Targeted layoffs might be the best thing long-term.

From my limited perspective of having recently worked at Intel, they had a lot of dead weight. There was a lot of talent, but it was also diluted by a lot of management and bureaucracy.

I honestly don't know if layoffs are an effective solution to this, but IMHO they could definitely benefit from lower headcount and greater urgency.

Thanks for this. I see that perspective and from my experience in larger tech corps it jibes.

I'm still trying to reconcile it with the CHIPs thing. Intel on one hand is admittedly bloated and inefficient, and so is choosing to reduce its headcount and therefore bandwidth to do work.

So what is the CHIPs money going to. Typically when you invest in something like this you are buying people's time to do the work and also materials and supplies to build the stuff like factories, etc. So if we are lowering headcount it's not going to labor.

I suspect as you say it's more about changing worker types, like getting rid of skillsets they don't need so they can hire those they do like people to build semi plants here.

So, in my mind, the only way it adds up is if Intel goes on a big hiring spree soon. Otherwise, where did the money go? We can buy equipment to build plants but who builds them?

I don't have any insider info on this, but here's my guess:

Gelsinger thinks x86's days are numbered. So Intel's historical moat of designing and manufacturing x86 chips is going away.

So now it's more compelling to treat design and manufacturing as separate business concerns:

- For potential foundry customers, it reduces fears that Intel will give higher priority to fab'ing its own chips.

- It frees Intel's chip designers to design chips without needing to assume they'll be built at Intel's foundries. E.g., they can design chips for TSMC's process if that would result in a better product.

And given all these factors, I suspect Gelsinger and the Intel board of directors are planning to separate the design and manufacturing functions into separate businesses. (Either practically speaking by limiting their interaction within Intel, or literally by making them separate companies.)

This matches what I've heard about Intel from friends who have worked there. I had a firmware programmer friend who's job was literally to copy data from excel files to .h files and when he offered to automate the process was told that the manual transfer was the process he should follow. He described his role as dead weight.

A friend that also worked at Intel until a couple of years ago confirmed what you say.

well-connected one-percent chick goes to work for Intel Security long enough to get health care for babies then quits. At the same time, competitive University grad-with-honors working class girl does protests and environmentalism, can't have children with no health care, ends up hanging out with lots and lots of other thirty year olds in the same boat, with debt. Just another day in America.

But what did they major in? Did the well connected person major in CS while the honors grad major in underwater basket weaving? We are in a free market system that is not perfect at evaluating talent but things like degree help provide signals to employers.

here is a basket-weaver :

----- B--- (Hupa) received a BA from Stanford and wrote her 2017 Berkeley PhD dissertation Wailaki Grammar on a Dene language spoken along the Eel River in northern California. Now an Assistant Professor of Native American Studies at Humboldt State University, her research focuses on Dene languages, and on historical-comparative linguistics for language revitalization within the Wailaki and Hupa communities. While at Stanford and Berkeley she also worked with speakers of Karuk, Yucatec Maya, and Sereer. B---- is a coauthor of "Xo’ch Na:nahsde’tl’-te: Survivance, resilience and unbroken traditions in northwest California" (2019, with Cutcha Risling Baldy, in Ka'm-t'em: A Journey Toward Healing), and has contributed to the Hupa Online Dictionary and Texts project (at UC Davis). She is also a traditional basket weaver and singer from the Xontah Nikya:aw in Hoopa Valley, and a member of the board of the Advocates for Indigenous California Language Survival.

Yah great point, the US culturally seems to really struggle with the idea that the business is just an arbitrary conception, what is important is the workers. We really have a disturbing sense that arbitraries collections of capital and legalese are more important than humans with passions and skills and it honestly scares me.

Yeah, it's amazing the number of people who believe that corporations exist in the same objective sense that puppies do, rather than the social-consensus sense that D&D characters do.

Imo the closest analog to a corporation is a program. It objectively exists like puppies do, but only when it's executed by the hardware, which are the people. Otherwise it's just a stack of documents without any real power to affect the reality.

You can easily guess what the purpose of the program is.

Mathematical constructs like programs and theorems are a third category; they have structure and properties that don't depend purely on social construction the way a company's bylaws do. Of course which sets of axioms we prefer to reason from is socially negotiated, but once you've decided on a set of axioms, you don't have any freedom of choice about the consequences. This gives propositions about them a kind of objectivity that corporations or D&D characters lack. It can never be objectively true or false that a given corporation is, for example, bankrupt, but it can certainly be objectively true or false that a theorem is correct, or that a program performs an out-of-bounds array access when executed on the input "-0".

1500 years ago, Aryabhata came up with an algorithm for approximating an integer ratio with (what we today call) continued fractions, and we use it today for that and for inverting elements of finite fields. Aryabhata was a Hindu, speaking Sanskrit, living in a monarchy; no corporation he formed part of still exists, or could exist, at least according to accepted legal principles. Yet his algorithm continues to be correct today, and like the movement of the Earth, it would continue to be correct even if nobody believed it.

Platonists believe that mathematical constructs like programs are actually more objectively real than things like stones or puppies, because propositions about them have a truth-value that is independent not only of the speaker's social context but also of time. Today this puppy is alive; tomorrow she will be dead. The Pythagorean Theorem doesn't do that.

The knowledge and talent is definitely very important (I mean in the long term everything in technology is knowledge and talent) but there are also the extremely expensive semiconductor fabs to be considered.

The CHIPs bill was pretty broad. It might make sense to subsidize a pure-foundry company as an ongoing issue (in particular, isolate these big investments from the boom-bust semiconductor trends). Or somehow try to come up with subsidies that go to Intel to the extent to which they act as a pure foundry, but that will be pretty tricky to work out I guess.

An ecosystem of open fabs seems to be a prerequisite of those small plucky startup chip design teams. Of course they can order from TSMC but then they have to wait however many weeks to test out each prototype...

> Then they announce huge layoffs to get rid of a lot of thr talent, people, heart of their business.

Intel has not announced layoffs. There is a rumor layoffs will be announced at the next quarterly earnings release.

>short term should be number of people working in semi in the US

it's not a jobs program and majority of the layoffs are targeting non-technical areas. Raw number of engineers isn't a great metric for government programs either, could be easily gamed. Focus should be on building up the ground level infrastructure needed so actual innovation can happen, not just giving money to Intel and other established players to further strengthen their monopoly

So far the rumor is that they're laying off marketing folks, not engineers. And we paid them to expand their microchip infrastructure, not create jobs.

Thanks for this. I'm curious to dig a bit deeper. What is involved in expanding their microchip infrastructure? What would funds be spent on? It seems like hard tooling, buildings, fab lines, and then they need people to build install and run all that stuff. I guess I'm asking - am I missing something, like can they expand their microchip infrastructure without the labor of people? In my experience designing automation for fabs there is a ton of skilled labor that goes into every single fab, both to design and set it up but also to run it.

Every business I have started or contemplated starting has always relied 100% on the people needed to run it. Otherwise, nothing happens.

>How did we dedicate money and resources to this thing of national importance and end up with fewer experts working on it?

How do you even know whos going to be fired?

What makes you think the government could run them better...?

You could run them exactly how they are structured today if you want. Youd reap savings by not having shareholder profits to maintain and could use that money that would otherwise be wasted on luxury spending for major shareholders for more R and D.

Without shareholders to answer to costs would increase out of control and productivity would plummet. It would become like a government

Sounds just like what China has been trying to do in their semiconductor industry for years without success

No thanks, I’ve been to the DMV

The DMV isn't a chip factory. It is an administrative office performing unskilled work. Most of which has been automated away.

Why wouldn't the government just start up it's own Intel-alike?

Because intel would lobby against it and snuff it

Couldn't any hypothetical governmIntel competitors also just lobby to run it into the ground? (or, perhaps one of the political parties will just want to run it into the ground in an ridiculous attempt to prove that the government can't work).

You mean like they did in communist countries and had a huge success at killing their economies?

I was thinking more along the lines of when the us government centrally planned the economy and won a world war as a result.

Committing to purchase X quantity of Y from a factory that already exists in a competitive market to make Z is different from removing the need for the company to make ends meet