GlobalFoundries to Sell 300mm New York Fab to ON Semiconductorby Anton Shilov on April 22, 2019 6:45 PM EST
- Posted in
- ON Semiconductor
GlobalFoundries and ON Semiconductor on Monday signed a definitive agreement for the latter to buy GlobalFoundries’ 300-mm fab in East Fishkill, New York. In addition to the production facility, ON Semiconductor will get a team of experienced engineers from GlobalFoundries as well as a technology transfer, development, and license agreements.
The total consideration for the Fab 10 takeover is $430 million. GlobalFoundries received $100 million today upon the signing of the agreement, whereas $330 million will be paid at the end of 2022 when ON Semiconductor will gain full operational control of the fab. Employees of the facility will also join ON Semiconductor. Meanwhile, ON will start using the fab’s services in 2020. Such a long transition period is not unusual for the industry, as in this case it will allow GlobalFoundries to finish filling orders from its current customers.
GlobalFoundries first received Fab 10 as part of its acquisition of IBM's microelectronics assets in 2015. The fab is used to process 300-mm wafers using various technologies, including 45nm and 65 nm technology nodes (as well as their 40 nm and 55 nm versions). Under the terms of the agreement, these technologies will be licensed to ON Semiconductor and will be the basis for the latter’s upcoming process technologies.
At present ON Semiconductor uses fabs that process 200-mm wafers, so the engineers from GlobalFoundries will help the company to transit it process technologies from 200-mm to 300-mm wafers.
Overall this is the latest move by GlobalFoundries to slim down the company as it pivots towards producing with more specialized process nodes. After Thomas Caulfield became CEO of GlobalFoundries last March, the company ceased development of bleeding-edge manufacturing technologies. Then the company started to consolidate its production capacities, selling one of its 200-mm fabs in Singapore (Fab 3E) to Vanguard as part of exiting MEMS business.
For its Fab 3E as well as Fab 10, GlobalFoundries will get $666 million in total (with $336 million in 2019). The company will use the proceedings to fund its specialized production technologies.
"ON Semiconductor is an ideal partner for GlobalFoundries and this agreement is a transformative step in our journey to build GlobalFoundries into the world's leading specialty foundry,” said Tom Caulfield, CEO at GlobalFoundries. “This partnership enables GlobalFoundries to further optimize our assets globally and intensify our investments in the differentiated technologies that fuel our growth while securing a long-term future for the Fab 10 facility and our employees.”
- GlobalFoundries to Sell 200-mm Fab 3E to Vanguard, Exits MEMS Business
- GlobalFoundries and Chinese Authorities Reconsider Plans
- GlobalFoundries Establishes Avera Semiconductor: a Custom Chip Company
- GlobalFoundries Stops All 7nm Development: Opts To Focus on Specialized Processes
- Change of Strategy: A New GlobalFoundries CEO in Dr. Thomas Caulfield
Post Your CommentPlease log in or sign up to comment.
View All Comments
Kevin G - Tuesday, April 23, 2019 - linkHuh, I took a look at the foundry list and oddly both Global Foundries and IBM (though only as a broker) have this fab on the list. I would presume this is to aid with the IBM -> GF transition and I would expect both IBM and GF to move to similar broker status when ON Semiconductor takes over.
Also amusingly there is a "NSA Microelectronics Solution Group" without any contact information, because of course there wouldn't be.
errorr - Tuesday, April 23, 2019 - linkThey have to have a wholly owned subsidiary with an independent Proxy-board that is composed of just US citizens. That is generally how most of the foreign-owned defence contractors work. Even major ones like BAE which are major contractors in our most trusted allies.
Sivar - Tuesday, April 23, 2019 - linkON Semi already has trusted foundries in the U.S., and I can see why they are "trusted" as IMO they go to great lengths to keep them secure.
It would seem this facility just adds one more to their existing USA-based trusted facilities.
Smell This - Tuesday, April 23, 2019 - link
I'm thinking this is what's left of the former IBM East Fishkill campus that Big Blue sold off for Big Bucks several years ago. Slick move by Mubadala Investment. They keep the cutting-edge IP, and transfer advanced processes to GloFo in Malta, Singapore and Dresden.
Yojimbo - Tuesday, April 23, 2019 - linkAs I don't believe GlobalFoundries are investing in building or expanding other fabs, this reflects a lower expected demand for their services. It may be a smart financial decision but I don't think it's some sort of win for them. They wouldn't be selling the foundry without replacing it if their business were growing.
As far as the IP, I have no idea how that will work but it sounds like the fab will transfer over with its staff intact. So I would guess ON has access to the IP necessary to use the workers' expertise.
Smell This - Tuesday, April 23, 2019 - linkMalta is adding 900+ jobs. You swung. You missed.
Yojimbo - Wednesday, April 24, 2019 - linkI'm not talking about their jobs, I'm talking about their production capacity. They have space in Malta because they cancelled their advanced finFET plans.
They laid off 900 employees last year, btw.
drunkenmaster - Wednesday, April 24, 2019 - linkCutting edge only stays cutting edge for years, maybe months. They bought this out from IBM in what 2015, IBM had a lot of IP but with where the industry is actually going both EUV and other things weren't ready back then and lots has been achieved and researched in the last 4 years, that information only has so much relevance, they've also paid a huge price for it and production wise they've gained nearly nothing from it. The singapore fabs are mostly 200nm and all much larger nodes, none remotely suitable for being upgraded for cutting edge nodes. Dresden is still making older nodes and was never ported to 14nm because those fabs while pretty damn good are still fairly old. The Malta fab is really the only truly modern day up to date fab you would go forwards with cutting edge designs and Global have stopped with 14nm. While 7nm is by many accounts, done and ready to go, it's the financial issue. For 7nm to pay off they have to invest billions in equipment, billions more in EUV, billions more in 5nm and then they need the volume of more than one fab to have any chance to make it pay off financially. They've basically drawn a line in the sand at 14nm and said we can't be profitable if we go forwards, so that IP is pretty worthless to them if they aren't moving forward with nodes. Stating they are sticking to 14nm means they basically stopped expanding and stopped investing, which means they were going to start to look to sell off all assets it can. Malta is likely to be sold eventually, the site has pre-approval for two more similar sized fabs on site that should be able to be built quicker than planning for 2 new fabs from scratch so I would bet they will sell Malta to whoever is interested in having 3 high output fabs in the states then sell off everything else they can and wind down the company.
If they don't want to expand or move to new nodes, they basically put the for sale sign in the window.
peevee - Monday, April 29, 2019 - link"Global have stopped with 14nm"
They have "12 nm". Close to "14 nm" technologically, but still denser as evidenced by new Zen cores made on it. Sorry, I have to use quotes as these are very, very fake "nm" (unlike 65nm node).
I wonder what use 65nm has these days. Simple microcontrollers? What is the point if something like A32 on "12 nm" is going to cost about the same (in cents), as energy efficient and have immensely more compute power and programmability for networking etc?
FreckledTrout - Tuesday, April 23, 2019 - linkPretty sad GloFlo bought the brain trust from IBM and decides to go legacy process only. I think this is a direct result of that stupid decision. They might as well sell everything off they as they slowly wither over the next decade. As TSMC and Samsung move to newer process nodes they leave behind legacy nodes for cheap which except for the really old 90+ nm stuff they will likely eat up most of GloFlo's customers.