What are production costs of ASIC chips and devices?

[User705]

Does anyone have some estimates regarding the bottom line production costs of ASIC chips per chip so to speak?  Just trying to figure out at what fiat point is the floor reached on some of these ASIC products.  For example the ASICMiner USB is bound to stop being produced once ROI says it should be $5USD.  It probably costs a couple bucks to make plus shipping plus packaging plus a retailer markup etc...  So if network doubles or triples shortly the USB will be unsellable with the current exchange level.

[1714005789]

1714005789

[tymothy]

So if network doubles or triples shortly the USB will be unsellable with the current exchange level.

Not necessarily. You have to assume people will not purchase a mining unit when there's a negative expected ROI which isn't true for economic and non-economic reasons. People may expect bitcoins to appreciate and anticipate a positive ROI. Or they may value being able to participate in securing the network, the novelty or the conversational value that a flashing USB stick can have. If USB ASICs drop to $5, these non-economic factors become even larger. People will pay $10 for a humping dog USB gadget which generates 0 bitcoins, totally plausible a USB ASIC could command a premium too.

Also, your analysis only applies to the current generation of USB ASICs. Chip advancements typically allow for faster and faster chips with almost no increase in fixed unit costs. If ASICs get substantially faster (which we would expect under Moore's law) the ROI will look less shabby.

[Rannasha]

So if network doubles or triples shortly the USB will be unsellable with the current exchange level.

People may expect bitcoins to appreciate and anticipate a positive ROI.

This is a bad reason to invest in mining hardware. If you expect bitcoins to appreciate, you buy bitcoins and either hold them or put them in low-risk investments.

If a mining investment (or any other investment) today costs 1 BTC, then it better be worth more than 1 BTC (profits paid out + resale value) at some later time, or it was a bad investment.

This doesn't take away the fact that there are plenty of people buying mining hardware using the same exact reasoning ("if BTC appreciates i'll make a profit, despite difficulty increases") and that's one of the reasons why mining is unprofitable for small-scale miners: There are too many people willing to mine at a loss.

USB miners are novelty items that can be fun to have and use, but not investment opportunities. But not every hobby needs to make you money.

[User705]

What I meant when I said unsellable is that the extraneous costs such as shipping and handling will be far greater then the device itself thus no manufacturer will make them since they can't sell them at a profit.  Yes I was referring to gen 1 but this will apply to gen 2 and also to the single chip on single USB stick setup.  Also when you talk about a humping dog USB that provides an unlimited amount of viewing pleasure and a USB miner is just a USB miner which does not provide unlimited BTC.  Trinkets are just that trinkets but I'd venture the majority of USB buyers bought them to make BTC.

[Pinwheel]

production cost of 1trh for ASICMINER blades was USD 10K few month ago. Cost of devices will not be mjor restrictive entry point factor soon. Profit from mining will be very small and unless you think on terahash scaleit does not make much sense to get involved at all. Once all major miners will be on same page with 28nm chips, only efficiency of devices and cost of power that what will matter. Cost of devices will cease to be major factor. You can trow more 10-20 trh, no problem it will be cheap and available ready to ship. In such situation people will look into two factors cost of power and cost of btc.

Suppose, I own wind power generation plant or hydro power generation plant I will keep adding more and more terrahashes, till everyone else will jump of the train.

[User705]

production cost of 1trh for ASICMINER blades was USD 10K few month ago. Cost of devices will not be mjor restrictive entry point factor soon. Profit from mining will be very small and unless you think on terahash scaleit does not make much sense to get involved at all. Once all major miners will be on same page with 28nm chips, only efficiency of devices and cost of power that what will matter. Cost of devices will cease to be major factor. You can trow more 10-20 trh, no problem it will be cheap and available ready to ship. In such situation people will look into two factors cost of power and cost of btc.

Suppose, I own wind power generation plant or hydro power generation plant I will keep adding more and more terrahashes, till everyone else will jump of the train.

1trh is 1000 Gh yes?  Recent prices for hardware are going quickly to .5 btc / 1Gh and lower which equates to 500btc per trh.  You are saying the cost is around 100btc per trh which means at same exchange rate it would no longer make sense for asicminer to produce gear even for themselves somewhere around five times current network if not sooner. 

[Satobit]

My friend who works in the ASIC industry told me that he estimate the cost of the avalon chip is less than $1 each.

so for a 4 module, the cost of the chips alone is less than $300. plus the PCB and other components and labor. should be less than $500 for a 4 module avalon miner.

For the people who has the control of the chip productions, they can still be profitable with their 1st gen ASIC even with the difficulty that is 10 times of the current.

[User705]

My friend who works in the ASIC industry told me that he estimate the cost of the avalon chip is less than $1 each.

so for a 4 module, the cost of the chips alone is less than $300. plus the PCB and other components and labor. should be less than $500 for a 4 module avalon miner.

For the people who has the control of the chip productions, they can still be profitable with their 1st gen ASIC even with the difficulty that is 10 times of the current.

That doesn't assume forward discounting.  Figure you need 50% gross margins so that puts a rock bottom price at 750 each and current prices are about 4k so I figure five times the total network rate and that's all she wrote for 110nm gen 1 at current exchange rates that is.