As you might have read, we are of the opinion that the two factors that need to be modeled when projecting mining returns are growth of the global hash rate, and depending on the audience, the price of a cryptocurrency with respect to base currency. As we’ve discussed, one company openly acknowledges to controlling at least 70% of the market for bitcoin mining rigs, and from our observations, a higher percentage in some alt coins. Therefore, there are at least some people who have access to a much better approximation of the global hash rate – you need only look at Bitmain’s production schedule.

But while Bitmain professes to not disclose such information, their fabrication node – Taiwan Semiconductor Manufacturing or TSCM – apparently openly shares this information with research analysts. A report was issued today from one of the major investment banks[1], having been distributed to their ultra-high net worth clients positing the virtues of energy-related investments due to the explosive growth in the bitcoin network infrastructure. Unsurprising that the opportunity uncovered by the assembling of their unfair advantage (“research”) was overlooked, in favor of a more sensationalist though less useful story.


To make these claims, they observed the production schedule reserved by Bitmain chips at TSCM. They claim that Q3/Q4 production has been 10,000 wafers per month – with one wafer supporting 27-30 units at 189 chips each – ramping up to 15,000-20,000 wafers per month in Q1. We had validated the wafer to unit conversion whilst conducting our own independent research this past summer (and also learned that wafers are not cheap), but have consistently underestimated the magnitude of Bitmain’s batches.

These numbers are reasonable if you assume that some subset of those units are used in producing their other devices. Some back-of-the-envelope math shows that at 16 exahash – if you assume everyone is using 14 TH/s Antminer S9s – then that’s around 1.1 million units[2]. So this would assume production at a high end of 270,000-300,000 units in each month in Q3/Q4, which is a little too high. However, if you assume that those assumptions also include the ~100,000 LTC rigs and ~100,000 Dash rigs produced, then perhaps the numbers do line up. If the hash rate is increasing by ~10% per month, this would imply roughly 100,000 new units put online in the Month of December – which we did observe.

So let’s assume this banking research department is right, and Bitmain is planning on releasing between 400,000 – 600,000 new units per month in Q1. Does this spell the disastrous end of Bitcoin Mining? Is the bitcoin network going to resemble the Dash network?


When we include the new assumptions in our model, we see that things aren’t so bleak. Bitcoin mining has been really profitable, as in >$800 per month in December profitable, before costs. Disregarding your costs to produce a month’s worth of mining proceeds for a minute, there’s only two ways to reduce the current ROI – a massive drop in Bitcoin price, or a substantial increase in hash power. If we hold the price fixed, then the hash power would have to increase by 100% to cut profitability down to $400 per month. As we just calculated, that would be the immediate introduction of 1.1 million 14 TH/s S9s.

So let’s adjust the model updating for the new information:

At 500,000 units per month, the mining reward per miner drops to ~.00025 BTC (.25 mBTC) per day. At $15,000 a bitcoin, this is $3.75. Not great, but even at these extreme production rates, still profitable given no change in BTC price. To be honest, this took us a bit by surprise. So, in light of this new information, how would we answer the question we inevitably get 4-5x a day: “Should I buy a bitcoin miner?”

Looking at a 1 year horizon, an S9 can be expected to produce just shy of .2 bitcoin. That’s $3k at today’s rates. Again, this assumes no change in the price of Bitcoin which may not be a realistic expectation if there are 500,000 new units on the market each month. Sure, much of that will be centralized among some of the largest players and some scary-large percentage may be hosted directly by Bitmain themselves, but new participants leads to more demand for a currency that by definition is fixed in supply. In all likelihood, .2 bitcoin may represent a large amount of fiat currency – just look at our price projection:

OK so that last screenshot was partly in jest – the exponentially weighted moving average is overweighting the ~$1500 bitcoin went up in the last 24 hours and projecting that volatility forward. But one can dream…

In summary, even if this banking research reports proves true, there is still vast amounts of money to be made in bitcoin mining in 2018.