9 ms·
TSMC warns AI chip crunch will last another 18 months
- sbbq 3y agoEveryone is buying up H100's (and A100's if they can't find H100's). Sure, in 18 month people may lose interest in buying H100's but then there will already be the next Nvidia model everyone wants to buy to get ahead of the competition in terms of compute capabilities.
- boringg 3y agoSounds like this will be the decade of chip shortages.
- motoboi 3y agoYeah, it sounds like they are saying we don’t know how much this will last, but at least in the next 18 months the trend will continue
- lofaszvanitt 3y agoTime to open source the chip making industry... ... ... This whole fabrication thing is a biiig chokepoint.
- wincy 3y agoAbout as likely as open source nuke design.
- alwayslikethis 3y agoIf we ignore the sanctions you'll be getting, open sourcing nuke design is much easier.
- hef19898 3y agoBuilding one the WW2 designs isn't really that hard today, is it? Getting the fission material on the other hand... Just ask Doc Brown, because it is not the 80s anymore were you get Plutonium at pharmacies, ah those days!
- pests 3y agoIt depends. Little boy would be easiest as the speherical detonation and materials science for the Fat Man implosion would still be difficult to design and machine. I also do not beleive the activator designs have ever been declassified.
- hef19898 3y agoDoesn't have to be an exact copy so, but yes design might be tricky. I assume solvable so, at least in a primitive and workable way. Machining, to be honest, beside size of the parts themselves, is something you can do in some reasonably well equipped maker space. And for size, CNC machines are readily available on the used market. And those are light years ahead of anything the Manhatten project had. Or Apollo for that matter (excluding the Saturn V of course, but that is rocket science).
- trashtester 3y agoFrom my undertanding, Fat Man requires a very precise design to focus the shock wave properly, or the bomb will never detonate properly. This design is quite non-trivial even today. It took several of the brightest minds of all time to get it just right back then (like von Neumann). Even with modern tools, you still need a team of quite capable engineers/scientists to pull something like that off, even if you do have the plutonium. The little boy design is easier to engineer, but bomb grade uranium is even harder to produce than plutonium.
- hef19898 3y agoTotally agree, the science / design part of it is still not trivial. Also I know next to nothing about it. The machining bit so, uaing proper machinery, is easy. And the necessary engineers and technicians readily available.
- trashtester 3y agoNot even a 1000 normal engineers can replace a single von Neumann. What makes it easier, though, is that the general principle is relatively well known, so one could always brute force the solution using computer simulations, even if unable to do the math by hand.
- FirmwareBurner 3y agoNorth Korea can build a Nuke without OS. I doubt they can build a semi fab even with OS.
- HankB99 3y agoIsn't the cost of the equipment and sophistication required to operate the biggest barrier to entry? How would open sourcing change that?
- FirmwareBurner 3y agoIt wouldn't. HN lives in a SW bubble where Foss is the solution to everything , but fails to see that capital is the main barrier in the semi industry and, unlike their Node library, the secret sauce of cutting edge chips is a matter of national security at this point and will be kept as close to the chest as possible.
- adrian_b 3y agoActually if the IC design tools would be FOSS, the cost of new designs would decrease considerably, even up to two times. The cost of the subscriptions to the design tools provided by Cadence, Synopsis, Siemens (Mentor) and a few others is ridiculously high and it can contribute with many millions of $ to the design cost of a state-of-the-art chip. The design tools themselves have been improved extremely slowly and they are of a quite low quality for so expensive professional tools. In an open-source environment it would be easy to make much better alternatives. The reason that only a few timid attempts exist to provide alternatives for only some parts of a complete toolchain, is that the semiconductor manufacturers, like TSMC, do not provide public information about the exact data that they need as input for manufacturing, about the design rules that must be implemented for valid designs and about the characteristics of their processes that must be known for simulations. All this information is available only to big companies like Synopsis, Cadence etc. and this perpetuates their monopoly and their ability of extracting huge amounts of money with minimal software development work, from all the companies that design ICs. While the companies that design ICs are not happy for this huge rent that they pay for the design tools, they are also extremely risk-averse and they do not want to make any change in their design workflows, no matter how small, that might increase the risks of failure in a designed chip. Because of this there are no chances of replacing the incumbents with open-source tools, at least not in the big established companies that design chips.
- 0_____0 3y agoThis is exactly the kind of thing that will never be open source. Any entity with the extremely high level of resources to fabricate chips at this node have zero incentive to open source anything, and no number of amateurs with 3D printers can change that.
- lofaszvanitt 3y agoWould be nice to fabricate your own chips.
- sloreti 3y agoRead Chip Wars [1]. The cost to stand up an advanced chip fab has reached tens of billions of dollars. It’s so capital intensive that even the US Department of Defense has exited the game. [1] https://en.m.wikipedia.org/wiki/Chip_War:_The_Fight_for_the_World's_Most_Critical_Technology https://en.m.wikipedia.org/wiki/Chip_War:_The_Fight_for_the_...
- champagnepapi 3y agoI’ve read it, and I agree with you 100%. Open sourcing won’t don’t anything with an ASML machine costs $300,000,000. Open sourcing software works because some many people already have the hardware (a computer)
- viscanti 3y agoAnd it's not just needing to buy the ASML machine. There's a ton of proprietary knowhow to actually get it integrated into a fab so that yields are reasonable.
- samtho 3y agoSemiconductor manufacturing requires a lot of capital expense just to open the doors, it also requires a staff of all levels to keep the machinery working and multiple suppliers for raw materials that require multi-thousand dollar minimum orders. The process these facilities implement are fairly well understood, it’s just cost prohibitive. If you’re trying to draw parallels between this is and, say 3D printing, you will not find many similarities. PCB fabrication is also a a different beast but a lot of people think it’s as easy as finding a fab house that takes gerber files and creates a board for them. If we want something that more closely resembles how we develop software, we probably will need to develop processes that enables us to flash FPGAs like MCUs, make them fully read only, and allow the customer to etch their own marking onto the chip. There facilities that produce ASICs like this already, but I can see something more standardized becoming more successful for general purpose. This does however require that we have a foundry create the original chips, the difference here is that they are producing the same standardized chip for everyone which would allow any foundry to make compatible alternatives.
- lofaszvanitt 3y agoYeah, we know this way. It's as convoluted and deliberately complex as building a nuclear plant.
- jrk 3y agoNote that the shortage now is different than the past few years: it is not in fabricating any of the primary chips, it is entirely attributed to insufficient capacity in TSMC’s CoWoS packaging. This is only used for the highest-end accelerator products that use HBM (and a few other things that use full silicon interposers like Apple’s M-Ultras), they are scaling capacity, and there are increasing numbers of alternatives from other packaging vendors that chip designers could use if they had known TSMC’s packaging would be a bottleneck. Demand spiked a lot all of a sudden, and it just takes time to build factories – there are no obvious constraints.
- wing-_-nuts 3y agoI wonder how this will impact consumer gpus? If nvidia can print money making h100's you have to wonder if they will continue to invest on the retail side or if they'll funnel most of their efforts to server and let desktop parts languish.
- 654wak654 3y agoThey're selling consumer GPUs with huge margins too. The markup on the 40xx series is insane but they're still shipping a ton of units.
- pandemicsyn 3y agois that the case? They sold a bunch early on but theres been lots of press about the lousy 40 series sales volume this year and their struggle to get people buy 4080/70's. In Feb they artifically constrained 4090 volume to try and push 4080s no one wanted. The 4070 was dead on arrival and had production cut by April. They've near stopped production now - theoretically because the AI stuff is so lucrative, but if they were really "shipping a ton of units" of the 40 series and didn't have excess inventory i doubt they'd have done that. - https://www.dexerto.com/tech/nvidia-rtx-4070-production-reportedly-cut-2120600/ https://www.dexerto.com/tech/nvidia-rtx-4070-production-repo... - https://www.dexerto.com/tech/nvidia-rtx-40-series-production-ai-gpu-boom-2235345/ https://www.dexerto.com/tech/nvidia-rtx-40-series-production...
- Miraste 3y agoI blame it on Nvidia's product segmentation falling apart. Only a few years ago, the 1060, 1070, 1080, and Titan Xp had price differences that matched their performance differences. In their current lineup, buying anything that isn't a 4090 is getting ripped off. The 4060 and 4070 are basically e-waste. The 4090 had a 100% performance improvement in a lot of tasks over its predecessor, and the 4060 managed 10%. Sometimes. I don't know why they bothered making them--it's no wonder they cut production.
- 3y ago
- TradingPlaces 3y agoThis certainly explains why Camtek got a surge of orders for their Eaglet-AP inspection machine for advanced packaging just in the last couple of months. https://www.camtek.com/news-and-events/camtek-receives-orders-for-45-systems-since-the-beginning-of-august-2023/ https://www.camtek.com/news-and-events/camtek-receives-order...
- samstave 3y agoCurious, does the shortage squeeze company stocks up or down, or hold them steady?
- Roark66 3y agoI'd bet $50 in 18 months "the crunch" will still be in full swing (and Raspberry Pi will still be "bringing huge volumes to the market any week now"). There is no way in hell all these expensive fabs will pay for themselves if chips are too cheap. So no one (especially not TSMC) will risk oversupply by building too many fabs.
- pclmulqdq 3y agoThe Raspberry Pi wouldn't have issues if not for the fact that they are explicitly an outlet for Broadcom's excess chips, and they have never paid full price for their CPUs. If they doubled the price of the boards, they would be able to get the parts.
- selfhoster11 3y agoIf they doubled the price of their boards, they would simply die to the entire hobbyist market.
- Miraste 3y agoWhy don't they? The current market price is 3 or 4 times the MSRP and has been for years--the Raspberry Pi has lost its place as the affordable hobbyist board. I suppose it would be admitting defeat, but if they'd actually pay for parts and make new Pis again, I'm sure they could at least bring the market prices down.
- nsteel 3y agoThe price in my market is back to normal for the versions I want. Has been for a month or so now. I don't see why the other versions won't follow soon enough, exactly as was promised at the start of the year.
- sangnoir 3y ago> the Raspberry Pi has lost its place as the affordable hobbyist board I keep hearing this - but I've seen no evidence of any successor board. Which board would you say is now the go-to board for hobbyists?
- baz00 3y agoThat's when the bubble will burst.
- dwaltrip 3y agoAh yes, just like when the internet died in 2001… —— Being a bit more fair, yes a bubble did burst then, but it was a “too much, too quick” type of bursting, not a “this whole thing is a scam” type.
- baz00 3y agoI'm not saying it's not useful, but much like the internet, claims have been overstated.
- shock-value 3y agoLLMs/transformers are a breakthrough on the level of convolutional neural networks — significant, and one that opens up lots of new interesting applications as well as invites lots more R&D — but like the neural net it’s not gonna fundamentally transform society or industry. (And as far as I can tell this current craze is entirely predicated on LLMs/transformers.)
- riku_iki 3y agollm/transformers are 5 years old already. llm companies raise lots of money on crazy evaluation, the question is if they will be able to build enough revenue stream to justify further money injections.
- the8472 3y agoAutonomous cars alone will be transformative. And LLMs are not in their end-state, we went from GPT-1 to 4 in 5 years over that timespan it gained lots of new capabilities. Extrapolate, don't just look at what they can do today. Atoms are more difficult than bits, so physical applications (e.g. in factories, robotics) will lag. I also expect specialist models like AlphaFold slowly finding their niches over time. > significant, and one that opens up lots of new interesting applications as well as invites lots more R&D — but like the neural net it’s not gonna fundamentally transform society or industry. Yeah, and semiconductor electronics that grad-students built from of germanium by painstakingly pressing them together are very interesting and worth additional R&D but won't transform society either.
- everdrive 3y agoAnother 18 months of relative peace. I wish it would last longer.