These posts have been on a hiatus for the past few months to devote resources to project work. Hopefully, this channel can get back to a regular cadence. Let's start with a relatively easy follow-up.
Since publishing "Can SiC hop a ride on EV growth?" almost a year ago, several readers reached out to discuss the topic further. After a handful of engagements and many more insightful discussions, it's worth expanding on the three takeaways in that initial note.
First, the forecasts for EVs are becoming more bullish and are crossing the threshold of 50% global penetration over the next fifteen years. Achieving that threshold means electric vehicles (EVs) must transition from a high-end premium offering to the mainstream consumer segment.
Second, EVs still cost more than conventional internal combustion engine (ICE) vehicles. It is hard to see how SiC, which is ~3-5x more expensive than silicon, can create downstream value if EVs are starting at a deficit to ICEs. Indeed, SiC offers economic benefits that offset its costs. Inverters using SiC have higher efficiency, and the vehicle requires less battery capacity, which can offset its added costs. But as long as EVs are more expensive than ICEs, bargaining leverage will eat away at that value. Those competitive forces are sharper on semiconductor suppliers, who typically have higher profitability than the rest of the automotive supply chain.
Lastly, RCD Advisors thinks a more likely way for SiC technology to break into the mainstream EV segment is through a vertically integrated supply chain. A vertically integrated supply chain can evaluate design tradeoffs better than a merchant supplier who depends on their downstream customers for market and technology signals.
So, where are we today? Let's review the 2024 landscape for SiC and the EV industry.
Slowing Momentum
xEV production boomed in 2023 to approximately 14M units. But the momentum is slowing as the premium segment saturates and mainstream consumers (particularly in the US) remain hesitant. In response, an EV price war has emerged. This environment isn’t conducive to a growing SiC market.
In the near term, it's not a big deal. Only premium vehicle models use SiC, and pricing is relatively inelastic in this segment. It's in the long term that forecasts could diverge. Most forecasts assume aggressive SiC penetration into mainstream and entry-level vehicle sectors. That is possible, but RCD Advisors' long-term forecast scenario includes a possible (probable) “luxury ceiling.” That is, only luxury vehicles have the pricing flexibility to adopt SiC.
More Candor About EVs and IGBTs
There has been a lot more candor about EVs and SiC. In a moment of brutal honesty, Maria Barra, G.M.'s CEO, last year admitted at an investor conference that EV costs were still not at the point where you can get to the mass market, which is a $30k-$40k vehicle in the US. Most semiconductor suppliers have begun emphasizing the continuing importance of IGBT technology. There is now much more discussion of minimizing SiC usage through Si/SiC fusion and auxiliary motors.
At last year's investor event, Tesla drove much of the discussion when it opened the door to using less 75% less SiC per vehicle. As further indication that Si IGBTs are not dead, in early May 2023, Denso announced a collaboration with United Semiconductor Japan (a foundry) to produce IGBTs on 300mm wafers for EV applications. There have even been some interesting new IGBT technology developments. IGBTs' incremental improvements (like asymmetric super junctions; see Ideal Semiconductor) could make these devices (modestly) more competitive with SiC.
SiC Capex Announcements Raise the Stakes
In 2023, SiC Capex announcements exceeded $15Bn (from company announcements). Whether these plans happen or not remains to be seen. Some of the funding for this Capex will come directly from sourcing agreements where automakers gain favorable pricing in exchange for the upfront investment. Sourcing agreements reduce some of the risks for the SiC supplier. But even with sourcing agreements, let's not mask the size of these bets. Assuming optimistic operating profits (~20%) and industry average returns on invested capital (11% for power discrete semiconductors), a $15Bn investment would require 36% CAAGR to 2030. Even the most optimistic forecasts fall short of that scenario.
Government subsidies will take on some of that risk. There were two joint venture announcements in 2023 between European and Chinese organizations. Infineon announced a joint venture agreement with TanKeblue, a starting substrate maker. And STMicro announced a joint venture with Sanan. Both collaborations are undoubtedly capitalizing on subsidies from the Chinese government to help offset the costs and reduce substrate and device pricing. But these companies are not the only opportunists. Wolfspeed is also anticipated to receive US and European CHIP Act subsidies to help fund expansions (they have already received subsidies from New York State).
It is difficult to justify industry Capex investments if the SiC market is constrained to the premium market, even if subsidies and sourcing agreements are included. Moreover, it is equally challenging to justify Capex investments if pricing erodes enough so that the technology is able to break through to the mainstream EV market.
The counterargument is that Capex investments are an attempt to gain a “first mover” advantage in a "Winner takes all" arms race. The first supplier to achieve scale and improve substrate yields (for 200mm wafers) will reap a competitive cost advantage. That advantage will allow the winner to enter the mainstream EV market and increase its profit pool.
Perhaps. But it seems like an increasingly implausible scenario using reasonable metrics for learning curves. The market size for SiC is constrained by its pricing erosion as well as its volume. A learning curve model can model a $9Bn market in 2030 by assuming low price declines (i.e. high learning coefficient) combined with high EV penetration. The analysis arrives at a lower market value in 2030 with more realistic learning coefficients and EV penetration.
This simple spreadsheet exercise has a large lever on Capex spending. It essentially describes a merchant market where the investment economics seem "upside down." Unless, of course, this isn’t a just a bet on innovation, but rather, like much of the rest of the EV market, a bet on subsidization.
Vertical Integration Goes Mainstream
The original note strongly advocated for vertical integration in the SiC supply chain. Indeed, suppliers have started to organize in that direction. In early April 2023, Wolfspeed and ZF announced a joint innovation lab combined with an investment to build a fab in Germany. ZF is quasi-integrating by collaborating on joint SiC and inverter design. Second, at around the same time, Bosch acquired the assets of TSI Semiconductor to integrate backward and produce SiC for EV production in the US. Third, in late 2023, Mitsubishi and Denso announced that they would acquire a combined 25% equity stake in Coherent’s SiC substrate business. Finally, throughout 2023, BYD has ramped up SiC to serve its internal demand. BYD’s vertical integration is likely to be a significant competitive advantage compared to competitive merchant supply chains.
Other consulting firms (link, link) have hopped on the bandwagon. For these folks, vertical integration is a way to remove layers of profit margin in a multilevel supply chain while also offering better supply chain resiliency. One firm mentions the potential for yield improvement from better feedback with the downstream inverter designer.
Having similar recommendations is encouraging. However, RCD Advisors' thesis is a bit different. In our assessment, vertical integration allows for coordination and free flow of information to make rational Capex investments. Complex market dynamics and design choices (800V bus, LPF batteries, vehicle weight, torque, driving range, etc.) are properly understood inside a single entity. These design choices impact the threshold pricing for SiC adoption differently. And it is much easier to coordinate all the moving parts inside one entity.
Moreover, winning in the merchant SiC market isn’t just about stellar execution on yields and efficiencies. It also requires your customers to execute, and their customers to execute too. This fact is also important to peripheral components and material suppliers who are planning to ride the SiC growth wave.
Vertical integration puts each entity under one umbrella, combining profit pools and investments. Information at each stage of the supply chain can control the main investment valve. It would avoid the possibility of over-investment, where every SiC supplier chases the same pool of premium EV buyers when counting their design wins.
No doubt, in most industries, merchant suppliers are more efficient. They react faster to external market signals from different sources. They develop experience faster than integrated operations that are often mired in bureaucracy. But merchant SiC suppliers don't have that advantage. Every major merchant SiC supplier depends almost entirely on the EV market. They are also forging close supply relationships and casting their lot with a limited pool of tier-ones and automakers. SiC suppliers are already integrated on the demand side for all intents and purposes. But except for some supply agreements (and some government subsidies), they are exposed to most of the downside risks on the investment side.
Some Suppliers Won’t Have a Choice
We have to be sober about how the industry structure could actually evolve. It's all too easy to offer strategic clichés about vertical integration. Tier-ones and automakers are making their own bets on EV. Realistically, they are not sitting on the finances to acquire SiC semiconductor suppliers. A more likely scenario is that these organizations will scavenge SiC assets after the shake out. This scenario may be the optimal end game because it would allow the auto industry to use outside capital to fund a part of the EV transition. To be clear, this isn't some sinister collusion in play. It is an investment bubble in the making.
Fortunately, there is a lot of hard-learned guidance on how to manage bubbles. They mostly boil down to either amassing enough resources to persevere, or getting out before it bursts. Weathering through a bubble is basically a solvency test. Getting out requires identifying the off-ramp. Maybe all these forged supply agreements and collaborations are not just funding sources after all. They could also be exit strategies.
If capitalizing on the SiC transition in EV is critical to your business, RCD Advisors can help. Contact us to learn more about the practice.
Disclaimer: The content in this note is for informational purposes only, and should not be relied upon as legal, business, investment, or tax advice. Any projections, estimates, forecasts, prospects, and/or opinions expressed in this note are subject to change without notice.