X-Fab looks to buy fabs as it sells out for the next three years
Nick Flaherty talks to Rudi De Winter, CEO of European foundry X-Fab, at the end of a roller coaster year on the plans for 2023.
X-Fab is a leading European semiconductor foundry making specialist high voltage chips for customers, and has been shifting away from the consumer business to automotive and industrial and medical markets.
Despite the downturn and the chip shortages, the demand is such that the existing capacity is sold out for the next three years, CEO Rudi De Winter tells eeNews Europe. This is despite capacity investment plans that started over a year ago.
The company has six fabs around the world, including Lubbock in Texas, acquired from Texas Instruments in the 1990s and in Sarawak, Malaysia from a merger with 1st Silicon, and is looking for more.
“The business has really changed because of the shortages,” said De Winter. “People are putting much more focus on the supply chain and recognising that semiconductors are strategic.”
“In the past our business was based on forecasts that were not binding so there was not the security of supply and security of business on our end. Now we see there is a lot of demand and we are signing a lot of long term agreements with a commitment to deliver and a commitment to buy quantities typically for three years so we are expanding quite a lot and those expansions are already sold.
“And it’s all allocated,” he said. “70% of the CMOS business is on long term agreements with larger customers. Based on those agreements and the outlook we should be sold out for the next three years.”
The problem is on the supply from the fabs and that is a result of long lead times for semiconductor equipment, he says.
“We are expanding in all our factories and investing more than ever so over the next three years that adds up to $1bn to support the long term agreements through advanced payments,” he said. “What I hope from the long term agreements is to take out the large swings in the semiconductor industry because of the long lead times and factory closures”
“The equipment lead time is still very long, up to two years, so the expansions are a long term plan,” he said. “It takes almost three years to ramp up. We started with that more than a year ago.”
One way around that is to acquire fabs, “We see very strong demand for automotive so we are permanently searching for further expansion,” he said. “We are looking at fabs that work well for the new things that we are developing which is BCD SOI 110nm HV so if we expand we want fabs that can produce that. And some of the 200mm fabs are not suitable.”
GaN
Perhaps surprisingly the company is a growing supplier of gallium nitride (GaN) for consumer power adapters but is moving this into the automotive market for on-board vehicle chargers. This is growing alongside the silicon carbide (SiC) and silicon on insulator (SOI) process technologies.
“We are very excited on the positioning of X-Fab with automotive and this is a very strongly growing segment,” said De Winter. “We have always been focused on integrating more high voltage than anyone else in the industry, We started with high voltage CMOS and the link in the last couple of years to SiC is closing the loop.”
“Today we focus on CMOS and SOI, that’s 80% of our revenue, but the SiC and GaN is growing,” he said. “We have some proprietary customer specific technology that we are running in volume for high efficiency small chargers. We started with that engagement seven years ago and that is running in high volume – the reactors are provided by partners and we start processing after the epitaxy for GaN.”
“We are prototyping and developing GaN-on Si on 200mm wafers and have several customer engagements. We are probably the most advanced on 200mm but in the near future that should become available,” he said.
These agreements also favour the automotive and industrial markets with long term plans. Half the business is in automotive and 22% in industrial.
“We are in the process of reducing our consumer business which will further decrease to around 10% from 20%, with this being replaced by higher value medical and industrial business which will grow as the capacity is really needed so we don’t mind that decline,” he said.
MEMS
Specialist micromachined sensors (MEMS) and photonics are also key areas for growth.
“Medical MEMS is growing well and one of our strengths is the combination of CMOS and combining that with MEMS, that gets a lot of traction in medical and we are pursuing that intensively,” he said.
“Typically these projects start on 200mm wafers at Erfurt and go to Itzehoe, near Hamburg, for further processing noble metals that are not suitable in a CMOS fab. You sometimes need to adapt the CMOS to work with the process and optimising all this in one go is more efficient for our customers than searching for different suppliers for the different elements.”
In 2021 the company announced a deal with Ligentec that would see the company be the largest photonics foundry.
“On photonics we implement the Ligentec their platform and are the foundry for them. This is opening up a wide variety of new opportunities for us and I wouldn’t be surprised if there is quantum computing in there,” he said.
Silicon Carbide
The company has been developing SiC on 150mm wafers and moving to 200mm at Lubbock.
“The appetite for long term rolling agreements is still there and we have long term agreements with our wafer suppliers, that’s not new,” he said. “For the time being supply looks OK.”
“Today we run 3000 SiC wafers a month and we are ramping up every month to double the output by end of next year and double again by the end of 2024 – that’s the kind of expansion we are planning.” That’s over 6000 wafers for 2023 and 2024 should be 12000 a month.
“We have over 90% yield from 600V to 3.3kV but we are prototyping higher voltages,” he said. “It’s custom devices and custom technology, so every customer has their own technology that we implement.”
It has been working on trench MOSFETS that are close to the final stage of qualification, following on from planar SIC transistors. “We are excited about the performance of the planar MOSFETs and the difference with the trench FETs is not so big,” he said.
Sustainability
The sustainability of the fabs is also important.
“We are following that closely. It’s something with a lot of dynamics, Everyone is searching for their path and roadmaps,” he said.
“If you want to go to a zero CO2 footprint that is very challenging for a semiconductor fab. A lot of people say they have a roadmap but that is based on buying certificates but in my view that is cheating. Buying certificates will work for a while but will not bring us to zero emissions in 2050. This is the challenge for the industry and for X-Fab.”
“It also depends where the factory is located. In France there is a lot of nuclear power, in Malaysia there is a lot of hydropower and while we can focus on reducing our consumption electricity is the responsibility of the society and whether they want to burn gas. Solar power doesn’t work 24/7 you cannot continuously power your clean room with solar energy and that’s a problem,” he said.
“But there are ways to optimise the power consumption of clean rooms and cut it by 20 or 30% so you need a zero CO2 supply 24/7.”
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
