Gasgoo Munich-"We can already produce good chips, but deploying them effectively at the system level demands deep coordination and joint development between device makers and customers."
At a YASC Technology Launch Event in Shenzhen on October 15, Li Gang, vice president of strategic operations at YASC, laid out a new challenge facing China's silicon carbide (SiC) industry in an interview with Gasgoo.
Under the theme "Above the Peak: A Four-Year Technological Long March," YASC unveiled a suite of new offerings: next-generation high-performance SiC MOSFET chips, automotive main-drive power modules, AI data center power devices, and a high-performance trench technology platform.
Yet more telling than the product launches was Li's broader read on China's SiC competitiveness: domestic chips have broken through on certain performance metrics, but when it comes to application solutions, local players still have a long way to go.
For years, the story of domestic SiC was defined by technological breakthroughs, capacity expansions, and import substitution. Today, local suppliers have secured sizable market share in new energy vehicles, with some products now matching—or even beating—their international rivals on performance.
But making a good chip doesn't guarantee customers will use it. And winning orders doesn't ensure sustainable profits.
The next phase of competition in China's SiC industry is no longer just about who builds the better chip.
Making the Chip Is Just the First Step
YASC's latest technology roadmap offers a window into this domestic progress.
According to the company, its 1,200V SiC MOSFETs have hit volume production: the fourth-generation PG4E and fifth-generation PG5E, with room-temperature specific on-resistance (Rsp) of 2.2mΩ·cm² and 1.9mΩ·cm², respectively. The next-generation PG6N pushes that figure down further to 1.7mΩ·cm².
Image source: YASC
At the same breakdown voltage, a lower specific on-resistance typically means achieving similar conduction performance on a smaller chip area. Yet as power density climbs, the demands on heat dissipation, short-circuit tolerance, and long-term reliability only intensify.
Li distills the past four years into two milestones: a breakthrough in chip fabrication, and a shift from copycat product development to self-defined design.
The DeepXE™ embedded PCB power module, unveiled at the event, is a product of that latter shift.
Unlike conventional packaging, the DeepXE™ integrates an AMB ceramic embedded unit with a PCB structure, re-architecting the power loop and thermal path. YASC claims the design cuts volume and weight by 50% and switching losses by 20% compared with a selected conventional package.
Those improvements, however, don't guarantee proportional gains across the entire electric-drive system. More importantly, the new module remains some distance from true volume production.
YASC noted that its relevant chips have passed AEC-Q101 certification, and its conventional power modules have cleared AQG 324. However, the DeepXE™ has so far only completed performance and concept verification, along with preliminary reliability testing. Full reliability validation, consistency and stability improvements, and industrial verification aren't expected until the first half of 2027.
A clear time gap remains between a technological breakthrough and scaled product deliveries.
Across the broader Chinese SiC industry, this dynamic is hardly an isolated case.
Data from Gasgoo Research Institute shows that from January to July 2026, Infineon led China's NEV passenger-car SiC power module market with a 14.3% share. BYD Semiconductor trailed at 12.2%, while STMicroelectronics and SMEC held 9.7% and 9.5%, respectively.
More notably, domestic suppliers now occupy six of the top ten spots by installation volume.
Chinese SiC is no longer just a lab-bound alternative; some suppliers have already built scale-ready supporting capabilities.
Yet companies sit at different stages of development. For those already entrenched, the priority is improving cost structures and boosting product margins. For those still gaining a foothold, the struggle is to lock in steady orders and secure spots on more vehicle platforms.
Import substitution buys market entry; commercial competition dictates whether a company stays.
From Capacity Races to Manufacturing Efficiency
Once in the market, manufacturing cost becomes an unavoidable hurdle.
In 2025, YASC's over 20 billion yuan Wuhan plant officially began production, taking just 21 months from groundbreaking to volume line.
At the event, the company further disclosed that CP test yields for its 5mm×5mm SiC MOSFET chips now exceed 98%.
These metrics showcase manufacturing progress, but they don't imply uniform yields across all products. Chip size, process flows, and test conditions vary—yield data isn't directly comparable.
Larger wafers theoretically boost per-wafer chip output and cut unit costs, but only if substrate quality, process yields, equipment efficiency, and line utilization improve in lockstep.
Notably, the SiC market in 2026 is showing a shift from the relentless price declines of previous years.
After a period of aggressive capacity expansions and price adjustments, 6-inch SiC substrate prices are stabilizing and even rebounding, with certain specs tightening in supply. Infineon and onsemi, meanwhile, have adjusted their own product quotes.
Image source: YASC
Still, localized price hikes don't signal a full-blown industry recovery.
Supply-demand dynamics vary across substrates, epitaxial wafers, devices, and power modules. Whether price improvements translate into profit growth depends on a company's product mix, order stability, and manufacturing costs.
Nor does the risk of heavy-asset manufacturing simply vanish.
In September 2025, U.S. SiC player Wolfspeed completed a financial restructuring, cutting total debt by roughly 70% and annual cash interest expenses by about 60%.
The case stems from a specific, highly leveraged capital structure and can't be directly applied elsewhere. But it proves a point: even with advanced tech and scaled capacity, capital investment, market demand, and operating cash flow must stay in balance.
For a company like YASC, scaling up, smart manufacturing, and process innovation are the main levers to pull down costs. Its embedded products, for instance, use copper plating on chip surfaces, reducing reliance on traditional precious metals and related process expenses.
Such improvements carry real value, but manufacturing advantages ultimately must be cashed in through stable production and sustained orders.
Scale can dilute costs—provided there are enough orders to sustain it.
The Real Barrier Lies Beyond the Chip
Manufacturing muscle and cost control solve the problem of steady supply. For the auto industry, though, a harder question remains: are customers willing to take the risk?
YASC already has multiple chips and power modules in the testing phase with Tier 1 clients.
Why does validation take so long even after a chip enters mass production?
Because an SiC power device isn't a simple drop-in replacement.
In a main-drive inverter, chip performance is just the baseline. Drive circuits, busbar layouts, thermal management, electromagnetic compatibility, and fault protection all shape the final system output.
Li offers an example.
Using the same PG4E chip, a conventional HPD module delivers roughly 430A of output current; switch to the new embedded package, and that figure hits 500A—a 16% gain.
The chip didn't change. The packaging and system design did.
This shows that the same device can unlock vastly different performance depending on the integration scheme. That said, these figures come from internal testing and don't represent a performance ceiling. The new embedded modules still require longer customer validation cycles.
Which brings the conversation back to Li's point about application capabilities.
Li readily admits that Chinese SiC firms truly need to shore up their application-solution capabilities built around the device.
Mature international power semiconductor suppliers don't just hand over a datasheet—they provide reference designs, drive schemes, system simulations, failure analyses, and long-term quality support.
These competencies are forged through countless client projects and years of mass-production experience.
For automakers and Tier 1 suppliers, swapping in a new vendor means re-matching systems, re-running reliability validations, and absorbing potential project delays. Even if a domestic chip is cheaper and specs better, customers must weigh whether those gains offset the switching costs.
YASC has begun trying to translate application know-how into product capability.
Li revealed that the company is exploring ways to use embedded technology to integrate drive circuits, peripheral devices, and even some control functions directly into the PCB, offering customers a more complete solution.
This signals that power semiconductor competition may stretch from device performance into system integration. The approach is still exploratory, but the mindset has shifted: it's no longer just about selling a chip—it's about lowering the engineering barrier for the customer to use it.
For Chinese SiC players pushing deeper into mainstream European, American, Japanese, and Korean auto supply chains, such capabilities are equally critical.
Li boils future SiC competitiveness down to three pillars: manufacturing capability, product R&D, and a solid foundation in applications.
Only by combining all three can a company hold its ground in the next stage of competition.
AI Data Centers: A New Value Frontier for SiC
Beyond autos, AI data centers are emerging as the next battleground for Chinese SiC firms.
For YASC, four strategic business pillars have taken shape: consumer electronics, industrial and energy, new energy vehicles, and AI and SST.
Among them, solid-state transformers (SST) stand out.
While new energy vehicles offer massive deployment volumes, power devices contribute a relatively small share of total vehicle value. High-power conversion equipment like SSTs, by contrast, promises far higher per-system device value.
For SiC players, the appeal of AI data centers isn't just potential shipment volumes—it's the chance to reshape their product value structure.
This opportunity has a technical backdrop.
In May 2025, Nvidia proposed an 800V high-voltage DC power architecture for next-generation AI data centers, slated to support higher-power-density AI rack systems around 2027.
As AI server power consumption climbs, higher-voltage DC supply could eliminate certain conversion stages, opening new application windows for wide-bandgap power devices like SiC.
YASC's newly launched PG4E discrete device, along with its EZ2C and EZ3C power modules, targets exactly these high-voltage DC power and SST scenarios.
Still, higher value potential doesn't mean the market is mature.
Li notes that data centers actually running 800V high-voltage DC architectures remain scarce. SiC's application outlook in this space is broad, but actual demand has yet to fully materialize.
By YASC's own progress report, industrial and consumer products are shipping in volume, and AI data center products are in normal supply—though customer applications remain in validation. The newly released EZ2C and EZ3C modules still need to clear market testing and customer integration.
Moreover, even if the 800V architecture gains traction, not every power conversion link will default to SiC. Silicon-based devices, GaN, and SiC will continue to compete on voltage, power, efficiency, and cost.
SSTs may open a new value frontier for SiC, but whether they become a steady revenue stream hinges on the actual rollout of new power architectures and customer procurement cycles.
New markets won't let companies bypass the old barriers of system design, reliability, and customer coordination.
Chinese SiC firms are juggling multiple tests at once: the auto market demands stronger application capabilities and broader customer reach; manufacturing must translate scale into cost advantage; and AI data centers require patience for new power architectures to commercialize.
None of these challenges can be solved by tech specs alone.
China's SiC industry has moved past merely proving it can make the chip. The next step is proving not just whose product can replace an import, but why a customer would keep choosing it—and whether that choice delivers real commercial returns.
