It depends on the stage of the supply chain. It seems the OP is is talking about 2 cents of sand in a silicon wafer to make a one dollar chip.

Your radio shack example, the finished "manufactured part" is a packaged resistor in a consumers bag on the counter next to a cash register. Counting only one raw material is foolish and not a standard to measure whether or not an enterprise is profitable. If it was, we wouldn't be talking nostalgically about Radio Shack -they would still be a profitable enterprise.

Unfortunately, manufacturing costs include, in addition to raw materials also include labor and energy. These threee items, Rams, labor and energy are typcially referred to as variable costs and are proportional to the quantity produced.

Even more unfortunately (especially for radio shack) fixed costs also exist. They can be (commonly are) much higher component in manufacturing costs. Interest on borrowed money, depreciation on assets, taxes, insurance, at the manufacturing location.

A boule of silicon can be converted into a pile of chips quite cheaply. But, doing it without a billion dollar Fab is quite a feat.

Also, a pile of resistors can be sold quite cheaply but, RadioShack couldn't compete with their network of brick and mortar stores when mailorder houses can do it cheaper -without the assets.

And another huge cost that is relevant to the TS's dream -- NRE (non-recurring engineering) costs. That has killed a number of these kinds of projects (even ones that were rationally thought out to begin with). People think that they will do all this in their spare time and often discover that they don't have that kind of spare time and that using what they do have for something like this amounts to an unacceptable opportunity cost for them.

As for ole Radio Shack, I would have loved to see them stick around largely as what they were back in the 70's and 80's. I don't know if they could have made that happen by changing their business model so as to tailor to and actively foster and create the hobbyist community, perhaps even partnering with schools at all levels. I think something like that would have been what it would have taken, and I'm far from certain that it would have been sufficient. I would not be surprised to learn that they considered many possible paths including some along these lines and concluded that they weren't viable.

 

I guess it would depend on exactly what is included in "manufacturing costs". Also, to keep it somewhat apples-to-apples, the TS appears to be talking about the price to the end consumer, which generally entails many companies buying parts from other companies, adding value, and selling them on to other companies before it gets to the end consumer.

So I think of the old classic of the two 1/8 W resistors that Radio Shack would sell for $1.19. Those probably well exceeded a 50x markup compared to what it cost to get them off the line. Small LSI chips probably fall in that same category today. Even if a 74HC00 chip were free to make, the supply chain needed to get it to the hobbyist consumer would probably leave it costing about the same that it currently does. The small PIC MCUs may approach that, especially when the end user is a onesy-twosy hobbyist and if "manufacturing costs" only includes the actual costs needed to get a functioning part made in a fully depreciated fab.

But I wholeheartedly agree that any single company having a 98% profit margin on anything that has any market and that any other company could conceivably produce is not going to have that margin for very long.

This is the financial principle of arbitrage. The ability to make unlimited risk free profits. Naturally as you would suspect the opportunities appear quickly and disappear as soon as market participants figure out what is going on. Does anyone remember the Hunt brothers and their ill-fated attempt to corner the Silver market?

https://en.wikipedia.org/wiki/Silver_Thursday

Margin call for $100e6 anyone?

 

...2 cents of sand in a silicon wafer...

You're getting robbed, GT. In SoFla, we get our sand for free.

 

You're getting robbed, GT. In SoFla, we get our sand for free.

I've heard about the eductational system in SoFla. I wouldn't trust the silicon from SoFla sand to make important calculations in my CPU.

 

You're getting robbed, GT. In SoFla, we get our sand for free.

Of course, as soon as you pay someone to scoop it up and dump it into a crucible, it is no longer free.

My dad worked for an air compressor manufacturing company most of his life and they had a sign over the parts counter that simply read, "Air is free... until you try to compress it."

 

Of course, as soon as you pay someone to scoop it up and dump it into a crucible, it is no longer free.

My dad worked for an air compressor manufacturing company most of his life and they had a sign over the parts counter that simply read, "Air is free... until you try to compress it."

This was my not-so-explicitly-made point. And it was for the benefit of OP with respect to his misunderstanding of pricing.

 

Example: an arduino MCU costs 3 dollars, an Intel MCU costs 300 dollars.

An Arduino can do about 100,000 floating point calculations per second.
An Intel chip can do about 1,000,000,000,000 (1 trillion) floating point calculations per second.

That is 10,000,000 the performance for 100 times the cost.

So the Intel chip is quite a bargain.

Bob

 

How can I make a custom graphics card with DSPIC or DSP TigerShark? Preferably DSPIC because its cheaper, I think its time we start thinking about our own computers and phones given how things are going. If we start slowly sooner or later we will reach where we need.

EDIT: I think we should make it appropriate for linux if possible. Maybe Ubuntu or for starters some DOS version of Linux?

It already exists. It is called a Raspberry Pi. It has the graphics controller -- the CPU, memory and the rest are there for free.