Automotive industry is still using 90nm chips. It's not even the generation before the current one. They have long production runs and require stability, and they also prefer to use standardized parts across many production runs.
Defense also uses older chips. Don't ask what kind of chips are in Amraam missiles or the F-35 -- although the F-35 is getting a technology refresh now (after a 14 year production run).
Only in bleeding edge consumer devices does it make sense to keep changing chips. In other systems, production runs can last 20 years and the life of the asset can be 40 years or more, and you want the same spare parts available throughout the entire expected life of all assets produced. And then when you look at fixed assets, such as thermal power stations, then you are looking at even longer time horizons.
For many devices (e.g. a power tool, or part of a vehicle) making a chip that lasts for many years would be more efficient than replacing it with one that drew less power but required a large batch of power to manufacture, plus the cost to dispose of the old one.
Also, there are many hardware components besides CPUs where durability and repurposability would far outweigh operating efficiency.
Companies in my industry have products they sell for 10-20 years. The chips will get used. Product updates often keep the same chip, and the same chip is used across multiple product lines.
EOL for these chips usually means a bunch of firmware engineers in sustaining need to scratch their head and read 10-15 year old documentation to find a viable replacement.
Also, the chip obviously doesn't need to wear out to become less preformant over time. Reliability is perhaps the wrong word, I think quality would suit better.
But in any case, if you have one chip that decreases in performance 10% per year due to security mitigations and another chip that remains consistently performant, it is something to consider when shopping.
All the big machines with computers in them run really old hardware. Airplanes, trains, etc have lifetimes measured in decades. The manufacturers stockpile parts, because these old chips are no longer for sale.
In addition to the other good comments regarding medical/automotive overlap, is that often those chip lines have long lifecycle commitments built into them - ie. we promise to make this set of parts for at least 10 years. For both both of those customer areas, not having to redesign your boards and recertify devices on an external timeline is of value.
It could be for reliability reasons as well, Something which was running for decades in a low-maintenance environment can be expected to continue doing so unlike modern semiconductors which are iterated and manufactured at very high volume that longevity is questionable.
I've been burned by semiconductors earlier due to bad design/manufacturing e.g. Pentium-D, SD 810 etc.
Also with current semiconductor shortage/supply chain issues/inflation, Anything to re-use (or) re-purpose old compute hardware is valuable to many.
This is true, but you may find that if you had to support something (and keep it relevant) across 20 years those specs and requirements may look much the same whether they are military or civilian.
For example, the amount of bending of the case of a device has to be DRASTICALLY less to allow effectively sealing contaminants out for 20 years vice 2, as well as the seals themselves being an order of magnitude better if there is no servicing involved. For most military equipment we have all of those AND regular servicing, something that consumers would absolutely revolt against nowadays.
One other thing people fail to realize on the electronics front: many of the chips in these older systems are getting very difficult to come by. About 10 years ago I was involved in repairing F/A-18 avionics, and one specific chip in that system was extraordinarily important. It was a radiation hardened 80286 CPU, and had a single production run for the entire budgeted lifecycle of the systems it was in. Unfortunately a design flaw in the power delivery systems meant that the CPUs were being destroyed at a rate roughly 4x as fast as expected and they had to figure out what to do. This specific chip was one of the many reasons (but a key one) that we retired that airframe.
Silicon chips age due to electromigration, which is exacerbated by small feature sizes. Chips made 20 year ago could take decades for enough migration to cause failure which is why you’ve never known or cared before. Today due to the much smaller processes we use it’s closer to a few years.
"What's the realistic life expectancy of modern hardware?"
Lower that it used to be. The design life for the Ford EEC-IV engine control system from the 1980s was 30 years. The program is mask-programmed onto the chip, and the parameter table is in a fuse-blowing type ROM. Many of those are still on the road.
With newer hardware, lifetimes are shorter. This is a big problem for long-lived military systems.[1] Electromigration becomes more of a problem as IC features get smaller.
There are embedded systems where 30-50 years of operation is needed. Pumping stations, HVAC, railroad signals, etc. have equipment which can run for decades with occasional maintenance. The NYC subway is still using century-old technology for signaling. It's bulky, but works well.
I would think "right to repair" has a lot to do with this. When you have to replace everything only after a few years, a lot more chips would be needed.
Wasn't there an article here on HN more recently? Anyway, check out https://semiengineering.com/the-rising-price-of-power-in-chi... which also describes a few issues with actual chip aging. That may explain why Intel for example gives only 3 years of warranty, despite the fact that people still believe in "that is solid state electronics, it will last a lifetime!"
Also, the overall cost of design and production is going up for years (>10 years already or so), which is why only these few fabs are left that run these advanced processes and they need massive scale to stay profitable. This "going up" is btw. non-linear, more like exponential and that also applies to power consumption of the resulting chips if you want to eek out "a bit more".
I wonder if the end of chip performance improvements could cause a short term boost in manufacturing investment and sales. Instead of chip investments in products and data centers surely becoming obsolete in a few years, useable lifetimes could be measured in decades if the chip designs and performance are not expected to change. If you can get 30 years out of a data center investment instead of 5, you can spend a lot more money and amortized it for longer.
The other issue is that the older nodes are more reliable. They're already matured where they understand how to make everything work right. On top of that, each process shrink introduces new challenges that can cause components to fail. The most modern nodes seem like they make everything somewhat broken coming out the door with designers building in mitigations for that. They don't last as long.
The older nodes don't seem obsolete if component reliability is a concern. All my concepts consider their potential. They're quite limited in performance, storage size, and energy, though. There's a tradeoff. Lots of companies want a cheap, reliable, simple CPU/MCU. That's where the oldest nodes shine. That said, the newest nodes are tiny enough that one might make a 2 out of 3 setup with extra error correction like Rockwell-Collins' AAMP7G CPU. Might still be pretty cheap... per unit (not development cost)... on 28nm CMOS or SOI process. Haven't seen an attempt.
Similar situation to why you still get spacecraft using chips from the 90s. They need chips which will remain reliable for a significant lifespan, in hostile environments. Way beyond what you're going to expect for a phone or PC.
Quite a challenge when the consequences for failure could be extreme.
FWIW, at least 10 years (maybe at 85°C) is expected of any µP/µC worth its salt outside consumer use. It's just that consumers have accepted somewhat junk-y electronics as a fact of life. (Ok, the improvement cycle is also still quite rapid, meaning a consumer might want to update after 3~5 years anyway.)
Considering warranty periods on some goods, you also really don't want to build stuff with lower rated chips—it'll come back to bite you. Imagine having to pay a professional to swap out the controller board on a smart heating system… people might have accepted swapping out their TV every few years, but for a stove or fridge we're not there yet (and honestly I hope we never get there.)
Leaving the consumer space, it's just expected. If your CNC mill has an x86 control PC, it's a ruggedized industrial build, and those use embedded CPUs with better ratings. Cars are off on a different planet for regulatory reasons, same for aviation or military use.
"Typically the product life of a semiconductor chip (nine to 12 months) is less than the time required to construct the facility and install the equipment for manufacturing (24 to 36 months).
That's an absurd underestimate of market lifetime. I'd bet that fully 80% of the chips available in 1999 when that report was written are still in production today (or would be, if not for the crunch.)
Chips don't suddenly get obsoleted though, they get put in "not recommended for new designs" for at least a year or two beforehand, and when the chip maker plans to phase the chip out they typically give a "last time buy" notice at least a month or two before it gets discontinued.
Chips - most chips - almost never fail from the aging mechanisms described in the article. The folks interviewed are not kidding when they are talking about applications with extended lifetimes past ten years and only seeing these issues at 28nm and below. In your typical laptop or whatever something else breaks first anyway. There is a lot else that can break in any piece of consumer electronics.
Defense also uses older chips. Don't ask what kind of chips are in Amraam missiles or the F-35 -- although the F-35 is getting a technology refresh now (after a 14 year production run).
Only in bleeding edge consumer devices does it make sense to keep changing chips. In other systems, production runs can last 20 years and the life of the asset can be 40 years or more, and you want the same spare parts available throughout the entire expected life of all assets produced. And then when you look at fixed assets, such as thermal power stations, then you are looking at even longer time horizons.
reply