Genuine question because I’m unfamiliar with the market. Why aren’t competitors undercutting these big players? 40% is ridiculous
Samsung, which is the largest IC manufacturer, recently increased prices: https://www.tomshardware.com/news/samsung-foundry-to-hike-wa...
If you look at prices for the stuff a lot of wafer capacity is going towards, particularly Flash and RAM, you'll see that the prices have gone up 20%+ over the last year.
I think people like to pick out TSMC because of the politics surrounding Taiwan and China, and maybe because a lot of companies use TSMC for their IC production and like to complain. People don't notice as much when other ICs like flash, ram, micros, etc are slowly increased in price.
Also there is only one supplier (ASML) in the entire world who makes a critical machine (extreme ultraviolet laser photolithography machine) needed for these cutting edge processes. It is unable to match the increased demand for their tools. And even if they were these super advanced might as well be magic machines have a lead time in years themselves as do the factories they go in to.
If you want to start marking modern semiconductors or increase your existing capacity to do so it isn't as simple as throwing a couple million at it and having it running in a few months. Its billions of dollars in a massive sterile factory using machines made by a handful of suppliers taking up acres of land and consuming impressive amounts of electricity and water to do so. Semiconductor fabs are massive sprawling industrial campuses https://images.anandtech.com/doci/13749/intel_kiryat_678_678...
actually iirc they've tried to get into the custom foundry business a couple times now (see: "Intel Custom Foundry" that got discontinued a couple years ago) but they never really attracted any major customers.
https://semiwiki.com/semiconductor-manufacturers/intel/7912-...
Part of this was due to the fact that Intel's design rules (the "rules" that govern chip design: how small wires and transistor features can be and so on) are completely different from every other foundry, so while porting from TSMC to Samsung or something like that certainly isn't easy, it's at least straightforward, whereas Intel's stack is basically totally different from everyone else's. One of the advantages Intel has is that the chip design and the process are tightly intertwined, if there is a problem with the chip design the designers can go down the hall and talk to the process people and see if they can come up with a solution at the process level. This obviously also became their downfall when 10nm slipped, and likely also it severely reduced their ability to sell their foundry services to other customers.
According to that article they were apparently also rather secretive about their process (seeing it as part of their "special sauce" I guess) and I've heard many customers doubted Intel's commitment to the program as well as their neutrality. And I think the final piece of the puzzle was probably price - I'm guessing Intel was asking a royal sum compared to GF or Samsung or TSMC. After all... they're intel, and their process is the best!
> One of the funniest stories I heard was about the first copy of the Intel 14nm design rules Atera got from Intel. They were heavily redacted, which is something I had never seen in the foundry business. After many delays Intel put their own implementation team on the first 14nm Altera tapeout and the result was a very competitive FPGA chip. If not for the continued delays, Xilinx would have been in serious trouble as the Intel 14nm FPGA, based on my experience with customers, beats the Xilinx 16nm in both density and performance.
They are now trying to get back into the custom foundry business of course. Different management, and a very different business situation for Intel - if they can get other customers to help finance their node development that's a big win for them at this point.
[1] https://www.anandtech.com/show/15428/asml-ramps-up-euv-scann...
Something doesn’t add up there.
No offense to his accomplishments, but the modern fabs like TSMC are so advanced you cannot just make the things they make without billions of capital investment. It just is not possible.
Unfortunately, many of the chips in short supply are on pretty small nodes, so you need extremely specialized equipment limited to a few of the big players to make them.
Certainly there's some stuff that can be made in larger nodes, but if you want something in 5nm you basically have one game in town.
The shortage won't last that long.
But in the meantime, anyone who owns a suitable factory can pretty much set the prices as high as they like. I'm really surprised the price hike is only 20%, rather than 10% per month for the next 3 years.
Some other government? This wouldn't have much impact on chips, except to make it a lot less likely anyone makes chips in that country ever again — which isn't to say no nation would do it; economic history is full of this sort of thing, particularly as nations start going the way of Venezuela.
For most uses of tech, demand is there even if the price doubles.
The entity that manufactures the product is compensated for it, meaning the money isn't being siphoned off by a random third party. Price controls grant a wind fall to consumers who are in no position to build another semiconductor plant. It's basically a wealth transfer away from the semiconductor industry into the consuming industries. If you want semiconductor prices to drop then either grant a subsidy or let the government build a plant that produces way more than the market really needs.
A single shortage plus price increase might not be able to pay for a whole plant, but any plans to expand capacity will be shifted forward rather than be delayed. If the profit motive exists the construction of the new plant will happen one year sooner and it will be bigger in anticipation of future shortages.