Is centralization of mining in China a problem? Advancements in hardware for mining, Moore's law, xenophobia.
Transcript
[ANDREAS] Hey, I am glad to be here. [AUDIENCE] I have been thinking about the centralization of mining in China. [Inaudible] [ANDREAS] I am concerned about any form of centralization. I would certainly like to see less, but that doesn't necessarily mean it is a problem now.
The centralization in mining that we have today in China is the result of a technology adoption curve, that went from CPU, to GPU, to FPGA, to ASIC over a period of four years. With each change in architecture, [there was a massive] increase in performance, which means the equipment is obsolete within two to three months. [Operating under those conditions] requires massive capital and direct access to chip factories. Selling chips to others doesn't make economic sense.
If you just keep them, mine for two to three months until they are obsolete, you [earn] more than if you sell them. Of course, access to cheap electricity is a huge advantage. All of these factors have created a [very steep] curve towards centralization, primarily in China. Yet [we have now reached] 16-nanometer fabrication, which means the 10,000x performance increases and two- to three-month obsolescence periods are over.
Moore's Law is [a law about the pace in growth of processing capacity]. But compared to Bitcoin's [transition from] CPU to ASICs], [the prediction] was actually too slow, about 5,000 times too slow. Over the next two years, we will not see 10,000x performance improvements. We will see a doubling, and that is really slow.
Instead of equipment being obsolete in three months, it will now have a shelf life of two years. That means mining with the chips you create, instead of selling them, is no longer an advantage. You will have all the risks of warehouse operations, government interference, and electricity prices; if you sell the chips, it is less risky. The [factors] that led to centralization of mining no longer exist.
Mining changed six months ago, when we reached 16 nanometers. We haven't seen [the ripple effect] yet in that industry. It will take about a year and a half. The Halving will be very important.
I don't think the centralization will last. Mining centralization was an artifact of performance increases and limited shelf life that no longer exist. Those advantages have now become risks. If you put a hundred thousand miners in a warehouse and lose power for two days, your profitability is wiped out for the month.
But if you put a hundred thousand miners in a hundred thousand kitchens, and use them to toast your bread, or cook your sausage (as Sven showed me today)... He cooked hot dogs on an Antminer S7. That was cool. [Laughter] [If you distribute the mining equipment], you will have the advantages of decentralization.
If the power goes off in one kitchen, it doesn't affect anybody and you are [not dependent on] profitability. So all of the [factors] that made Chinese centralization happen are [being] unmade. I want to make another point: if mining centralization had happened in Sweden, [certain] people would be a lot less "concerned." Not just because the Chinese government is a threat to Bitcoin. If you think about it, the people mining in China are directly opposing their government by doing it, and they are taking enormous risks.
They are entrepreneurs of the finest capitalist attitude, who should be applauded for their incredible engineering and operational skills. They are not their government. They will resist and fight to maintain independence. We should applaud that, not fear it.
Somehow, I think if they were Swedish we would be less concerned. There is an element of racism and xenophobia in this, of stereotyping an entire race, and I think that is wrong.