If it's the latter, I think that would be something to welcome - the official manufacturer can't make enough, so others will. Whether that's legal or not, it's good for the chip buyer. (And chips should be open source anyway, so I don't much care about them violating copyright...)
This article found ATMega328s chips that were actually buck voltage regulators: https://www.sparkfun.com/news/395
Here's an STM32 fake that "works", but has various issues (some serious) listed near the bottom: https://github.com/keirf/Greaseweazle/wiki/STM32-Fakes
Somewhat funny, the very bottom of the page above mentioned that "Note that this chip fixes Erratum 2.3 of genuine STM32F103". So the clone is arguably less buggy than the genuine STM32F103, at least in one aspect.
If I'm paying for a specific chip, I want all the reliability and QC I expect at that price.
You choose whether your wanted the propriety program or the open source program, fully aware that they were different programs with different abilities, features, limitations, and bugs. You could compare them and decide which suits your needs better and then reliably obtain that one.
With hardware the fakes are being sold as the genuine part. You really don't know what you are getting and really don't know if it follows the spec. You might for example think you are buying a 3.3 V microcontroller that has 5 V tolerant input pins (meaning you can hook it straight up to a 5 V sensor you are using), but the fake is not 5 V tolerant and won't work for you unless you add a level shifter to your design.
There was an article submitted to HN nearly a year ago [1] about fake/counterfeit DS18B20 temperature sensors. That article gives a good look at all the different kind of chips you might end up with when you think you are buying a DS18B20, with lots of details on how they differ from the real ones.
> These differences mean the fakes likely have higher power usage and lower sensitivities, even though they are functionally identical. The foundry could have marked these devices as Si24R1, which is compatible with the nRF24L01 and no one would have been wiser. But the lure of higher profits was obviously too tempting.
[0] https://hackaday.com/2015/02/23/nordic-nrf24l01-real-vs-fake...
It would be nice to have a chip that works or doesn't work at all. But there's a large amount that mostly works until it doesn't someday, somewhere and that can make millions of expensive electronics completely useless because some fake $0.02 chip died, plus the aftermath of who's at fault is another can of worms.
A fake chip is a chip that isn't physically what it claims to be. It will not work properly. A counterfeit chip is one that was not made under appropriate licenses. It might be exactly the same as the legit chips. It might even come from the same factory line that makes the legit chips and have been tested right alongside legit chips. The only difference between a counterfeit and legit chip can be a line in a contract on a computer thousands of miles away from the physical object with absolutely no physical imperfections. If you are using a computer, or any other complex machine, chances are that at least some tiny corner of it wasn't made in 100% accordance with IP laws.
If an article wants to talk about "fake" chips then talk about fake chips. Don't go on about counterfeit chips and violations of intellectual property laws. Show me the chips that are not doing the jobs they are supposed to do.
Or that third-shift might, although using the same tooling and having access to IP and processes, cut corners on sourcing or QA.
Or it might use the exact same everything, but skimp on the QA testing and culling.
Or (not third-shift) it might be from a genuine run, but have failed QA.
That's mostly just third-shift. The problems can get worse from there, even if the product at least initially appears to do the job it's supposed to do.
From the article:
> As part of his research, Das regularly monitors counterfeit reporting databases like ERAI, and although it is too early to notice a surge, he is confident that the number of reports will start growing in the next six months as companies realize they have been sold illegal parts.
Further down:
> Although only based on his own observations, Calabria is also convinced that this is only the start of a wave of counterfeit semiconductors creeping into the market. "I think we are on the cusp of a major problem here. The worldwide shortages have opened the door for criminals to exploit the electronic component marketplace, and I'm seeing early signs this has already started to happen," Calabria tells ZDNet.
This is much more a “how do I know the 683rd chip on this reel will meet its sleep current specifications [so our device will make its battery life specs]?” rather than “how do I know this is an IC at all and not a fancy Lego brick?”
That's if they're not just straight up a different (or non functional) part that's been re-marked and then you find yourself re-working hundreds or thousands of boards!
Others have talked about it mainly affecting high-end silicon rather than all components but I assume that most equipment now needs at least 1 high-end component as CPU or microcontroller?
Add some confounding factors like, it takes a long time to build a foundry, in general the CAD is specific to a foundry, not every foundry can make every part, and even if they could, they're not licensed to then: You get a global chip shortage that takes many months to unravel.
So it's not just a capacity shortage per se. Rather, a situation where real demand was wildly different (higher, and lower) than predicted demand, and re-spreading work is hard.
A good OLED TV that is just a panel and doesn't come with a slow horrible computer built in would be nice.
...to the semiconductor equipment sector. We can't get components to make the controllers that control the machines that make semiconductors.
More people are discovering it because they are desperate for chips and they have to take a bet from unknown supplier. Normally you get like 10 box of solid chip with 1 box that are either counterfeit or fake. Now may be two or three.
Some Manufacture are willing to bet because even if the chip cost 5x or 10x as much, it is still relatively small increase from their total BOM. Think Car as an extreme example, who cares if you pay $100 for chips that used to cost you $10 or even $5 when your car is worth $10K+? Not to mention your operation cost losses when you have no cars to sell.
Apparently a lot of the AVR328s on low-end Arduinos (which are still useful, despite much better microcontrollers existing now) are fake too, though I've never had a problem with one.
