Hacker News Reader

Comment by ksec | Hacker News Reader

Intel said the same thing for Arizona as well. Initial $20B investment which could expand to $100B. So they are building it in two places. I wonder why they dont keep it all in Arizona? ( May be Tax Break, but I have no idea how Tax Break works in US )

And if you include their (expected) expansion in EUR, Intel is layering the foundation for Foundry Services. Compared to its half assed Custom Foundry in 2012/2013 with little to zero CapEx increase. Foundry Services 2.0 is very real.

Note: Investor Notes from ASML are also interesting, suggesting Intel may be ahead in terms of High-NA EUV orders. Initial shipment expected in 2023.

View on HN

Increasing supply chain resiliency by having multiple, geographically separated, sites this more of a thing now.

https://hbr.org/2020/09/global-supply-chains-in-a-post-pande...

"The obvious way to address heavy dependence on one medium-risk or high-risk source (a single factory, supplier, or region) is to add more sources in locations not vulnerable to the same risks."

also, resiliency against adverse action in the Senate. they now have 4 senators who are unlikely to vote on anything damaging to Intel

That's similar to the method defense contractors use, they spread out their manufacturing locations to have influence over as many politicians as possible.

I'm a space nerd and the geographic distribution of the Apollo contractors was downright impressive. They built major systems from New York to California.

This is an interesting point I never thought of before. Now I wonder if the tech giants would have faired better politically if they weren't all co-located in one of the most politically polarizing areas of the country.

Honestly? They should be spread out. There is no reason for FAANGS to be located in mega campuses in the Bay Area. Let people who eg graduate from Midwestern colleges have the option of working in the Midwest (or returning there), pay them equally well. This would create so much economic activity in areas that really need it instead of simply enriching the landlords in the Bay.

It is a part of the reason why all the big government contractors are spread out all over the place. Even around a single metro area they'll have different offices in different congressional districts.

If you are on the winning end of that bloat life is pure happiness and joy. But if you are on the losing end (ie. not employed/a beneficiary of these firms) then you are paying dearly for the joy that others have.

I would think a lot of people benefit from the motivation for companies to geographically diversify. A lot of the bad things about the Bay Area boil down to people willing to put up with a whole lot because of a group delusion that a certain kind of thing can only happen there.

>I would think a lot of people benefit from the motivation for companies to geographically diversify.

Yes like I said there is the "in group". Is that "a lot" of people, I wouldn't say so. Namely the people who are employed or benefit financially from these orgs. If you are paying for the bloat via the government which is the customer of these defense orgs, it is effectively a tax on everyone else.

It’s not a delusion. While it’s true that you can start a business anywhere, being in the vicinity of people who have a ton of experience is very useful. There’s a reason why people move to LA to make it in entertainment or to NYC to make it in finance.

That said, for a non startup technology job, yes, no reason to be in the Bay. That work can be done from anywhere.

6. Intel is the largest private employer in Oregon as well.

Right. Interesting take on that below re: Taiwan/TSMC, etc:

https://www.supplychainbrain.com/articles/32482-the-world-is...

> why not all AZ?

Game theory between US states playing to get the most gov’t assistance?

Finite pool of talent? Our plants are in rural areas and everyone is struggling to hire, from auto to bakeries.

The Great Lakes region has a surprising concentration of advanced manufacturing skillsets. With large manufacturing facilities with extremely stringent quality standards like a bleeding edge chip fab there's going to be a enormous need for people with traditional manufacturing-specific support skillsets that are simply not available in large quantities anywhere else in the US.

If the current administration's support of the transition to electric vehicles doesn't falter, that's going to drive an increasing need for microprocessors in vehicles. Intel locating themselves in central Ohio puts their production facilities on an interstate highway hub that puts them within a 6-12 hour drive of nearly every major auto manufacturing facility in the US, which is an enormous incentive for the auto manufacturers to source from them. Additionally, Columbus has the Rickenbacker International Airport which is a dedicated air freight airport that lets them get their product global faster.

This announcement of two facilities makes me think that Phoenix is going to be a smaller facility handling the more advanced, smaller run chip designs, but host the R&D and design offices (To take advantage of the city's concentration of tech talent) while Columbus is going to be the primary production facility.

I was surprised to learn Minnesota has a fab that's critical to the US govt! They are one of the only secure trusted chip makers.

Growing up there I had no idea. My dad loved to talk up MMM, Honeywell, Monsanto and a bunch of other legacy companies in MN that still play huge roles that I have overlooked.

Just one article at top of search I found even talks about the public-private partnerships (tax & other incentives)

https://www.aroundosceola.com/news/skywater-seeks-dod-approv...

They’re going to fab there and then ship the output to Malaysia, Costa Rica, or the Philippines for package and test. Being close to Detroit doesn’t count for much.

the only thing an ev actually needs chips for is charge controllers which can be old fab tech, and motor controllers which aren't normal silicon or tiny feature size. Tesla is using silicon carbide, and gallium arsenide will probably be next.

The Bollinger takes this route of minimalism. Ironically, an EV has less reliance on processing than a modern ICE vehicle.

Automotive EE here…

> the only thing an ev actually needs chips for is charge controllers which can be old fab tech, and motor controllers

I wish I had the time to explain how this could not be more wrong.

Quickly, I’ll mention that the vehicle I’m working on has 8 separate modules for the driver’s seat. Inside these modules are a combined total of 10 processors with as many bootloaders and flash procedures and validation records and crypto and peripheral drivers, 17 CAN-FD modules, 6 system basis chips, 9 switching power supplies with a number of accessory LDO supplies, at least 40 more “chips” of various functions and capacities… for just the driver’s seat alone

The steering wheel buttons for radio left and right side and the cruise control comprise 3 modules and power supplies and etc etc.

>The Bollinger takes this route of minimalism. Ironically, an EV has less reliance on processing than a modern ICE vehicle.

I know what you are trying to say, but it is largely incorrect. The EVs still have “engine” control modules and powertrain systems. Instead of figuring out injector pulse widths, they’re calculating deliverable torque. The ABS/brake controllers are more complex, the battery heating and cooling systems require modules and processing, the HVAC, the transmission, steering systems, everything is more complex compared to ICE with an old ample supply of hydraulics and coolants and a simple fuel that is stable and usable at all storage temps fed into a mechanical device.

I’m not trying to be confrontational, but you couldn’t be more wrong about automotive electronics.

The previous poster is definitely massively oversimplifying things, and I do value your input as someone who actually works in the field. Obviously, there's still a lot of the same kind of controllers when it comes to all the switches and displays and motorized chairs and turn signal logic and all of those things which are practically the exact same as an ICE. And obviously there's some places where you don't need controllers and sensors on an EV where you would have them on an ICE, like mass airflow sensors and fuel injection controls and stuff like that. Meanwhile you do have additional circuitry like the inverters to actually drive the motors, BMS systems, there's still sensors on the motors and wheels and what not, brake regen, etc.

You mentioned the transmission is more complex compared to an ICE, but that does not make much sense to me at all. To my knowledge most EVs are essentially fixed gear ratio gearboxes, so you're either in drive, reverse, or neutral. Meanwhile ICE transmissions are becoming more and more like computer-controlled manual transmissions what with the rise in popularity of dual-clutch automatic transmissions. To me it seems a ton more complicated keeping track of operating that DCT than just "the dial is in D, change transmission to the forward mode."

I think it's gallium nitride (GaN), not gallium arsenide (GaAs).

you are absolutely right. I couldn't figure out why GAN didn't make sense so I spelled it out and got the wrong one. It is amazing how much smaller and cooler GaN made usbc power supplies and rf power amps. I'm waiting on lunchbox welders based on it. It is so insanely efficient.

Also the infotainment. And real questions because I'm genuinely curious: do driver assists (ABS, ASR, TCS) and their adjustments rely on processors? What about in-car adjustments to suspension height, ride firmness, throttle response (e.g. eco vs sport vs comfort modes)? Lane keep assist, blind spot checking, adaptive cruise control all seem like things that would require processing power as well.

> do driver assists (ABS, ASR, TCS) and their adjustments rely on processors?

If I'm not mistaken, microprocessors aren't the only semiconductor components that have been in short supply recently. Things like shaft position sensors are also solid-state devices.

Most current vehicles use separate embedded microcontrollers for those features. In principle it's possible to build a car with a single powerful processor controlling everything, but the engineering becomes far more challenging and you need a higher level of vertical integration on the parts supply chain. There are also safety risks: you don't want an LKA code defect to cause a TCS failure. And wiring harness would have to grow to carry all the separate data signals.

Tesla is partially going the integration route but they still have many separate processors in the car.

Even headlights and brake pedals (brake by wire) have ICs in them these days, I doubt getting rid of a PCM will make an EV less IC heavy.

The only infotainment a car needs is interfacing with CarPlay/Android Auto.

Do EVs not have all of the ancillaries hanging off CANBUS or something? Or is the lion's share of chips in an ICE vehicle engine and transmission control?

Of course.

There isn’t an EV in the world that is less complex than it’s ICE counterpart. GP is very mistaken. I wrote a longer explanation above.

Most of the complexity of EVs is just generic modern car complexity, though. EVs don't have to be any more complicated than ICE vehicles, and EV conversions especially are usually really simple (though to be fair a lot of the vehicles being converted are old classic cars that didn't have any computers in them to begin with).

The EV drivetrain might have more or less computers than an ICE drivetrain, but really managing batteries and a motor isn't inherently all that complicated. On the ICE side of things you have a lot more moving parts and the necessity of keeping everything properly tuned so it can pass emissions checks.

Braking and traction control I could see being fairly complicated, but it's a similar problem whether it's an ICE engine or an electric motor.

Self-driving is hugely complicated, but that's not complexity that's inherent to EVs specifically, it's just a feature that's usually only implemented on EVs for some reason.

All the other stuff: infotainment, tire pressure sensors, back up cameras, and so on are just modern car complexity.

The braking is way way more complex because of regenerative. That isn't a factor with ICE.

Trust me - it's not simpler. I know you think we just apply power to the motor and go, but that is not the case. There are facets of "soft start" and it ties in to traction control, there are efficiency processes, etc. Instead of map tables for rpm and injection, you have battery heating and cooling targets with current sensed feedbacks and torque targets, and bla bla bla. Stop making me bring work home with me! ;)

EVs are not simpler. It's literally impossible in automotive to ever get simpler. At least since the 1970s. I think we're hitting a point of unsustainability and the people in charge think OTA firmware is going to fix that, I think it'll make it much worse.

EDIT: fwiw, I wish I could go back to a time when backup camera was even mention worthy of "things I consider complex in a vehicle". It's just a couple shielded wires and plugs into the radio/telamatics. I can tell you without a doubt that the starter in most vehicle is now far more complicated, to the point it too has it's own microcontroller and data bus (typically LIN, not sure I've seen a CAN yet).

I agree it isn't simpler. I only assert that it could be if you ignored all those efficiency gains from regenerative braking and battery temp management. Things that aren't often in ev conversions. But you made me realize that is apples and oranges, because an ICE that gave up on efficiency could be an old vw beatle with aie cooling and a carburetor. I feel you on the seemingly endless stairway of complexity. I work in embedded safety critical and I often tell people to never do anything in software they could have done in hardware. Hardware is cheaper bat this point...

Most conversions have regenerative braking nowdays. Most AC motor controllers support regen as a basic feature. A lot of the old series-wound DC motor conversions didn't have it because it was hard to do on that kind of motor (similar how it's hard to reverse; you have to flip the polarity of one set of leads but not the other, which requires a bunch of contactors or IGBTs or something).

Where regen gets kind of complicated is that you generally need to turn regen off when someone slams on the brakes. Unless the traction control / antilock braking system knows how regen figures into its calculations and can apply the proper braking force to each wheel taking it into account. I don't know what OEM EVs typically do in that situation, but for conversions you usually just have a brake switch so that regen can be automatically disabled when you're actually braking.

What does the software industry look like for this stuff? I've got two decades of embedded c++ experience, some of which is safety critical. How much of automotive is offshored these days? Is it still mostly C and a bsp and a text editor, or is it gitlab and pipelines now?

I don't know, I'm just familiar with what the motor controllers do because I'm trying to do an EV conversion.

I would imagine that the people working on parts for the conversion market have a somewhat different development process than a regular auto manufacturer, in part because they're making things in low volume and in part because they just have to design generic reusable parts that do one thing properly. If you're building a whole car, you have to make sure everything works together as a cohesive whole. In some ways that would be easier because you know exactly what parts are being used together.

I would hope that most of the safety-critical real-time software that controls a car's basic functions like braking and traction control and so on would be relatively simple and straightforward, but maybe that's too much to expect. My understanding of the way the auto industry works these days is that the major manufacturers are increasing dependent on a complicated network of 3rd-party parts suppliers, and I'd expect that to influence the character of the software stack somewhat.

Ohio is situated on the great lakes and therefore has a lot of cheaply available water, while Arizona does not. Chip production is water intensive. Not to mention the vast expanses of cheap land, and their electricity costs are pretty low as well.

These could be reasons.

Modern Fabs can recycle its water usage by up to 90%. Electricity cost for Business in both state seems to be about the same. Land are cheap as well ( From what I have been told ).

You do get operational efficiency from having Fabs all located within one place. And that is a quote from TSMC's ex-CEO Morris Chang. That is why I wonder if there is something missing. Or is it the simplest explanation Not to put all eggs in one basket.

> Or is it the simplest explanation Not to put all eggs in one basket.

I think this is important part. If something happens to one site (problem with electricity, etc) other will work.

Also, it maybe easier to find workforce that way.

Also it will be good for development for that regions.

Arguments can be made that TSMC's choice to consolidate locations is part of the reason that we have a semi shortage right now (concentrated water shortage risk, concentrated worker illness risk, etc. etc.)

One guess: if the reliability of Intel is considered a national security issue, they may get funding specifically to be more reliable to offset these costs.

They're building the plant near Columbus, which isn't near the Great Lakes. If they wanted Great Lakes water they would've chosen Cleveland, which is very close to Lake Erie.

Are you from California? It rains a ton in that part of the country, water is not scarce. It’s hard for Californians to wrap their heads around but water is not universally rare.

> Cleveland, which is very close to Lake Erie.

It's on the shore of Lake Erie. If it were any closer to Lake Erie it would be underwater!

Lake Erie is not so far from Columbus: When I was a prof at Ohio State, my wife and I would go to Lake Erie just for a pleasant Saturday afternoon. So we'd go to Cleveland, ..., Sandusky, etc.

Semantically, damning water in the Great Lakes Watershed (how Columbus get's its water) is not THAT different from letting that water flow into Lake Erie and then piping it back.

Point being, Ohio has very few long-term drought concerns.

Columbus is not in the Lake Erie watershed. It is in the Ohio River watershed. The Great Lakes Compact does not allow Columbus to get its water from Lake Erie.

> Semantically, damning water in the Great Lakes Watershed (how Columbus get's its water)

Wrong. Columbus gets most of its water from the Scioto River, part of the Ohio/Mississippi River watershed.

https://en.wikipedia.org/wiki/Scioto_River

You might be thinking about Cleveland instead, but I don't think anyone dares drink from the Cuyahoga River, since its had a history of catching fire.

https://en.wikipedia.org/wiki/Cuyahoga_River

Akron’s reservoir is on the Cuyahoga River just North of Kent. Though this is far upstream from “catching fire” location.

https://www.akronohio.gov/cms/Water/Watershed_Cuyahoga/index...

Is this the same Kent, Ohio that Nestle uses to bottles their water with? I may be mistaken here.

The Cuyahoga is pretty darn clean these days.

There are actually extremely specific and intense rules about which parts of the country can take water from the great lakes and what burdens they incur when they do.

It's not necessarily aligned with any specific jurisdictions either, I believe it's largely based on the boundaries of the watersheds. Or at least where the boundary was ascertained to be at some point.

So yes it's very different, regardless of the semantics(?).

The idea that the Great Lakes are going to be tapped to solve water problems outside the watershed is fall flat when they realize those Great Lakes states aren't going to sell off access like that.

I remember some moonshot proposal to pipe the water to the Southwest, which is wild.

There might be redirecting (or excessive drawing) of tributaries though.

It's surprising, actually, how small the Great Lakes watershed is, compared to the surface area of the lakes. You really aren't going to get much of anything from tributaries; if you want some of the water you draw it from the lakes.

I thought that while chip production uses a lot of water, the water is mostly not consumed in the process and due to environmental regulation the water has to be cleaned at the plant so they form a closed loop system. It takes a lot of water when the plant is built, but after that it requires only a small amount of water to continue running.

It's political distribution, to be called upon as needed. Same reason NASA is sprawled the way it is.

You can more easily get numerous Senators to go to bat for you and your various causes (Arizona is a growth state, Ohio still has a lot of clout (not as much as CA, TX, FL obviously)).

Water in AZ is some of the cheapest in the country. At least until Lake Powell runs dry. Not a bad idea to spread the risks around however.

At least until Lake Powell runs dry

Which is coming a lot quicker than anyone expected. It's why all of the cities and water authorities hooked up to Powell and the other Colorado River reservoirs are in panic mode these days.

Just a few weeks ago, the various states came to an agreement about cutting water use. Lake Mead is at historic lows. There are mandatory federal restrictions either in effect, or about to start.

No wonder they grow all that lettuce in Yuma.

Seems extremely stupid from a long term perspective, but what do I know. I'm just a software guy.

They grow lettuce in Yuma during the winter because it's warm enough to grow it there during the winter. Something around 90% of the USA's winter grown leafy greens produce comes from Yuma.

Why not florida?

When you look at Lake Mead and Lake Powell lately it seems they are well on the path to running dry.

Yeah, but they are building in Columbus.

All this sounds irrelevant when an ASML machine costs $100 million a pop.

> ASML machine costs

It seems that cost maybe around $300 million each: https://whbl.com/2022/01/19/intel-orders-asml-machine-still-...

Those are the new high-NA EUV machines. The current gen ones are more in the range of 150 million.

I agree about the game theory/race to the bottom as an incentive. I think there are more great reasons compounding these decisions that others point out too.

Large corporations love to hold cities and states hostage and make them bid against each other. AMZN is a great example with their campus 2 bidding war. I am glad NY had the cajones to stand up to them.

Most cities and states even have dedicated EcoDevo staff to actively seek out these companies.

Cities and states compete to give the largest tax benefits, usually something like you don't have to pay ___ taxes for __ years (like 10, worth tens of millions). Plus they'll spend on infrastructure and give preferential zoning, city regs, etc.

Sure having more jobs probably creates more incoming/housing taxes for the city.

But states like Texas win this race to the bottom (TSLA and Musk personally) and in my opinion the costs of these corporate incentives (imho handouts) are just passed along to citizens in the cost of less societal services and needs, less school funding, less everything. Just a funnel up.

Musk may be an outlier. He is part of a collective of libertarian techbros who have taken what they can out of California and are now looking for a new set of suckers. Texas has a history of playing second fiddle to California so they were an obvious choice.

Yes other tech companies are also moving to Austin and I think that is more a sign of some millennials and other young professionals opting to leave California for a lower cost reasonably sane area. Its still does not seem like massive California sized numbers though. We just have to wait and see. I'm still long California though.

Sucking the Colorado river dry probably has a lot to do with it.

Taiwan was (is still?) in a drought and semiconductor manufacturing requires so much fresh water that TSMC was bringing in huge tanks on ships from the mainland.

>Our plants are in rural areas and everyone is struggling to hire, from auto to bakeries.

This is a solved problem isn't it? Just pay people more and treat them with dignity and respect. The pandemic has given a lot of people the opportunity to realize that they do not have to settle for poorly paid jobs that treat people poorly. You should have realized and adjusted by now. If you cannot accommodate to these requirements, then you have no place in this market anymore. Just close up shop and move on.

That’s true if you’re talking about, like, Burger King or something. But even places that treat their workers well are even having problems. Not everything is black and white.

Can you give some examples? From what I see, most of the lack of supply is at the unskilled level.

Where I live—the Seattle area—it appears most or many places are having staffing issues. From basic grocery stores to higher-end New Seasons/Metropolitan Market type stores, for example. Same with restaurants. Heck, it’s even difficult to hire software engineers right now.

The Intel plant in AZ plant is in Chandler, which is more suburban than rural - it can draw on much of the Phoenix Metro population (4.5million) for hiring.

National security considerations encourage geographic dispersal of strategic assets.

Here is a much better article from Anandtech

https://www.anandtech.com/show/17221/intel-announces-ohio-fa...

And a Talk by Jim Keller on Future of Semi Conductor for those interested.

https://www.youtube.com/watch?v=32CRYenTcdw

I wonder why they dont keep it all in Arizona?

Maybe water?

I've read a bunch of newspaper articles from that region recently about how people in Arizona and New Mexico are getting very pissy about water going to new industries. And from what I read on HN, chip manufacturing needs lots of water.

One thing Ohio seems to have plenty of is water. At least compared to places like Arizona.

Reply on news.ycombinator.com