Yes. But you're talking about a very serious, ultra-skilled startup that will burn millions in a blink. Not easy, not low risk. The expertise needed to stand any chance of success is considerable. Sure that experience exists, but even the best minds get tripped up by the speed of change in crypto, the market instability, etc, etc.
Yes, you are right - this is a very serious project. In our country they say "this is not selling seeds in the market".
But in our country there is another proverb: "The pots are not molded by holy clergymen." This means that even very complex work (for example, spacecraft) is made by ordinary people.
About millions.
At the initial stage, when I will develop a logical cell and a transport cell, a lot of investment is not necessary. It is enough for me one ordinary salary.
Further, when we have cells and there will be a grid of cells that can be shown in a visual simulator and give it a simple program - you will see that everything works. Then it will be possible to connect another 1-2 engineers, so that together with them we would finalize the devices necessary for the chip: input / output, clock frequency oscillators, data distribution, block control, programming bus, memory blocks.
At this stage, you will need to pay 2-3 engineer salaries.
Now we will have a ready chip, its simulation, source code and testbench.
There comes the third stage - the manufacture of the prototype. This is a multi-project wafer. If you make a project on technological standards up to 22 nm, I think $ 100,000 is enough.
While porting the RTL code to the topography ... while photolithographic plates are being made ... while test chips are being manufactured - we develop a motherboard for the chip and produce several test samples + we write a test bitstream for this chip for any cryptocurrency we choose.
After that we get the chip, we program it and we get the result of its work.
According to the result of the work we make calculations of productivity and economic advantage.
Then you can attract third-party investments, because people will see the real prototype and its capabilities. I do not think that the modern cryptoindustry will not be able to collect several million dollars if it is confident of a profitable result.
And the risk in this project is minimal. I already wrote about this.
Thus, in the early design stages, you need only a few thousand dollars to pay a couple of engineers.
Next you need about $ 100,000 for the manufacture of test chips and motherboards for it + bitstream development and testing.
Everything. These are all costs that early investors will incur.
And the result - it will be a guarantee of solid income from the mining of altcoins or the sale of miners on the FPGA.
I don’t think that Bitmain or Canaan companies regretted designing chips. But it all began with the same idea as here. Then there was an industry request for ASIC. Now there is an industry request for cheap FPGA. Nobody is doing this yet.
So it needs to start doing this now !!!
If we don’t start doing this, then someone else will start doing it later and remove all the cream from the market.
About the best minds.
For FPGA, it doesn’t matter how complicated the cryptocurrency hashing algorithm it processes. FPGA has a fixed device.
The question is: how much LUTs (LABs) is available on the chip for programming, how much RAM is available on the motherboard, how much static cache memory is available on the chip and what is the frequency of operation.
My job as a developer is to make as fast cells as possible. Everything else is a long-standing standardized problem, which is solved in many micro-electronic devices. For example, even the integration of the ARM core into a chip is no longer a secret innovation. Much of this is already ready. It is only necessary to take and connect together.
In our country they say: "The wolf is not as scary as it is depicted in the pictures." This means that many tasks that seem very difficult are actually simple. Only one thing adds complexity to them — few people take on these tasks.
But do you really think that it is too tough for us?
If the crypto-community can finance the development and manufacture of the test chip + board, then I will prove that we can do it!