Good news?
Lots of cheap(er) shiny shiny on the way?
The chip industry is on course for an inventory correction in the second half of 2022 or early 2023 with steep inflation, signs of end-user demand slowing, and companies building stockpiles among the causes. This is according to a report from analysts at Jefferies Group, which advises investors on where to place their bets. …
Components may become cheaper, but I doubt that finished goods will, or if they do, it won't be to do with the cheaper components, but with poor consumer demand (companies will really want higher margins on lower volumes, but may drop prices just to maintain share).
We've seen this type of thing before. Manufacturers have bumper years because of some external event, and then expect that subsequent years will be the same even after the event stops. We saw this with flat-screen TVs when TV switched from analogue to digital, and we're seeing it again because of the skewed market Covid caused.
It's interesting looking at sites like ebay to see what bargains can be had. Companies appear to be dumping systems into the second-user market because they've suddenly got more IT equipment of various types than they need. I've recently bought a couple of Windows terminal/lightweight clients (to become low-power *ix systems) and some monitors at knockdown prices.
The difference is that nobody believes Gartner, not the people paying for the research and definitely not the people reading it. Nobody knows why anybody still pays them
The 'chief economist' of XYZ bank goes on CNN to tell the public something in the definite hope that enough people will believe it and act on it to change the price in the banks favour. If this didn't work they wouldn't pay him.
Good luck buying chips from STMicroelectronics
These costing $11 are sold for $300+ from scalpers. Meanwhile distributors give lead times 52+ weeks, that keep getting extended.
People just buy whatever is available, often just in case.
These chips are build on old 40nm+ nodes. Why can't we build something like that in the UK?
There is crazy demand for these type of chips. No need for fancy 5nm processes.
Why can't we build something like that in the UK?
There's no reason you can't, but you don't make decisions like that based on situations like this where there is more demand than supply. You make them accounting for the entire business cycle, including the lean years where you are running below capacity and may be losing money.
>These chips are build on old 40nm+ nodes. Why can't we build something like that in the UK?
Because even a 40nm fab costs $$$
Then you have to find workers. If you aren't allowed to import workers and you have an economy where all your skilled workers are concentrated in one city you are going to have a job getting 1000s of engineers to work in your desolate former industrial size 'new enterprise zone'
then the parts you are making at 40nm are going to go to 3rd world country for packaging/integration so you haven't improved your strategic situation very much
Try buying ANY FPGA. Almost all of these are out of stock, basically with "fantasy" lead times (52+W)... and we got told by one of our suppliers: Don't expect anything from us in the next 4-7 QUARTERS.
Meanwhile we are running completely out of stock
(we have open orders since September/October last year).
Yeah I totally see that this chip crisis ends soon...
Chip supply is horrendous out there, and no sign of it getting better until next year at least. As people above have been saying, 52+wk lead times on lots of stuff, we've been quoted 104wks for some parts. Design them out, then the new part suddenly disappears as well.
Cannot see any sort of decent improvement until mid next year, and even then, it's not going to be great.
If you're going to redesign a module to use a new part, why would you not buy sufficient quantity of that to produce a batch at the time you decide on the new part, rather than wait N weeks till you've done the design work? In former days, waiting would make sense, but now it definitely doesn't!