Back in the day, the step from 55nm to 28nm chips was considered a blistering progress. Some of the modern ASICs like the BW-L21 are based on 28nm chips and they offer some impressive numbers hashrate-wise.
Nowadays we also have access to 14nm ASIC miners such as the Innosilicon A4 Dominator for Scrypt and the Ebit E9 Miner for Bitcoin. Those offer even better hashrate/kilowatt ratio than ASICs based on larger chips.
As expected, technological progress does not stop there. Apparently, IT companies are rushing to start manufacturing 7nm chips and hardware that would be based on those.
(Lower numbers are better)
The four main competitors are Samsung, Intel, TSMC and Global Foundries. After investing about $ 4.8 billion USD into the development of 7nm facilities, Samsung took the lead in the race and apparently is a couple of nm ahead of its competitors (no pun intended).
GMO Reaps the Benefits
Most of the mentioned companies expect to start manufacturing products based on 7nm chips by Q1 2018. For instance, AMD already announced that it is releasing “Starship” in 2018, a 7nm-based GPU with 48 Cores. So far so good, but what about 7nm ASICs for Bitcoin and other cryptocurrencies?
Recently, GMO – a Japanese internet giant and Bitcoin exchange operator announced that they are partnering with one of the four mentioned companies to be able to develop a new ASIC that will be based on 7nm chips.
Good news is – GMO will be selling those ASICs to the public. I will talk about it in a minute.
Bad news is – the company will also create their own mining farm in Northern Europe. Their plan is to have 50,000 7nm chips to deliver a total hashpower of 500 petahashes per second (PH/s). Once built, the farm will indeed be a considerable difficulty bomb and it will rank among the top ten Bitcoin mining pools in the world.
Having such a solid background, it is very probable that GMO will eventually achieve their goal and become yet another mining giant. This is indeed bad news for the mining community for obvious reasons.
Now, let’s take a closer look at what might be considered good news:
GMO ASIC for Bitcoin
According to official announcements, the upcoming ASIC will deliver 10 TH/s with a power consumption of only 500 W. It is supposed to be twice as efficient as Bitmain’s Antminer S9 that outputs 14 TH/s with a power draw of 1,372 W.
Also, since GMO is based on Japan, we might expect better customer service and higher product quality. The implementation of cutting edge technologies will also means that GMO’s ASICs will probably be the most expensive on the market. We will see.
Many experts agree that 7nm chips will bring mining hashsrate to a new level. However, personally I am still uncertain whether the change will be as big as some claim.
Let me explain why. Back in the day, there was a lot of hype around 14nm ASICs. For instance, the A4 Dominator by Innosilicon was expected to deliver 850 MHS per kilowatt of power when mining Scrypt. That was amazing, considering that its predecessors with bigger chips could deliver only 1/10 of that hashrate.
Now, when the A4 Dominator was finally released, it was able to output only 25% of the expected megahashes per second. While that was still an improvement compared to its 28nm predecessors, the leap was not really that big.
Why MHS/kilowatt and Not MHS/$?
As you could notice, I base my calculations on MHS/KWS, rather than MHS/$. I do understand that ROI depends a lot on the manufacturing price of the new chips and on the hardware that will support those.
Smaller chips are harder to manufacture than bigger ones, which is why I do expect 7nm-based ASICs to be rather expensive.
However, I speculate that 7nm chips are here to stay for a while; Mostly because we are about to hit the ceiling of Moore’s Law. The next step after 7nm is 5nm chips. Now, in order to manufacture 5nm chips and make them reasonably priced, our industry has to make some important technological advancements. I do not expect this to happen anytime in 2018 or 2019.
Therefore, I do believe that 7nm chips might become the “sweet spot” for ASICs for the next couple of years. This means that since we are talking about a sort of long-term investment, initial price does not matter as much as maintenance and electricity costs. Obviously, I am making this conclusion based on what we know so far.
The Bottom Line
I believe that 7nm chips will be a step forward, not a leap. At most, we can expect a 100% improvement in power efficiency, compared to 14nm chips. In reality, the number we might get might be only a 70%, 50% or even a 25%, as it happened with the Innosilicon A4 Dominator.
While we all agree that 25-100% better power efficiency is still a lot, it is not exactly a game changer of any sort.
This also means that even though the upcoming 7nm ASICs are definitely a difficulty bomb, those with 14nm chips won’t die off straight away. I suspect that the real difficulty bomb here will be GMO’s farm in Northern Europe.
Thank you for reading. As always, your comments, suggestions and questions are welcome.
Subscribe and stay tuned for further updates!