Taiwan III-V IC firms gearing up for car LiDAR demand Julian Ho, Taipei; Willis Ke, DIGITIMES Thursday 27 May 2021
Taiwan's III-V semiconductor players including foundry Win Semiconductors and epi-wafer supplier Visual Photonics Epitaxy (VPEC) have cut into the supply chains of major carmakers and China's EV vendors by providing ToF-based LiDAR components, and are poised to enjoy strong demand for automotive LiDAR applications, according to industry sources.
Win Semi and VPEC have started shipments of VCSEL chips as light source for ToF LiDAR sensors to some of the world's top-10 automakers and China EV vendors, the sources said. VPEC reportedly has landed big orders for VCSEL epi-wafers for car-use LiDAR applications, with the order volumes equivalent to total annual amounts recorded in 2020, the sources added.
Taiwan III-V supply chain can now offer high-power VCSEL epi-wafers used in both transmitting and receiving ends of automotive LiDAR devices, allowing a light detection range of 200-300 meters, the sources said.
The sources continued there are at least 70-80 Taiwan makers engaged in the development of LiDAR systems seeking to explore immense business potentials, and they now adopt either edge or surface emitting laser as light source.
Leveraging CMOS image sensors alone can hardly materialize Level 5 fully autonomous driving, which can be achieved only through the collaboration of LiDAR, mmWave radar and traditional image sensors, the sources stressed.
Cwl, thanks alot for the interesting information above. How do you interpret the news, can you help me? Does this broaden the application areas of VCSEL from smartphones to cars as well? What does this mean with regards to demand for Aixtrons manufacturing equipment? Rising orders short-term or mid-term as well?
"...VPEC reportedly has landed big orders for VCSEL epi-wafers for car-use LiDAR applications, with the order volumes equivalent to total annual amounts recorded in 2020, the sources added..."
VPEC has 52 MOCVD's and announced in March that it would be buying 5-10 more this year. A 6" GaAs wafer produces VCSELs that satisfy about 10,000 smartphones but only a few hundred car's LiDARs. If the EPI wafers order volume for LiDAR equals VPEC's total production of 2020, then more would be needed.
that bought OSRAM and a major VCSEL producer is hiring a manger with Aixtron MOCVD experience:
Epi Operations Manager - Singapore ams Singapore, Singapore 14 hours ago Be among the first 25 applicants
Your Tasks And Responsibilities Lead and manage Epi Operations maintenance and process engineers Drive Epi defect reduction and Yield-up initiatives to improve value Collaborate with Production and Plan Equipment Maintenance schedule Transfer existing Epi products and process to receiving plant Work closely with Epi design house on RnD pilots to provide refinement on upcoming leading-edge Epi designs
Your Education And Experiences Masters degree in Materials, Electrical/Electronics EngineeringBachelor degree for candidates with relevant extensive experience will be considered Experience in Aixtron MOCVD tools
CWL, that is indeed a very interesting topic that you raise. My understanding - based on some analyst reports - is that AMS is an Aixtron customer while Osram is using Veeco products. Following the ams osram merger one could see this as an opportunity for Aixtron (ams as the dominant partner trying to achieve synergies by streamlining suppliers/kicking osram suppliers) or as a threat if Aixtron?s ams business will be shifted to Veeco. Third option: it stays as is currently. Of course, I have no clue which scenario might be the most realistic - if you have any view on this, please let us know :-)
I don't have any real insight here. It is very difficult to achieve high yields on VCSEL. Once a customer qualified it would be very hard to switch to a different MOCVD tool. Eventually better tool wins. I know very little about OSRAM other than it was mainly a LED lighting company that tried to move into other areas relating to MOCVD.
"...According to the micro-net news, on June 5th, InnoSec (Suzhou) Semiconductor Co., Ltd. held a mass production and R&D building groundbreaking ceremony in Fenhu High-tech Zone. The first phase of Innosec Suzhou project is expected to invest 8 billion yuan, and the plant construction and equipment will be completed in 2020, and mass production will begin today, becoming the world's first company to achieve mass production of 8-inch silicon-based gallium nitride . Production capacity will gradually climb after the start of production. The production capacity will reach 6,000 pieces/month by the end of 2021. After the project is fully operational at the end of 2022, the Suzhou plant will achieve an annual production capacity of 780,000 pieces of 8-inch silicon-based gallium nitride wafers, with an estimated annual output value of 15 billion yuan. Profits and taxes exceed 1.5 billion yuan..."
Previously I posted in # 663 about Innoscience which is a customer of Aixtron:
"...InnoSec combines the AIXTRON MOCVD ( G5+ ) and 8 -inch CMOS compatible wafer manufacturing line to solve the industrial technology bottleneck of low yield, small capacity and unstable process of compound semiconductor wafer manufacturing, and successfully realized 8 -inch silicon-based For the mass production of gallium nitride devices..."
"Innosecco plans to reach a production capacity of 12,000 pieces per month in Suzhou (the world 's largest 8 -inch production capacity) and Zhuhai plant (the world's first 8 -inch) by the end of this year"
"In addition to international manufacturers such as ASML , AIXTRON , TEL, and South Korea?s SK Group, the scene also saw Chinese local equipment..."
I wrote in #1011: "6000 monthly 8" wafers need 40 Aixtron's G5 assuming one batch per day. If 780,000 annual wafers by the end of 2022 is realized, we are talking about 400 G5's! "
In #1012 the news says: ""Innosecco plans to reach a production capacity of 12,000 pieces per month in Suzhou (the world 's largest 8 -inch production capacity) and Zhuhai plant (the world's first 8 -inch) by the end of this year"
So the total Innoscience GaN-on-Si capacity will actually be 12,000 8" wafers by years end. That would mean 80 G5's. My assumption of run rate was one batch per day. The plant could make two runs a day or whatever depending how thick the GaN layer needs to be.
Even if only 200 G5's would be needed by 2022 by Innoscience alone, we are talking about the total 2022 projected revenue by the analysts already (397m euro).
The SEMI INDUSTRY NETWORK news mentioned contract signing ceremony at the scene. Aixtron among the other equipment companies could be there to sign contracts. Mr. Song the Aixtron China GM was there.
Excellent research as always, CWL! Do i read correctly ? 397m in potential orders from this single customer?
Although I remember from my discussions with the company very well that Asian customers tend to order in large badges (as in 2009/10) to achieve good pricing. So in a worst case scenario we might have to apply a 25% (?) haircut which would still leave it at an astonishing 300m? order potential.
certainly! Bosch is a customer for Aixtron, running Beta tests on the new SIC machines.. I think they disclosed that in the last 3-4 conf calls.. so absolutely, this is positive for Aixtron.
More positive though is a recent interview with the CEO/CFO from Bosch in the Financial Times who said that Auto OEMs have to put "money on the table" to ensure better supply and a more stable Chip order situation in the future.
Clearly more capacity coming in the Chipsector in Europe!
The Semi.org news referenced in post#1012 has been modified. The latest version has taken out the section that has the names of the companies attended. The latest version has gone through two revisions already and certain MOCVD information has now been deleted. I guess someone is monitoring here and there and realized certain information should not be out there. Aixtron should make a press release soon since Innoscience is already making news with Aixtron in there. Look at the price today!
@Alex, agree, that is the news we have been waiting for?even though it has not been a complete surprise (ref. fel?s excellent postings following the latest quarterly results).
Thus, the real good news in my view is the timing (I would have expected such a news only in a month time or with the 2Q report) and in particular the magnitude of the guidance increase! It?s not just a 360 to 380ish increase, but rather an astonishing 360 to 420m! Following the 3Q20 result, the 2021 revenue consensus number stood at 290m, following the 4Q20 numbers it was 340m, following 1Q numbers it was 360m ? now we will probably end up at a consensus number of 420-430m. If you need a definition of an ?inflection point? going forward ? take Aixtron as an example.
As a quick side calculation: 360m revs = 18% EBIT margin 420m revs = 21% EBIT margin => incr. revs of 60m yield incremental 23m of EBIT, thus incremental business comes with an EBIT margin of approx. 40% (selective view, I know).
EBIT consensus as of yesterday (source: marketscreener) stood at 73.5m, the new Aixtron guidance is at 88.2m, i.e. some 20% above. Impressive.