ASICMINER: Entering the Future of ASIC Mining by Inventing It

[Franktank]

Just did my second report here.

[woodles]

For guys that believe in Bitcoin, you have surprisingly little faith in the free market.

What?

The capability of the competition to rapidly assemble and bring products to market has been a much-maligned topic in this thread in the past few months. The "everyone else is incompetent" mindset flies in the face of free market theories, which suggests that competent competition will arrive in any sector where there is profit to be made.

Clearly the evidence of competent competition is here and the opinion has shifted. But your surprise remains - did you really think this opportunity for profit wasn't going to get exploited? That isn't a very free-market way of thinking.

It seems to me that even free market people couldn't predict just how fast the network would grow. When even staunch supporters are proven to be too conservative, I think that shows just how powerful a free market is. And yes, it is surprising when the market out performs the expectations of pretty much everyone.

[Jutarul]

Holy cat, I don't believe my own calculations!  But then I didn't believe them in July either, and the growth just keeps accelerating...

Calculate again (http://en.wikipedia.org/wiki/Sigmoid_function). The trick is to know where you are in the curve (hint: the asymptotic level depends on bitcoin production costs). Expect a conversion from exponential into linear growth after we cross 1-2 PH, given current price levels. (bitcoin price increases will move the asymptotic line and cause bitcoins to rotate back into mining investments, delaying the exponential to linear growth conversion).

[Vycid]

Holy cat, I don't believe my own calculations!  But then I didn't believe them in July either, and the growth just keeps accelerating...

Calculate again (http://en.wikipedia.org/wiki/Sigmoid_function). The trick is to know where you are in the curve (hint: the asymptotic level depends on bitcoin production costs). Expect a conversion from exponential into linear growth after we cross 1-2 PH, given current price levels. (bitcoin price increases will move the asymptotic line and cause bitcoins to rotate back into mining investments, delaying the exponential to linear growth conversion).

Undamped and underdamped systems do not follow sigmoidal growth trajectories.

There is a huge impetus to increase hashrate, because the yearly block rewards are worth something like 100M USD. In the short run, models of the increase in hash power may cause the rational observer to consider mining hardware a good investment. The problem is, if every rational actor in the market sees an opportunity for profit, then the growth will necessarily exceed projections; there is a time lag between when the hardware is ordered and when it comes online (very large for preorders). Greed is considered a rational market response.

Here is a simple mathematical representation of overshoot: http://en.wikipedia.org/wiki/File:Sine_integral.svg

The larger the growth rate, and the larger the time lag between the investment in growth and the return on that growth (relative to the growth rate), naturally the larger the overshoot.

This is the core of my bearish outlook on any mining security, as anyone who has been paying attention already knows.

The simplest example of overshoot in nature I can provide is the "rabbit example". Suppose you have rabbits in an ecosystem, and their population is stable over time, living in healthy balance with their food supply and their predators.

Suddenly, a plague wipes out their predators, which was previously the downward force on their population (here, the BTC/USD price exploded alongside the ASIC revolution). What happens, of course? The rabbits reproduce exponentially, vastly exceeding the carrying capacity of their environment - food (block rewards). Then, after all of these rabbits reach maturity (hardware delivery), starvation occurs (failure to meet ROI).

In industry, we deal with overshoot in situations where the setpoint (difficulty) is changing with cascaded PID controllers (http://en.wikipedia.org/wiki/PID_controller). That produces a damped scenario. The controls in the ASIC market (rational actors) are equivalent to a single PID controller, and we are headed for overshoot.

[Jutarul]

The larger the growth rate, and the larger the time lag between the investment in growth and the return on that growth (relative to the growth rate), naturally the larger the overshoot.
... and we are headed for overshoot.

There will likely be an overshoot - but only if price levels grow slower than hashrate. One scenario where an overshoot may be avoided is when the friction to scale is sufficiently large and/or jurisdictions interfere with the deployment or the sale of new hardware. Companies who are able to minimize this friction will be the ones who benefit the most from the network growth.

In industry, we deal with overshoot in situations where the setpoint (difficulty) is changing with cascaded PID controllers (http://en.wikipedia.org/wiki/PID_controller). That produces a damped scenario. The controls in the ASIC market (rational actors) are equivalent to a single PID controller, and we are headed for overshoot.

I disagree. I think there is solid proof out there that a great portion of the mining community has different assumptions about the market - ranging from expert to naive, from optimistic to pessimistic. This causes the mining investments to more or less experience an expectation value, i.e. the smart money is moving in or out first in the right direction.

Supplier companies like ASICMINER are facilitators in the market and just need to make sure that
a) they are a competitive option for getting invested in mining
b) survive times of depression in mining (hibernation ability), e.g. caused by the bitcoin price to collapse below production costs

I see a lot of companies out there striving for a). It will be interesting to see how many companies are prepared for b).

[Vycid]

The larger the growth rate, and the larger the time lag between the investment in growth and the return on that growth (relative to the growth rate), naturally the larger the overshoot.
... and we are headed for overshoot.

There will likely be an overshoot - but only if price levels grow slower than hashrate. One scenario where an overshoot may be avoided is when the friction to scale is sufficiently large and/or jurisdictions interfere with the deployment or the sale of new hardware. Companies who are able to minimize this friction will be the ones who benefit the most from the network growth.

Since BTC mining is an "embarrassingly parallel" computational problem (http://en.wikipedia.org/wiki/Embarrassingly_parallel), and since the hardware manufacturing requirements for miners pale in comparison to the size of the market for traditional electronic devices, I cannot anticipate much friction to scale. Perhaps single entities - like AM or ActiveMining or whatever - will be unable to scale because of logistical reasons, but the thousands of individuals buying Baby Jets or Bitfury boards will have no trouble.

Furthermore, there is a huge difference in electricity prices by region, something FC has not acted to take advantage of, but other actors have. This will encourage the overshoot scenario (in profit terms) for those individuals who have not exploited cheaper regions.

I agree that there is the possibility of regulatory interference, but predicting that is nigh impossible, and I seriously doubt a single region's regulation can stop this train.

In industry, we deal with overshoot in situations where the setpoint (difficulty) is changing with cascaded PID controllers (http://en.wikipedia.org/wiki/PID_controller). That produces a damped scenario. The controls in the ASIC market (rational actors) are equivalent to a single PID controller, and we are headed for overshoot.

I disagree. I think there is solid proof out there that a great portion of the mining community has different assumptions about the market - ranging from expert to naive, from optimistic to pessimistic. This causes the mining investments to more or less experience an expectation value, i.e. the smart money is moving in or out first in the right direction.

Supplier companies like ASICMINER are facilitators in the market and just need to make sure that
a) they are a competitive option for getting invested in mining
b) survive times of depression in mining (hibernation ability), e.g. caused by the bitcoin price to collapse below production costs

I see a lot of companies out there striving for a). It will be interesting to see how many companies are prepared for b).

This market is irrational and I do not believe it reaches reasonable equilibrium expectation values. Here is why:

In traditional equity markets, the "expert investors" control a much larger portion of the equity, because they have been previously successful and profited, whereas the novices tend to lose money and do not exert as much force.

However, BTC is an immature currency, and AM shares have exploded from very inexpensive to very valuable. Many of the people who bought in at AM's IPO (or early in the run-up) were unsophisticated. Now they control a large amount of equity and fail to price things in - especially because they are going to be in an extremely optimistic mindset after such a large return on their investments. We are witnessing a strange irrational market that is the product of a non-equilibrium state. It will eventually adjust as more sophisticated actors profit from the irrationality.

I do not believe AM will stop benefiting from its position as a market facilitator, but it is vastly overvalued.

[Vexual]

In traditional equity markets,

.. inexpensive to very valuable.

Are you a few dollars short of a bitcoin?

[hlynur]

Just did my second report here.

thanks for your efforts.
after quarrelling about my AM investment for last two weeks it felt good to read an objective summary which recalls the longterm perspective.

[bitfair]

Furthermore, there is a huge difference in electricity prices by region, something FC has not acted to take advantage of, but other actors have. This will encourage the overshoot scenario (in profit terms) for those individuals who have not exploited cheaper regions.

Just one comment in regards to the electricity prices: it won't be much of a factor for a little while still, it is more profitable to focus on pumping out hardware at the moment than worrying about electricity costs.

If you do the math, the electricity bill for AM hardware to produce 1 BTC is about $1.32. So "strategic positioning" with respect to power prices will come into play eventually, but it makes sense to focus on other areas first.

In other words, hashing with AM will have a positive marginal cost until the network difficulty is about 100x what it is today.

[Jutarul]

I do not believe AM will stop benefiting from its position as a market facilitator, but it is vastly overvalued.

That begs the question of HOW to value a company.

The way I see it is that in an informed and transparent market, valuation is easy and players have a clear understanding on where the fair market value lies. However, in an unpredictable or opaque market, valuation is difficult and players tend to try to "position" themselves, instead of maximizing profits. (e.g. look at the dynamics in the gold market right now).

It it my understanding that the ASICMINER equity derives a majority of its valuation from it's role as a strategic asset class, because of the difficulty in the market to predict the future.

[Pale Phoenix]

Vycid, on the last page, you mocked a gentlemen for his surprise over the velocity of hashrate increase, framing it as a lack of faith in the free market. Presumably because...

...if every rational actor in the market sees an opportunity for profit, then the growth will necessarily exceed projections; there is a time lag between when the hardware is ordered and when it comes online (very large for preorders). Greed is considered a rational market response.

Then, you assert that, in fact, the market is not behaving rationally (mostly because of nouveau riche AM shareholders, but we'll leave that bit alone).

We are witnessing a strange irrational market that is the product of a non-equilibrium state. It will eventually adjust as more sophisticated actors profit from the irrationality.

So, don't you think you owe the gentleman an apology for maligning his inability to accurately predict hashrate growth in this greedy, but unsophisticated, rational, but probably irrational market environment?

[Vycid]

Vycid, on the last page, you mocked a gentlemen for his surprise over the velocity of hashrate increase, framing it as a lack of faith in the free market. Presumably because...

...if every rational actor in the market sees an opportunity for profit, then the growth will necessarily exceed projections; there is a time lag between when the hardware is ordered and when it comes online (very large for preorders). Greed is considered a rational market response.

Then, you assert that, in fact, the market is not behaving rationally (mostly because of nouveau riche AM shareholders, but we'll leave that bit alone).

We are witnessing a strange irrational market that is the product of a non-equilibrium state. It will eventually adjust as more sophisticated actors profit from the irrationality.

So, don't you think you owe the gentleman an apology for maligning his inability to accurately predict hashrate growth in this greedy, but unsophisticated, rational, but probably irrational market environment?

None of these statements are incompatible. The market is irrational, despite the actors acting rationally (from their perspectives).

My point is that an expectation of anemic hashrate growth is a lack of belief that the free market can quickly and easily produce the competition that all free-market philosophies depend on.

I do not believe AM will stop benefiting from its position as a market facilitator, but it is vastly overvalued.

That begs the question of HOW to value a company.

The way I see it is that in an informed and transparent market, valuation is easy and players have a clear understanding on where the fair market value lies. However, in an unpredictable or opaque market, valuation is difficult and players tend to try to "position" themselves, instead of maximizing profits. (e.g. look at the dynamics in the gold market right now).

It it my understanding that the ASICMINER equity derives a majority of its valuation from it's role as a strategic asset class, because of the difficulty in the market to predict the future.

I admit that nobody can value a security with perfect accuracy, and certainly not myself.

My methodology essentially consists of: what is the current shareholder's equity, plus the annualized IRR expectation value - discount rate (the discount varies based on industry) over ten years forward, with appropriate adjustments for certain risk types. Basically, it's a discounted cash flow method. http://en.wikipedia.org/wiki/Discounted_cash_flow

Based on my observations that the hashrate growth is being constantly underestimated, it is not difficult to arrive at how AM has become so overvalued - forward projections of revenue are unreasonably optimistic. Furthermore, I see reductions in margin coming much sooner than most. Within a couple of years I suspect we will see margins of 20-30% for companies operating in the cheapest electricity regions.

In light of this, and because AM has little shareholder's equity, I believe it must return its market cap within the next few years to be fairly valued. In combination with more pessimistic growth models for the hashrate, I obviously do not see that happening.

Normally, the expectation value calculation (http://en.wikipedia.org/wiki/Expected_value) lends itself well to risky companies. For example, I own a significant position in Allied Nevada Mining (http://finance.yahoo.com/q?s=ANV) which I mentioned a while back. I believe it is undervalued on shareholder's equity alone, but the real key is the expectation value element. Here is a very simple (and therefore inaccurate) demonstration:

• We suppose there is a 20% chance that gold goes below ANV's cost per ounce at ~$1250, and ANV goes out of business;
  
• a 70% chance that gold stays around ~$1400/oz, and ANV's price level does not change because ANV is fairly valued at a P/E of ~10;
  
• and a 10% chance that gold returns to $2000/oz, and ANV quadruples in price based on the enormous profits per ounce.
  

If for ease of calculation we arbitrarily set the price level with a current P/E of 10 at $5/share (this is pretty close to the actual one), then our expectation value is (0.2 * 0 + 0.7 * 5 + 0.2 * 25) = $8.50/share. Therefore we consider ANV undervalued.

(A more sophisticated approach uses the premiums on gold options to calculate the odds. The P/E is easy to obtain for a given gold price based on published figures for costs per ounce and capacity.)

So why can't I apply this thinking to AM? Because the underlying "commodity" for AM (the Bitcoin) is a currency denominated in BTC. By contrast ANV is denominated in dollars, not in ounces of gold. The opportunity cost of buying AM is Bitcoin-denominated, so even if bitcoin goes to $10,000/BTC and AM maintains 10% of the hashrate, you don't make much profit over BTC.

I am making no commentary on the valuation of BTC here. I do not know. But AM is overvalued.

[Pale Phoenix]

None of these statements are incompatible. The market is irrational, despite the actors acting rationally (from their perspectives).

You seem to be talking in circles.

Are you now saying that these unsophisticated AM investors, who fail to price in risks (and thereby maximize utility), are acting rationally? Your earlier post strongly implied that they are not.

My point is that an expectation of anemic hashrate growth is a lack of belief that the free market can quickly and easily produce the competition that all free-market philosophies depend on.

Here you are being obtuse. Under no honest appraisal can his initial expectation/estimate be called "anemic."

[Vycid]

None of these statements are incompatible. The market is irrational, despite the actors acting rationally (from their perspectives).

You seem to be talking in circles.

Are you now saying that these unsophisticated AM investors who fail to price in risks (and thereby maximize utility) are acting rationally? Your earlier post strongly implied that they are not.

Those investors are acting rationally, yes. They don't know any better so it isn't irrational. I apologize if I made that unclear somehow.

It is the market that is not rational.

My point is that an expectation of anemic hashrate growth is a lack of belief that the free market can quickly and easily produce the competition that all free-market philosophies depend on.

Here you are being obtuse. Under no honest appraisal can his initial expectation/estimate be called "anemic."

Then you're calling my appraisal dishonest and there's no more discussion to be had. I have defended the reasoning behind my projections much more thoroughly than an exponential trend on an Excel spreadsheet.

By the way, the difficulty rate increase started showing double-exponential curvature about a month ago.

http://en.wikipedia.org/wiki/Double_exponential_function

[101111]

Under no honest appraisal can his initial expectation/estimate be called "anemic."

Correct.

[tkone]

vycid thank you for helping keep the price down

[romerun]

The "leaked" slides talk of the second large scale deployment (~1 PH/s)

I thought I see the official update saying 200Th in the new few months?

200TH is nothing. At least 1PH makes more sense in the near future.

[stefffe]

Will price ever rise above 3.5 again?

[JimiQ84]

The "leaked" slides talk of the second large scale deployment (~1 PH/s)

I thought I see the official update saying 200Th in the new few months?

200TH is nothing. At least 1PH makes more sense in the near future.

200TH is being used (deployed, franchised, sold) right now (first two batches were 62TH/s, i believe our current hashrate is around 45TH/s and AM sold around 20TH/s).

Fourth (and probably final) batch of 130nm technology will be ready later (probably start of october?) - 1000TH/s. Then around december/january first batch of 2nd gen chips. (55 or 65nm)

At least that's my view of things. I am not sure of anything 'cause of lack of updates from FC

[101111]

Will price ever rise above 3.5 again?

No one can predict the future, but personally it's worth at least 5.