Intel will shut down its automotive business, lay off most of the department’s employees
Posted by LegoGuy23@reddit | cars | View on Reddit | 93 comments
needmoresynths@reddit
Really glad we gave this company billions of tax dollars last year
taylorswiftttttt@reddit
If Intel goes down, the US loses the ability to manufacture leading edge semiconductors, we will never get it back
xsvfan@reddit
And having two desktop/server cpu designers is a good thing. We gave billions to AMD to keep them afloat to prevent an Intel monopoly.
JDragon@reddit
AMD has never been the recipient of government largesse on the level of Intel. AMD literally had to sell their fabs (GlobalFoundries) to the UAE to stay alive.
ocean_800@reddit
AMD is not a fab and we still had intel so
Maximus_Aurelius@reddit
🚀
EZKTurbo@reddit
Nevermind the fact that Nvidia surpassed intel years ago
snollygoster1@reddit
That's blatantly false. AMD, Apple, Nvidia, GlobalFoundries, Qualcomm, Texas Instruments and many other semiconductor companies are US based. Intel could try actually producing products people want.
Dan6erbond2@reddit
The first three use TSMC however which is based in Taiwan and I believe so does Qualcomm.
hypersonicboom@reddit
TSMC has now at least one fab already operating on US soil
Corsair4@reddit
TSMC's cutting edge stuff is not in the US. Their US production lags behind the Taiwan foundries a decent amount.
hypersonicboom@reddit
The stuff that is putting out is still better than Intel's, and is the node used by the products currently utilized by Intel and AMD's consumer products. Weapon's grade chips generally utilize even older, but specialized nodes (for environmental, EM hardening reasons for example)
Corsair4@reddit
Cutting edge is increasingly concerned with machine learning and related computations.
Nvidia owns that space, through a mixture of better hardware, and several decades of setting up cuda as THE environment for ML work.
If you want to be on cutting edge for ML stuff, CPU doesn't really matter, so AMD being better than Intel doesn't really matter. AMD is way behind Nvidia on that front, and Nvidia uses TSMC or Samsung for that work - neither of whom will bring their cutting edge fabs to the US.
hypersonicboom@reddit
TSMC has more fabs coming and Nvidia could abso-f.cking-lutely still design and manufacture leading edge products using those fabs, if the ones in Taiwan or Korea became inaccessible for them (and the rest of the world) due to blockade or outright war.
Corsair4@reddit
Ok, lets look at what they're planning, shall we?
Straight from their website.
High volume production on N4 started in Q4 2024.
Volume production of N4 started in Summer 2022, so Arizona is 2 years behind.
They are planning volume production of N3 in 2028. Volume production of N3 was in winter of 2022, so the Arizona Fab, according to their own timelines, is 5+ years behind on that one.
They just broke ground on the fab intended for N2 or 16A production. N2 is expected to enter volume production this year, and 16A is expected to enter volume production in 2026 or 2027 in the main Taiwan facilities. Arizona is targeting "end of the decade" for volume production there, by which point the Taiwan division will be looking to the next development.
TSMC is very clearly not intending for these US fabs to be on the cutting edge. The US foundries are literally several years behind the cutting edge.
High value industries like this are a national asset to the relevant government. They encourage countries to play nice with each other. Taiwan and Korea have manufacturing and IP, US has IP, ASML has manufacturing equipment, China has rare earth mining and processing.
Any country that makes an effort to sequester the whole process to themselves is going to get shit on in 5 different ways from other countries who have dominant positions in other parts of the process, or other industries.
hypersonicboom@reddit
That's why I said that you can absolutely manufacture leading edge designs on n-1 node, just look at RTX 3000 series vs. the Radeons of that time. Although the Samsung node Nvidia was using was actually n-2 vs TSMC
Recktion@reddit
Even all of Intel's recent consumer products use TSMC.
Dan6erbond2@reddit
Consumer. However it's about an American company fully being capable to design and fabricate certain components especially in military applications.
snollygoster1@reddit
TSMC is (or was) in the process of building a foundry in Arizona. While it is a foreign based company that foundry will provide Americans with jobs. Intel can't just get a free pass "because murican".
Recktion@reddit
AMD, Apple, Qualcomm, & Nvidia don't manufacture chips, all those companies are 100% dependent on TSMC and would not be able to produce any products for year(s) if they couldn't use TSMC anymore.
Global Foundries & Texas Instruments are at least a decade behind TSMC, they do not manufacture high end chips now and will never manufacture them again. They have a fraction of the R&D capabilities and will eternally be behind China & SMIC now.
Intel is the only American company able to put out what TSMC did 10 years ago. Instead of 10 billion, the US would need to pump hundreds of billions into a company to even have a chance of competing with TSMC.
gumol@reddit
you might be forgetting about Samsung foundries
Recktion@reddit
Only used for memory, no one wants to use them high-end chips.
Corsair4@reddit
Nvidia bounces between TSMC and Samsung processes for their GPUs, and are currently planning on using Samsung's 2 nm process for at least some of their GPUs.
Recktion@reddit
They don't use Samsung for anything they currently make and that's a click bait title. Nowhere does it say Nvidia is going to use them but in the title of the article. It's just rumors, of them making low (for them) margin chips.
Corsair4@reddit
The point is that it's not like TSMC is the only game ever for high end chip manufacturing. Samsung is in the ball park.
And besides, Nvidia absolutely gets their memory from Samsung on their high end products, and that is absolutely not produced in the US.
So if your argument is "Nvidia couldn't produce products at all if TSMC went under," thats just not true. They've worked with Samsung for chips in the past, and Samsung remains an option.
And if the argument is "what companies are relevant for the whole functional GPU", then Samsung is absolutely involved in that conversation, because they along with SK Hynix are probably the most advanced RAM manufacturers around, and VRAM is critically important for cutting edge compute on ML techniques.
Sythe64@reddit
If only those companies were encouraged to invest hundreds of billions dollars into the USA instead of China.
Corsair4@reddit
TSMC has invested 165 billion into the Arizona Fab.
The Arizona Fab's timeline is several years and several nodes behind TSMC's own flagship production in Taiwan.
Samsung has also poured a ton of money into US fabs, and they too keep the cutting edge stuff in Korea.
Cutting edge semiconductor processes are a national asset. You can get TSMC and Samsung to do things in the US, but they will NOT bring over the latest and greatest to the US.
Koomskap@reddit
Sure, then we'll lose our advantage in the applications of chips too.
I agree with you on the idea, but the cat is out of the bag now.
Noobasdfjkl@reddit
Good job mentioning a bunch of fabless companies that all rely on Samsung and TSMC.
Saying “Intel is not ‘leading edge’” while having just mentioned GloFo and TI is hilarious.
taylorswiftttttt@reddit
I said "manufacture leading edge semiconductors". Manufacturing leading edge semiconductors is a uniquely difficult task. There are very few other industries that compare in the insane amount of investment needed to become a competitive manufacturer. Manufacturing widebody airliners is comparable in some ways: only two companies can do do it, and China is investing billions of dollars and decades of time to catch up.
There are only three companies that competitively manufacture the most advanced semiconductors - Samsung, TSMC, and Intel. TSMC and Samsung are building fabs in the US, which is certainly good, but the R&D for those countries will always stay at home.
JSTFLK@reddit
Intel fabrication is quite competitive. Their reliance on x86 architecture is comparable to Kodak and the market for film cameras.
piddydb@reddit
Not to be a pessimist but it seems like that’s going to happen with or without taxpayer money
Agree-With-Above@reddit
I see this argument all the time, especially about the automotive Big 3 back in 2009.
People that argue against the bailout are just only looking at things superficially. If the government didn't bail them out, hundreds of thousands of people at the OEMs will be out of jobs. Then tier1 suppliers will lose significant business, and they will have to lose people too. The ripple effect downstream will mean millions of people will lose their jobs and it will take a decade to recover.
Not to mention the loss of IP, US market share, etc.
guy-anderson@reddit
Eh, they only get the money until after construction of their new facility is done. As far as I can tell, Intel has yet to get a single tax dollar, and if their construction projects fall through it doesn't cost US Taxpayers a dime.
Rude_Thought_9988@reddit
I don’t think they have received a single cent from that deal yet.
V8-Turbo-Hybrid@reddit
This company also gets better CEO as well.
StatusCount7032@reddit
The company said it will fulfill existing commitments to customers but will lay off “most” employees working in Intel’s automotive group.
Ok. Why would the customer stay with Intel if they're laying off the people that develop/support the products and/or services?
LegoGuy23@reddit (OP)
I think that statement means that they'll fulfill existing orders and/or support contracts. Certainly no one will be out to place new orders, but existing customers won't be left high and dry; there'll be a skeleton crew left for a while.
LegoGuy23@reddit (OP)
It isn't too surprising given Intel's ongoing financial troubles and the continued rise of ARM, but it's still a bit of a shame.
holzmann_dc@reddit
Recalling the 1990s and up to about 2007, I never thought they would fall as hard or as fast as they have. I mean, we use to refer to the world in WinTel terms. How the fuck did they get it all so wrong?
ob_knoxious@reddit
The failure to move off of 14mm and abandoning the "tick tock" model is the main reason. They had a lot of failures in the 2010s but this is the real failure. TSMC beat them at their own game.
nucleartime@reddit
Also not doing more than quad cores for consumers for the longest time.
handymanshandle@reddit
Man… Intel had a LOT of trouble in releasing a viable 10nm fab. It really wasn’t until Golden Cove where they finally had their lineup sorted out over having a bunch of chips that were either 10nm or 14nm+++++. It all happened rather quickly, too; AMD found a nice spot to counter Intel in the enterprise sector and ran away with it seeing how quickly Epyc chips exploded once they moved to Zen 2 and TSMC 7nm.
holzmann_dc@reddit
Tick Tock is what you do when you think you're top dawg and always will be. In other words, resting on laurels.
ob_knoxious@reddit
I mean its kinda the other way around. Tick Tock was what Intel used since 07. Die shrink (Tick), micro architecture (dock), repeat. They still maintained 90% market share in desktop and enterprise with this. When 10mm failed Intel decided they could get away with what was basically Tick tock tock tock.... believing they could basically get away with not upgrading their lithography because they were so far ahead of TSMC and others. That is truly resting on your laurels. And with failures in mobile and security issues, they end up where they are now.
Sir_Sir_ExcuseMe_Sir@reddit
Adjustable rate mortgages?
Oh_ffs_seriously@reddit
Nope, mortal enemies of the CORE.
LegoGuy23@reddit (OP)
Advanced RISC Machines (originally Acorn RISC Machine) is a type of microarchitecture and assiciated instruction-set becoming increasingly popular compared with Intel's X86 system. They're much more power efficient and rapidly catching up in terms of raw CPU horsepower.
For example, your cell phone uses an ARM processor, and a couple of years ago Apple switched from Intel (x86-64) to their M1, M2, etc. line of CPUs. While Apple designed their own CPUs, they're ARM-compatible processors and they licensed the IP from the ARM company.
Basically, they used to only really be used for low-powered embedded devices, but now they're in everything from game consoles, to laptops, to cars, to data-centers.
mk1power@reddit
Advanced RISC Machines in this case lol
mustangfan12@reddit
Yeah, I never even knew they had a automotive division. They probably couldn't attract any customers, and throw in the fact that x86 isn't ideal for low power devices
WCland@reddit
About 10 years ago there was a ton of competition among chipmakers to get into the auto market. Intel, Qualcomm, Broadcom, etc. Nvidia pretty much owns it for high end chips now. As noted by OP, ARM has a good bit of the market. There are also many simple processors around cars for things like power mirror adjustment.
poopoomergency4@reddit
intel has an automotive business?
rideincircles@reddit
Mobileye.
ice0rb@reddit
Automotive compute not cars
hi_im_bored13@reddit
They also used to own mobileye
ACiD_80@reddit
not used to... Intel owns about 80% of mobileye.
Recoil42@reddit
They still own Mobileye.
SpaceghostLos@reddit
Didnt Ford purchase Mobileye?
hi_im_bored13@reddit
ford, bmw, gm, vw, etc. all have partnerships with them and purchase their product but mobileye is a public company
blacksantron@reddit
Everything's computer!!
tangledwire@reddit
Just like those Teslers...
dnyank1@reddit
Yeah, pretty much all the first gen "android cars" from Volvo, Polestar run intel. Can confirm it's just not a performant solution relative to the expectation set by modern phones.
And none of the gen 2 cars from GM or other brands went Intel, so...
fak3g0d@reddit
Do you know what chips they're using now?
dnyank1@reddit
Qualcomm automotive, Nvidia
fak3g0d@reddit
cool, thanks
markeydarkey2@reddit
The infotainment in my car uses an Intel Atom X5-A3930.
IWantToPlayGame@reddit
I'm sorry
p90rushb@reddit
If they made a car, they would probably name it Intel Acceleron.
IWantToPlayGame@reddit
Dude that's brilliant! I legit lol'd
driveawayfromall@reddit
They make really shitty automotive grade processors like the ones in early Teslas and polestars
Szymon_Sz@reddit
Not anymore.
sri_peeta@reddit
It's amazing how much Intel has fallen in the last decade or so. Car industry is in the midst of 3rd generation of its automotive-computing era, nearing the cusp of AI-Self driving features rolled out to the masses in the next few years and they still cannot make money out of all that's been happening. GE, Boeing, Intel...wonder what's next?
holzmann_dc@reddit
Actually, GE Aviation is doing pretty good. Intel, Boeing, etc. all bowed to the Jack Welch school. FAFO.
manystripes@reddit
Having worked in automotive electronics I'm trying to remember the last time I saw Intel silicon in an ECU and I'm drawing a blank. Maybe a Windows based head unit from like 15 years ago but those have all moved onto better platforms
LeifEriksonASDF@reddit
They were on old Teslas. It was pretty bad. They went to AMD Ryzen later on and it was a night and day difference. The Ryzen change was around 3 years ago and there's already features they can't port over to the old Intel ones anymore cause it'll lag too much.
AndroidUser37@reddit
Eh, I have a family friend with a 2018 Model 3 and the Intel Atom chip and it works fine enough. I don't notice many problems with it.
ducky21@reddit
Old Tesla Model S were famously the cars using consumer grade electronics and not automotive, so they just Didn't Work in Texas and had to have special A/C routines added to keep the infotainment LCD from melting.
Those same cars, right?
godlyhalo@reddit
Tesla's engineering standards for their PCB's are laughable in comparison to other OEM's. If the engineering standards say a component should only meet the bare minimum requirement of a class 3 electronic device (As defined by IPC, an international organization), then they tend to have some issues that arise with longevity, thermal cycling, and other environmental conditions that automotive PCB's experience. All other OEM's state that they must meet class 3 specific, and a bunch of their own requirements regarding pass / fail criteria after different types of testing. The legacy OEM's have refined their engineering standards to fit the uses of their vehicles. Tesla is basically, "Yup it looks good! Ship it", and then people wonder why they have major electrical issues 5 years later.
LeifEriksonASDF@reddit
Those might have been the even older Nvidia Tegra ones. The Intel ones were merely slow as hell but the Nvidia ones I think were guaranteed to fail given enough time.
godlyhalo@reddit
Never seen an Intel chip on any automotive ECU, control module, or other component in recent years. It's all ARM / Nvidia / ISSI / Texas Instruments and a couple of others. I do work specifically on these types of boards and components for a variety of OEM's, I didn't even know Intel had a presence at all. Intel may be too specialized in high performance CPU's for the automotive market. The auto industry is still using 160 NM transistors on some components, I actually got to see those once under an optical microscope recently.
IWantToPlayGame@reddit
What are headunits using now? Linux? Android?
manystripes@reddit
Unsure of the whole spread but Android, QNX, and linux distros like Yocto at minimum
funnyfarm299@reddit
There’s also Android Automotive.
potatohead_v2@reddit
It's an awful time to be an automotive industry worker. Companies across the globe and across the entire supply chain are having all sorts of issues and people are getting laid off. The automotive landscape will look very different 10 years from now
handymanshandle@reddit
I'm surprised that people don't remember that Intel did a lot of embedded hardware for automakers. Their Atom chips were all over a bunch of Linux-based infotainments for the longest time (and they seemingly still are to some degree?). Guess it shows how much publicity can matter when more people recognize AMD's rather small contributions to this sector much quicker than Intel's here.
That said, I can't blame Intel for pulling out here. They've been in quite the pursuit to make budget cuts lately and I guess they haven't done so well here lately, With the low-power part of the automotive embedded CPU cornered by what are presumably more cost-effective and better ARM options and the higher end part of that segment cornered mainly by Nvidia, Intel has no real place to go here.
r_golan_trevize@reddit
Ford used Intel chips way back in their long lived EEC-IV and EEC-V control units starting in the early 80s, replacing previous Toshiba and Motorola processors in earlier EEC versions dating back to the mid 1970s. Intels been around in the business for a while.
coconutpete52@reddit
Today I learned Intel has …. Had … an automotive business!
No_Stay_4583@reddit
Probably replacing them with AI, nice move of Intel
Bderken@reddit
Man the Reddit ai parrots… worse than ai itself lmao
2016KiaRio@reddit
Are you blind?
j_demur3@reddit
All I can really find about what Intel Automotive were up to prior to this news is just their marketing bumf about Software-defined Vehicles and how current cars have '100 ECU's' and how their upcoming platform will create an 'iPhone moment' by cutting that down to 'less than 50'.
Plus some vagueness about Zeekr, Stellantis specifically in regard to Formula E and Karma 'announcing support' but I don't know whether that relates to, didn't Karma show some concepts cars and announce further versions of the what was the Fisker Karma and then declare they were relaunching in a different direction or something? If that was Intel's whole roster at the end, there's no wonder this wasn't seen as a necessary part of the business.
THE_GR8_MIKE@reddit
It's a good thing touch screens and the like aren't unresponsive and laggy as shit in most cars.
Wait-
LeopardHalit@reddit
lol no more Intel Atoms