Hi, I've been trying to estimate future AM market-share based on current evolution, and I came up with some interesting numbers; I'm curious what everyone else thinks about this.
First, the assumptions:
- Difficulty is rising approx. 50% per month
- AM is working on 2nd gen chips which will be available in December
- AM 2nd gen chips will be built on 65/55nm, compared to 130nm for current gen
I. EfficiencyGenerally, every die shrink allows for a roughly 40% increase in the number of transistors; coupled with a more optimized design and potential increases in clock-speed, all while keeping the same overall architecture - we're probably looking at an 80% performance increase for the same power use. Given the two steps of shrinking (from 130nm to 90nm to 65nm) we would achieve about a 3.2x increase in performance.
Existing chips perform at 332MH/s, so this hypothetical 2nd gen. design would then hash at ~1.1GH/s. With overclocking that becomes ~400MH/s, or ~1.3GH/s for the 2nd gen. chip.
That would give us the following:
- USB Erupters v2: 1.1 - 1.3 GH/s
- Small Blades v2: 17 - 20 GH/s
- Normal Blades v2: 35 - 40 GH/s
Of course, this all assumes a realistic evolution of the current AM chip design. At this very moment, significantly better designs exist on the market -- the BFL chips actually perform at a top speed of 3.75 GH/s and are also built on the 65nm process. It is possible that AM does a complete overhaul of their chip and approaches (or even exceeds) those speeds; however there is no way to predict that.
II. DifficultyFar from being science by any definition, predicting difficulty is more like a black art. That said, it's safer to err on the side of caution here and assume a 50% monthly increase (and even that might be conservative). 4 months from now, in December, that would give us:
- Difficulty: 5x what it is now or ~190Mil.
Given I + II, the resulting price for a 2nd generation AM blade comes out between
4.5 BTC and
8 BTC - using e.g. the TGB calculator, a 40GH/s device would yield a maximum of ~4.5BTC if the difficulty keeps rising with the same rate, and even at 40% diff. rise it would not yield more than ~8BTC.
If AM can maintain a profit given the NRE and production costs for the 2nd gen. chips, they could still keep up selling them at the above prices. But what about self-mining? Given a difficulty increase of 5x and performance gains of just 3.2x, AM would need to increase their deployment area by 50-60% and replace all boards with the new chips just to keep their current share of network power; whether that will be possible or not remains to be seen.
Of course there are many unknowns that could influence the outcome for the better or worse (difficulty rising slower or faster, BTC prices increasing or decreasing, 2nd gen. chips having 10x performance, production/deployment delays etc) but overall I'd say the above analysis covers the middle ground, and if it comes to pass we'll be looking at similar profit levels in December compared to August. If anything happens and AM misses some of the targets, we could be looking at a 60% decrease in mining revenue as well as sales income.
As for impact on share prices, after a rise in the coming weeks due to increases in blade sales, they will likely level off and see a decline towards the end of the year, until production-level 2nd gen. chips start coming up en masse. What do you think?
DISCLAIMER: I currently own shares in AM as well as other mining ventures.