$450, a 25–33% drop that pushed many suppliers below their own cost of production. The slide was even steeper for the emerging 8-inch wafers: once the pride of next-generation roadmaps, their sticker price collapsed by more than 60% in twelve short months.
Why? Because those glowing furnaces were multiplying far faster than demand. Global nameplate output now hovers near 3 million wafers per year, while realistic, yield-adjusted demand sits closer to 2 million. And the build-out isn’t over; projects already funded could lift capacity above 5 million wafers a year by 2027. Manufacturers, desperate to win sockets, have responded with price cuts that sometimes feel more like a freefall.
The ripple effect: GaN feels the squeeze
Walk two blocks down the tradeshow aisle, and you’ll find startups committed to an alternative wide bandgap process: Gallium Nitride (GaN). A year ago, their booths buzzed with confidence. GaN, they claimed, owned everything below 650 V: fast, efficient, and drop-in compatible with silicon fabs. Additionally, GaN can integrate with CMOS logic for combined power and control on a single die, a claim that SiC cannot make.
However, the SiC price war has redrawn those boundaries almost overnight:
- EV onboard chargers and powertrains, SiC’s largest single prize which according to various sources is projected to reach between $5.8 to $8.4 billion in 2025, are now an option for both 400 V and 800 V systems.
- Industrial power supplies that once favored GaN’s high switching speed are now evaluating SiC modules priced within pennies of GaN transistors, driven by the 25–29% CAGR trend lifting the entire segment.
- Even the hyperscale datacenter architects, who have long been enamored with GaN for 48 V front ends, are quietly testing SiC in the 900 V bulk stages of next-generation AI racks.
Yet GaN is far from being cornered. Nothing surpasses a GaN IC for packing 100 W into a pocket-sized wall brick in the bustling consumer charger pavilion. Its silicon heritage helps keep costs low and supports high volumes along with small form factors – areas where SiC’s ruggedness simply isn’t sufficient.
Markets swell despite the turmoil
Zoom out, and the prize is swelling quickly. Analysts estimate the combined SiC and GaN power device market at $1.4 to $2.2 billion today, racing toward $11 to $24 billion by 2034 – a compound growth rate of roughly 23 to 27 percent each year. The very oversupply that troubles producers is, paradoxically, expanding SiC’s market potential and hastening adoption in renewables, mobility, and high-performance computing.
What comes next?
Back inside that Shanghai manufacturing facility, managers are already weighing the costs of idling furnaces against the risks of losing design wins. Some will hesitate; a few may merge; more than one might go bankrupt. Meanwhile, GaN engineers are refining higher-current HEMTs, and system designers are increasingly combining SiC and GaN on the same power board: SiC for the heavy lifting, GaN for finesse.
In other words, the story is far from over. But for now, the lesson is clear: When capacity floods the market, prices do not fall gradually. Instead, they crater, rewriting the rules of competition and forcing every player – from EV makers to phone charger brands – to rethink which process truly powers the future.
We have seen this movie many times in the semiconductor industry. While it is surprising that new processes are undergoing this supply/demand shift, the semiconductor industry continues to cycle. In the long term, cycling fosters breakout growth. In the short term, however, it can be painful for some and incredibly rewarding for others.
We will report further on this topic as it evolves. Please write to us with comments, suggestions, or questions.