On July 13, 2026, Bosch announced that it has begun sample production of silicon carbide (SiC) semiconductors using 200mm wafers at its plant in Roseville, California. The company has signed a direct funding agreement with the US Department of Commerce worth up to $225 million under the CHIPS Act, with total investment in the plant reaching up to $2 billion. However, what has started moving now is the process for producing samples for customer evaluation—commercial production is the next stage, targeted for sometime in 2026. What matters more than the completion of mass-production equipment is that Bosch has entered the phase of building yield and automotive-grade quality in the US.
Roseville is Bosch's first semiconductor manufacturing site in the US. In 2024, the US government indicated that once this plant reaches full operation, it could account for over 40% of US domestic SiC device manufacturing capacity, potentially making it Bosch's largest SiC site globally. But the market environment today differs from when the investment decision was made in 2023. EV growth has fallen short of initial projections, and SiC supply capacity has expanded, particularly in China. The success of this $2 billion plant hinges on whether Bosch can improve yield and utilization in a market with loosened supply-demand balance, and bring prices down to levels customers can adopt.
The Reality Behind "Production Start": Samples for Customer Evaluation
The start of sample production means that Roseville's new manufacturing equipment has been connected and reached the stage where wafers can actually be processed through Bosch's SiC process. Bosch acquired the former TSI Semiconductors plant in August 2023, and has progressed this far in roughly three years. By using an existing facility, the company was able to secure buildings and personnel faster than it could have with a newly constructed plant. Employee count has grown from about 250 during the conversion period to over 300 currently.
Even so, there remains a long distance between samples and commercial products. Automotive discrete semiconductors require reliability testing based on AEC-Q101, including temperature cycling, high-temperature/high-humidity testing, and electrical stress testing. Even when changing manufacturing locations or processes, qualification data must be accumulated according to plans agreed upon with customers. The reason Bosch's announcement distinguished between "sample production" and "commercial production sometime in 2026" is that this process ramp-up and customer qualification still remain.
Regarding the third-generation SiC chips to be manufactured in the future at Roseville, Bosch states they offer up to 20% higher performance than the previous generation with a smaller chip area. However, the company has not disclosed whether this 20% figure refers to on-resistance, power density, or conversion efficiency. Judging the actual value of the products will require waiting for datasheets and customer adoption. The start of plant operations and product competitiveness should be viewed as separate matters.
At 200mm, Area Is About 1.78x Larger—Yield Will Determine Profitability
Roseville's aim lies in the shift from 150mm to 200mm diameter wafers. Comparing circular areas, a 200mm wafer is approximately 1.78 times larger than a 150mm wafer, meaning that for the same chip dimensions, the number of chips obtainable per processing run can be substantially increased. STMicroelectronics has explained regarding its own 200mm SiC development that the usable area is nearly doubled, while the number of good chips increases 1.8 to 1.9 times. Bosch has also moved its Reutlingen, Germany facility to 200mm, aligning both Roseville and Reutlingen on the same wafer generation.
Expanding wafer diameter does not automatically lower costs. SiC is more difficult to manage in terms of crystal defects than silicon, and if an internal wafer defect crosses a chip, that area becomes unusable. Even if larger diameters allow more chips to be produced per run, profits disappear if defect density or process variability worsens. This is why STMicroelectronics cites crystal defects, dedicated equipment, and manufacturing ecosystem as conditions for the 200mm transition.
Bosch's SiC manufacturing involves a mask structure with more than 10 layers, and the process involves over 300 steps. While Roseville had previously manufactured 200mm silicon ASICs, many pieces of equipment needed to be replaced for SiC production. Even though the roughly 130,000-square-foot cleanroom could be reused, the conditions for ion implantation, etching, thermal processing, and inspection are entirely different matters. The advantage of an existing facility lies in avoiding the need to build a new building, not in purchasing SiC yield.
$225 Million Supporting One of the Largest SiC Facilities in the US
The direct funding cap of $225 million represents 11.25% of the planned $2 billion total investment. Bosch has also indicated it will receive a $25 million tax credit from the State of California, as well as utilize the federal CHIPS Investment Tax Credit, which allows a 25% credit on qualified capital investments. While public support reduces startup risk, the bulk of capital expenditure is planned to be borne by Bosch itself.
Comparing this finalized announcement with the preliminary agreement from December 2024 reveals another difference. The preliminary agreement included, in addition to direct funding of up to $225 million, a loan facility of approximately $350 million. Bosch's July 2026 announcement mentions only the direct funding agreement and makes no reference to the loan. While it cannot be confirmed that the loan has been withdrawn, the only confirmed support at this point remains the direct funding.
What the US side is seeking in return is production capacity and jobs. A 2024 US government document projected up to 1,000 jobs during construction and up to 700 jobs in manufacturing and related roles once operational. The plant plans to handle not only wafer front-end processing but also testing, sorting, and dicing. If front-end through some back-end processes can be consolidated at the same site, North American automakers could diversify supply that had been concentrated at Bosch's German plant.
However, domestic US manufacturing and complete domestic self-sufficiency are not the same thing. Bosch has not disclosed the sourcing of SiC substrates, the origin of raw materials, or the locations of all packaging processes at this time. What Roseville reduces is concentration in device manufacturing sites—it does not immediately resolve overseas dependence across the entire supply chain.
Headwinds Against the High-Growth Assumptions of 2023
When Bosch announced the Roseville acquisition in April 2023, the company projected the SiC market would grow at an average annual rate of 30%. In EVs, SiC primarily reduces losses in drive inverters, onboard chargers, and DC/DC converters, allowing for smaller cooling systems and passive components. Compared to silicon semiconductors, Bosch stated that SiC could reduce energy consumption by up to 50% and extend driving range by an average of 6%. For EVs moving toward higher voltage systems, the advantages of SiC remain clear even today.
The question has shifted from whether these advantages exist to whether demand will grow as fast as capacity plans anticipate. SiC specialist Wolfspeed explained in its fiscal year 2025 annual report that slowing growth in EV applications combined with global capacity expansion, particularly centered in China, has created a supply-demand imbalance that has intensified competition especially for 150mm products. The company has announced plans to close 150mm facilities and is consolidating production toward 200mm. Infineon also began shipping 200mm SiC products to customers in Q1 2025. Bosch is not leading the charge toward 200mm adoption, and must compete for customers in a market where rival companies are also advancing toward larger wafer diameters.
In this environment, production capacity size alone does not translate directly into advantage. If customers' vehicle model plans are delayed, utilization rates for expensive equipment drop, raising fixed costs per wafer. On the other hand, if sales channels can be expanded from EVs to data centers and industrial power applications, demand fluctuations can be smoothed out. Bosch cites AI data centers as one such application, but has not disclosed order values or customer composition for Roseville.
Criteria for Judgment After Commercial Production
The first milestone for gauging Roseville's progress is whether it can transition to commercial production sometime in 2026. Following that will be which third-generation products get qualified and to which customers they are shipped. Production volume alone does not reveal profitability. Only when 200mm process yield, equipment utilization, and per-chip costs all improve together will the larger wafer diameter justify the $2 billion investment.
Evaluating the supply network also requires specifics. If the same products can be mutually qualified across both the German and US facilities, customers will find it easier to select sourcing locations by region. Conversely, if product generations or process qualifications are split by facility, having two plants provides little substitutability. The real strength of the supply resilience Bosch speaks of will show up in how much production can be shifted between Roseville and Reutlingen during disasters or demand surges.
Sample production is evidence that a three-year-old acquisition plan has turned into actual wafers. If commercial products pass customer qualification and 200mm yield and utilization reach prices customers demand, Roseville will add a practical SiC sourcing option for US automakers. Only once that point is reached will the $2 billion in equipment and $225 million in subsidies transform from figures for plant construction into a foundation for genuine competitiveness.