3 ms·
I'm a little bit confused why it has to be so expensive and so hard to produce chips... Is it because of the scale required to sell them at a low enough price t
by taf2 5y ago
I'm a little bit confused why it has to be so expensive and so hard to produce chips... Is it because of the scale required to sell them at a low enough price to be competitive? If so, is there maybe a market for smaller fabs that only produce small scale batches of specialized chips?
- mensetmanusman 5y agoIt takes on the order of a thousand steps to build them, each of those steps is supported by million dollar equipment, hence the cost.
- akvadrako 5y agoThe machines that can make modern chips are extremely expensive. On one hand they are sensitive enough that shipping them costs millions of dollars since they need to stay in a vacuum. On another hand they need a room-sized laser installed nearby.
- SenHeng 5y agoWhile I don’t disagree, saying that expensive equipment is expensive doesn’t really explain why it’s expensive in the first place. The next two statements are more helpful.
- thedday 5y agoAnd water. They need lots and lots of water. And everything has to be ultra clean and to start that way in creation, shipping and in use.
- sbierwagen 5y agoA new fab would be competing with old fabs that are already amortized/depreciated, and thus have lower capital costs.
- Cyph0n 5y agoA few reasons: 1. Extremely expensive equipment, both directly related to IC fabrication (e.g., lithography) and indirectly related (e.g., air filtration, water processing, stabilization). 2. A large amount of skilled labor onsite to operate and maintain the complex equipment. 3. Constant supply of resources, including power, water, and sand.
- MangoCoffee 5y agofoundry is a capital intensives business. the processing steps that is involved is like 300+. that's why you hear the term "yield" in the semis foundry business. each step that went wrong is going to affect your yield. here is a short video that show the semis fab process. https://www.youtube.com/watch?v=c9arR8T0Qts https://www.youtube.com/watch?v=c9arR8T0Qts
- hinkley 5y agoThe more times you have to fiddle with something, the more opportunities you have to break it. It's compound interest essentially, and that can add up real fast. If you have 300 steps and by some freak coincidence, each step had exactly the same error rate, so that you lose N% at each step, then here are the overall yields: 1% loss: 4.9% yield 0.5% loss: 22.2% yield 0.1% loss: 74% yield 0.01% loss: 97% yield You have to have an error rate in the range of X failures per thousand per step to get above 50% yield, and X/10000 for a really good yield. And all of these things are so small that a spec of dust causes failures.
- ksec 5y ago>I'm a little bit confused why it has to be so expensive and so hard to produce chips. You are talking about something that is forever edging towards atomic size manufacturing. $7B is hardly expensive in today's terms. Not to mention the scale. Annual Smartphone sales is 1.2B unit. You have multiple silicons within a single smartphone. >If so, is there maybe a market for smaller fabs that only produce small scale batches of specialized chips? Smaller Fabs will just be as expensive. You built larger fabs ( as you have noted ) to try and amortised over larger volume. ( Lower Unit Cost ) Which is the problem why NAND and DRAM capacity planning are hard, their Fab size have grown to a scale that it is hard to build additional one without some careful consideration. ( High Cost, High Risk )