Hacker News Reader

Comment by TheCondor | Hacker News Reader

It’s too late. It was too late 18 months ago. Nvidia is over 30x bigger. Qualcomm is 2x, AMD is 2x. Intel needs a world beater, like tomorrow. Qualcomm offered to buy them, no idea what the terms are but he’d have to beat those terms, right? That had to be the death blow.

AMD has already shipped ARM parts in the recentish time frame (Opteron A1100) there are rumors that they have internal arm projects that can be ready quickly. All the other part makers are all in with ARM. Intel is on an island. Throw the customization that MS and AWS are doing, unless they have something magical, tomorrow, it’s just hard to see.

View on HN

> It’s too late. It was too late 18 months ago. Nvidia is over 30x bigger. Qualcomm is 2x, AMD is 2x.

AMD came back from odds much longer than this. At its nadir, AMD's market cap was about a billion. A billion! Jensen's jacket collection is worth more than that.

Anyway, it's not looking good for Intel but it's certainly not "too late".

If they can dominate with process, no it’s not too late. If their process is on par with TSMC? I don’t know. Like I said, they need a world beater else margins erode dramatically. They have assets, they have value and Intel has done it again and again for decades but unless there is something that is a tier better than anything else margins drop and they just forced their “technical CEO” out.

They have had ultra success in the past, I don’t see anything on the roadmap to suggest a repeat of that, there is a lot of institutional memory of that success. With a long term plan, a board that is knows it’s going to be a journey and the CHIPS money, there is a story, make X86S a reality… I think there will be incredible pressure to cut and sell it though. If the new fabrication technology is only as good as the competition, I think that makes Intel an even more attractive take over target for the fabless giants

> nadir

Great word, TIL.

> All the other part makers are all in with ARM. Intel is on an island.

Intel also has an ARM architecture license… they sold Marvell the Xscale assets, but retained rights to produce ARM chips should they desire to do so in the future.

https://en.m.wikipedia.org/wiki/XScale

Sure, but then they're competing with everyone else in a fair fight. Which means either taking a huge hit on margins, or having a product with enough of a performance delta to justify prices somewhere near today's.

The problem isn't so much x64 vs ARM but that, more broadly, high-performance CPUs are becoming commoditised.

Which is why Pat was on the right track imho with foundry. It’s too bad he won’t have the tenure needed to see the outcome of 18a, good or bad.

Even Arm is tenuous at best with risc-v right on its heels.

definitely not too late. plenty of cases in tech where worse comebacks were achieved. amd is a good example mentioned elsewhere in the thread. jobs brought apple back after it was months away from bankruptcy. even Nvidia itself was close to being fucked when its cuda bet was early and wasn't paying off yet -- Jensen talks about this in interviews

> It’s too late.

Hardly.

Intel has fabs and, as we found out from Covid, we don't have capacity in old nodes either. Intel can very much still print money.

Fabless companies are really cool--until shit hits the fan and nobody will run your chips anymore--but, hey, who cares, it's not like that ever happens, right? I mean you'd have to get an enemy invasion or terrorist bombing or tsunami or earthquake or fire in the fab line or global pandemic--and, pfft, how common are those, really?

All it takes is some high-level dumbass trying to curry favor with Pooh Bear over Taiwan and suddenly Nvidia, Qualcomm, Apple, etc. have a net value of zero for the forseeable future.

But, hey, it's super more profitable to not actually have all that icky manufacturing. Ewwwwwww.

> All it takes is some high-level dumbass trying to curry favor with Pooh Bear over Taiwan

Or just one large earthquake to disrupt production for a few months. I'm sure that TSMC has planed for earthquakes, but they are still highly centralized, where as Intels have their fabs spread out nicely (not all the same node size so if you really need 5nm you might still be in trouble).

Maybw centralization is the competetive edge.

Their best people can easily move from fab to fab.

> suddenly Nvidia, Qualcomm, Apple, etc. have a net value of zero for the forseeable future.

That is an exaggeration. There is still Samsung, Nvidia even fabbed there few years back. It's worse than TSMC, but it's not like Intel is the only alternative.

Re-engineering a product for a different fab often takes years. And, if TSMC goes dead, nobody, not Samsung or Intel or anybody, would be able to absorb all the volume required, anyway.

Yeah, you are right. Though I am not worried about latest and greatest. We can easily run everything on Sandy Bridge from 2011. It's the other stuff that is underappreciated.

Everything has chips in it. From thermometers through vacuum cleaners, dishwashers, fridges, doorbells. And that is just consumer stuff. The machines that make the goods are full of electronics too. All infrastructure, from water to red lights, medical equipment...

We got a very, very small taste of it for cars in last crisis. And nobody is building fabs for legacy chips... Except China.

Thats a great idea until China and NK launch a coordinated attack on both Taiwan and South Korea

We should definitely be prepared when China attacks USA. Supersonic missiles to the fabs, ASML, Tel facilities too.

Maybe we should make fabs on moon, just as an insurance.

That's not a joke BTW. If China attacks both SK and TW, that's a war with US and this will happen. Fabs in USA will be a very valid military target.

In a pacific escalation scenario, which I don't think is farfetched in the next decade, the CONUS would still be safe at least short of nuclear exchange.

Samsung is in TX too

Yup. Fabless is also totally dependant on churning out good designs. Intel's current gen chips are fine, compared to AMDs. Not great, not terrible, just competitive.

The reason Intel are worth less is because investors are actually assigning negative value to their fabs - until they see Intel churning out advanced chips again it's just a money pit.

Are AMD chips wearing out after a few years?

They arent fine.

Intel seems to be cutting corners to be faster what made their tech full of security issues. Now those issues are fixed by microcode.. and chips become less efficient.

They also oxidize?

Market cap are by no means an indicator of the future of a company! I bought AMD shares when their market cap was 20x smaller than today and everyone was telling me that AMD didn't stand a single chance vs Intel. I'm pretty sure that you were that guy then.

Reply on news.ycombinator.com