Should you buy an Alpha Technology ASIC for Litecoin mining?

alpha technology viper

Short answer, probably not unless you point it to another Scrypt-based token like dogecoin.  I’ll show you the math and hypothetical situations below, but quick story.

This past spring I helped build a couple simple Sapphire 7950-based mining rigs in Shanghai for a few friends.  At the time the new Vapor-X cards cost around $300 each and could be tweaked to run at around 600 kh/s (in line with these hardware expectations). [Note: This was also just before Cryptobadger, who has a great series of how-to guides, highly recommended if you’re interested in doing-it-yourself.]

Fast forward to early December, virtually all physical stores in any big city throughout China were sold out of the following Radeon models: 7950, 7970, R280, R280X.  My friends spent hours calling up and chatting to online shops from Taobao and Tmall to try and locate any supplier with product.  But all were backlogged for the next couple of weeks.  Why?  What had happened is that the price of litecoin tokens had popped on exchanges (namely BTCe and OKCoin) by a factor of 10x in less than 3 weeks ($2 to $20).  Yet the difficulty rating was still (temporarily) at around a mere 1,000 meaning that the return-on-investment for even a small rig composed of 7950s was relatively quick.  However that capital expenditure / token profit information traveled rather quickly, ending up on a variety of domestic evening newscasts.  And thus, there was a mad dash to get these GPUs.  As a consequence, I spent one early December weekend during this time combing the PC malls in Changning district trying to find any owner who could supply a couple dozen 7950s all in an effort to help some of my friends build a litecoin mining farm.  Yet each laoban explained (with a smile) that some large buyer had bought the remaining stock en masse — store by store.

We didn’t build that farm and in the short-run that may be okay.  While some miners simply look at the short-term seigniorage of flipping a few blocks, long-term mining operations probably will hold onto whatever tokens with a view that the tokens will appreciate by an order of magnitude.  Thus, while the collective hashrate for the capital expenditures for the 50 GPU cards my friends wanted to buy would have certainly been considered a profitable investment at the 1,000 difficulty rating, the longer term capital risks are substantially higher because the spread in crypto mining, like other commodity gathering, tends towards equilibrium (e.g., the cost of mining eventually equals the financial returns).  That is to say, the extra units of profitability (or unprofitablity) sends a signal to market participants to either continue one particular activity (like mining) or to shut down mining operations altogether.  You see this frequently in other capital intensive resource gathering spaces such as petroleum extraction as well as precious metals (e.g., if the revenue / barrel increases, other competitors will invest in new extraction techniques and/or open fields for development yet if the revenue / barrel decreases, competitors may shut off all production in a particular field).1

Again, crypto mining involves a scarce resource (block discovery in which the coin or token are part of) and in order to mine you need capital investment, in the form of a GPU.  Thus the same issues of supply, demand, price discovery and profitability now exist.2

ASIC mining

In 2011, Scrypt was adopted as the proof-of-work mechanism used by the Litecoin protocol.  It was purposefully chosen by Charlie Lee due the understanding that it was more resistant (not necessarily immune) to GPU and ASIC mining than the SHA256d proof-of-work used by the Bitcoin protocol.  And after two years into this altcoin experiment this assumption seems to have empirically played itself out as there have been no known litecoin ASICs.

Yet now that litecoin tokens are trading at ~$25+ each, the return-on-investment for physical design engineers (the people who actually design integrated circuits) to develop an ASIC has become within the realm of profitability for even something designed to be resistant such as litecoin.  That is to say, some group of investors believes that the capital costs of hiring a team to design, fab and sell an ASIC to litecoin miners is profitable (otherwise they wouldn’t do it).

At the end of last month, Alpha Technology in the UK announced that they were at the finishing stages of design for two products, a 5 MH/s and 25 MH/s ASIC for litecoin (remember, the ASIC you buy and use for bitcoin mining cannot be used for litecoin mining because it is designed specifically to work on one particular proof-of-work).  The 5 MH/s is expected to need 100W and the 25 MH/s version is targeted at 600W.

According to the announcement, the cost for the Viper 5 MH/s is £1350 ($2209) and the 25 MH/s is £5450 ($8918).  Not including electricity, taxes, shipping and pool fees, the top Viper at 25 MH/s works out to be around 2.8 khash per dollar.  This compares with a ~$300 for a new Sapphire Vapor-X 7950 @ 600 kh/s which is around 2 khash per dollar. [Note: most 7950s are no longer assembled as such and are now rebadged as R280, the big exceptions are HIS from Taiwan.  Cryptobadger has a run-down of the best cards available.]

Risks and variable factors

Despite looking legitimate through press releases and 3D renderings, the product might never come out.  Or when it does get released (probably late), the real numbers might be way off the estimates.

For example, last March a good friend of mine paid 50 BTC for 100 GHash/s ASIC from Butterfly Labs.  He received it more than 6 months later at the end of November.  If instead he had held that 50 BTC he would have been able to sell for $40,000 – 50,000 on many exchanges.  Instead, if you plug that 100 GH/s rate into a mining calculator, he will not even be able to mine 1 BTC for the next year at the current difficulty rating let alone ever be able to mine 50 BTC it cost to buy them.

So here is the likely scenario with Alpha Technology.  They will not ship for at least another two months.   Why?  The inner engineer in me asks: Has it passed verification process?3 Has it been taped out? What about maskmaking?  Those they do ship to are those who ordered first and there is a wait list (here’s a widely inaccurate waitlist).  Each minute you wait is another minute someone else will be ahead of you.

Now, assuming you were able to get the Viper 25 MH/s today, looking at the mining calculator to see how many litecoin tokens you would receive at the current difficulty (3366.7), the number is: 2408 LTC / year.4 Assuming the network hashrate does not collapse, that is the most optimal scenario you have this year.  And the highest price an LTC has gone for on an exchange so far was $50-$60 back in late November/ early December last year.  If this price level is ever reached again and the difficult rating never changed, then you would stand to make ~$125,000 which would make your $8,918 investment very fruitful!

But alas, this is not how it works.  Again, assuming the product is made and even shipped on time, the difficulty rating will continue to increase proportional to the additional hashrate.  So as more and more of these Viper’s (and/or other GPUs, FPGAs and ASICs) are added to the network, the higher the difficulty rating is adjusted to.

The next estimated difficulty rating is expected to be 3700 which knocks off 20 LTC more a month, dropping you down to around 2150 LTC / year.  But this is not the entire story.  You still need to factor in electricity costs, the transportation and shipping fees (unless you live next to the manufacturing and distribution center) as well as the pool fees.  Some pool fees, like at Coinotron which I used are 5% (it is this high in part because of the maintenance and admin costs needed to protect against DDOS attacks). 5

In all likelihood, unless you are the very first person on the list when the product ships, you would be better off either building an off-the-shelf GPU mining solution or buying LTC on an exchange and either hold for speculation and/or arbitrage.

Why?

Even if your electricity was free, you lived in the UK/India/China where they are manufactured and were the first person on the pre-order list, your first mover advantage would be quickly eroded by other Viper owners.  To give you an idea why, look at the current Litecoin Mining Pool hashrate.  If 200 other consumers bough 1 of each Viper, they would collectively add 6,000 MH/s which would place these ASICs alone as the 6th largest pool, increasing the hashrate proportionally (this is actually conservative because difficulty typically trails hashrate).

If litecoin difficulty doubled to 6,000 at current price levels ($27) and a 25 MH/s hashrate you would generate 125.7 LTC / month and earn enough monthly ($3,423) to pay for the machine in 3 months.

But instead, let’s assume for the moment that other market participants have access to similar mining calculators and how-to Cryptobadger guides.  And that over the next 6-9 months the difficulty rating jumps to 30,000 (9x the level today).  Impossible you say?  Last April when I got the initial litecoin rigs up and running, the difficulty rating was 300.  So in less than 9 months the rating has gone up 11x.

If it reaches 30,000 you would only generate 25.15 LTC / month which at ~$27 / token would generate $684 / month.  That means you would likely only generate enough litecoins (at current prices) to cover the costs for the Viper in the first year (ignoring all other pool fees, electricity costs, taxes, duties, etc.).

Sure the tokens could appreciate and increase in value, but as we continue to see, if price levels increase so too would competitor hashrate as others see a similar seigniorage opportunity.

That said, if these numbers are real, this Viper ASIC is only 40% better than GPUs in terms of hash/dollar. Of course this is just the first generation and other companies might be able to make more efficient chips. But this definitely is a positive sign that Scrypt hashing might be able to keep ASICs from totally dominating mining like it does with sha256d.

Money well spent?

One thing to constantly remind yourself is that like any investment, you should only spend money you can lose.  That is to say, as bullish as you may be on any particular asset class (including cryptocurrencies) there is always a statistical possibility that its liquid price could sink below whatever level you have bought at (e.g., underwater).  Perhaps even falling to zero.

If history is any guide (and perhaps it is not) looking back at the California Gold Rush (the 49ers) the firms who ended up financially in the black were merchants and service companies such as Samuel Brannan, Philip Armour, John Studebaker, Levi Strauss and Wells Fargo.6 Those who also made and sold mining equipment (picks, axes, shovels, sluices) had mixed results.  Yet the group of people who typically fared the worst financially were the miners themselves as they were nearly all exposed to various types of risks (upfront capital costs, land title lawsuits, inclement weather, sickness, etc.) and as a consequence, most ended up bankrupt.

Does this mean you should not purchasing crypto mining equipment?  No, but you are probably more exposed to risks with fewer potential upsides than downsides.  Your capital is tied up into a depreciating asset, a machine which unlike a GPU that can be resold, has a singular use that may or may not be delivered on time with unknown hashrate performance deltas.  Or you could be thinking, just like the first people who managed to get an Avalon batch last winter or a KnC miner when they ship new updates throughout the year (like the upcoming Neptune), perhaps you might be lucky enough to get a Viper that lives up to its paper reputation.7  But the odds are you won’t, especially if you are reading this and have not pre-ordered it.

One other option for HNWI is that you could invest in an IC design company such as Alchip which does the physical design for KnC.8 Or create your own engineering team to build ASIC machines for internal use only and sell public shares just like ASICMiner in Guangdong did last year.

Lastly, for entrepreneurs there are other areas to focus on beyond the token such as the financial instruments and applications discussed by Mike Hearn in 2012 that utilize the Bitcoin or Litecoin protocol (e.g.,  secure time-stampingproving ownership of tangible propertydecentralized DNS and new ways to sign contracts).9

Exhibit A:

Below is a very rudimentary table that utilizes this Litecoin difficulty calculator.10  The calculator is nowhere near advanced as the dynamic dashboard over at Genesis Block (which is BTC only).  In fact, this chart below does not include in its calculations the long tail of the difficulty curve.  It is an end-to-end snapshot (what it is today versus what it will be 6 months from now).  But that is unneeded as this shows you that in every instance, building GPU-based systems instead of buying the ASIC is probably more profitable in the first 6 months.  In some cases, merely buying LTC and holding is actually more profitable.

I should point out that for this activity I negated electrical costs which obviously are non-negligible especially for a large GPU farm.  Obviously an ASIC will come out per watt more efficient, so feel free to factor in whatever electrical costs you may pay on a monthly basis in your region.  I also assumed the consumer would be able to buy 32 new or used Vapor-X 7950s for $200 each and then simply build a barebone system using milkcrates as per Cryptobadger (Friendly reminder: anything that is not the GPU is not generating tokens and is therefore a money sink — you do not need fancy cases or i7 CPUs).  It is probably very difficult to locate 32 of these GPUs, let alone 42 for Scenario 4.  But you could likely find batches of used versions on eBay, Craigslist and other etailers.  They do not even need to be the Sapphire brand; see this chart for more comparisons.

The biggest difficulties for a massive GPU farm like that however will be logistics, cooling and storage.  You would need to have access to reliable power and internet sources.  You would also need to keep an eye on each of the GPUs throughout the day to make sure they are performing up to snuff (I actually used LogMeIn but I think Cryptobadger’s method is much more efficient).

Still, I would speculate that in all likelihood, the Viper is unlikely to be delivered to your door within the next 90 days.  If that is the case your two profitable options (based on this chart) are to buy and hold LTC and/or build a rig or two (depending on if you can get them used and what your electrical costs are).

One last note, I do not predict that LTC price levels will reach the numbers listed in this chart this year (if ever).  These are for illustrative purposes only.  In contrast, if price levels do continue to increase I would expect the difficulty rating to increase in a corresponding manner and that the lopsided disconnect in the last column would never germinate.  Baseline difficulty is 3300 and starting LTC price is $27.

Investment Option ETA Setup cost in USD and LTC Difficulty increases same as 6 mo. historical avg and LTC stays at current price Difficulty increases same as 6 mo. historical avg and LTC increases at 6 mo. historical avg Difficulty increases less than 6 mo. historical avg and LTC increases at 6 mo. historical avg Difficulty increases more than 6 mo. historical avg and LTC increases at 6 mo. historical avg Difficulty increases same as 6 mo. historical avg and LTC increases at less than 6 mo. historical avg Difficulty increases same as 6 mo. historical avg and LTC increases at more than 6 mo. historical avg
Difficulty x4 @ 13200, LTC @ $27 on July 6th 2014 Difficulty x4 @ 13200, LTC x9 @ $243 on July 6th 2014 Difficulty x2 @6600, LTC x9 @ $243 on July 6th 2014 Difficulty x8 @ 26400, LTC x9 @ $243 on July 6th 2014 Difficulty x4 @13200, LTC x4.5 @ $121.5 on July 6th 2014 Difficulty x4 @13200, LTC x13.5 @ $364.5 on July 6th 2014
Scenario 1 Preorder 25MH/s Viper today  1/6/2014 Delivered in April $8900 or 329.6 LTC (no fees) Begin April 6th and after 3 months hashing = $4,890 or 171.45 LTC $41,662 or 171.45 LTC $83,324 or 342.9 LTC $20,511.63 or 84.81 LTC $20,831.17 or  171.45 LTC $62,493.52 or 171.45 LTC
Scenario 2 Preorder 25MH/s Viper today  1/6/2014 Delivered in July $8900 or 329.6 LTC (no fees) 0 0 0 0 0 0
Scenario 3 Buy mining rig with OTS GPUs worth $8900 today Start mining in 1-2 weeks $8900 or 329.6 LTC (no fees) 32 Vapor 7950s (Milkcrates) after 6 months hashing =  $7512 or 263.34 LTC $63,991.62 or 263.34 LTC $127,983.24 or 526.68 LTC $31,995.81 or 131.67 LTC  $31,995.81 or 263.34 LTC $95,987.43 or 263.34 LTC
Scenario 4 Buy mining rig with OTS GPUs @ 25 MH/s today Start mining in 1-2  weeks $8900 or 329.6 LTC (no fees) 42 Vapor 7950s (Milkcrates) after 6 months hashing =  $9780 or  342.9 LTC $83,324.7 or 342.9 LTC $166,649.4 or 685.8 LTC $41,662.35 or 171.45 LTC $41,662.35 or 342.9 LTC $124,987.05 or 342.9 LTC
Scenario 5 Invest $8900 (cost of Viper) into LTC today Buy and hold 329.6 LTC $8,900 $80,092.8 or 329.6 LTC $80,092.8 or 329.6 LTC $80,092.8 or 329.6 LTC $40,0046.4 or 329.6 LTC $120,139.2 or 329.6 LTC

See also: Why it is impossible to profitably mine bitcoin (BTC) with GPUs — but still quite profitable to mine litecoins (LTC)

  1. The Mountain Pass rare earth mineral mine is an example of a mine that was recently restarted due to these economic conditions of supply, demand and profitability. []
  2. While you can read through the developmental history of both Bitcoin (the network) and bitcoin (the token), the original miners and early adopters from 2009 and 2010 mined for a variety of reasons and motivations.  Perhaps accumulation and appreciation were among those, yet the first “real” exchange didn’t occur until May 21st, 2010 — a $25 pizza was exchanged for 10,000 BTC.  See This Pizza Cost $750,000 from Motherboard. []
  3. See Automate and Control the Functional-Verification Process from Chip Design, Interview: Adnan Hamid Addresses Trends In Chip Verification from Electronic Design and Chip verification made easy by Laurent Fournier []
  4. Difficulty rating for Bitcoin adjusts every 2016 clocks or roughly every 2 weeks []
  5. Not to mention there is always the potential downtime in the even there is a net outage or electrical problem where the machine is located.  Unless you put it in a colocation, your machine’s uptime will be directly effected by where you live. []
  6. Contrary to popular myth Sears & Roebucks did not exist at this time and in fact was founded much later in its modern form in 1893.  It was Richard Sears’ father, James who went to California during the gold rush and failed to become rich. []
  7. See Engineering the Bitcoin Gold Rush: An Interview with Yifu Guo, Creator of the First Purpose-Built Miner from Motherboard and That Swedish Bitcoin Mining Company Has Sold $28 Million-Worth Of Its New Mining Hardware from Business Insider []
  8. See Alchip, KnCMiner team up for Bitcoin mining machine with 28nm ASIC from DigiTimes []
  9. There may also be other opportunities for a startup to focus on: hedging exposure, quantifying and qualifying risks and perhaps even insuring or re-insuring []
  10. I recommend reading through this Litecoin community thread which includes a very detailed chart based on estimated hashrates and difficulties. []

5 thoughts on “Should you buy an Alpha Technology ASIC for Litecoin mining?

  1. Spot on with this write-up, I truly believe that this site needs far more
    attention. I’ll probably be back again to
    read more, thanks for the info!

  2. Great article! I am surprised that no one in the U.S. is attempting this and we have to rely on UK and India. India’s VAT (tax) is currently about 12.5% !! What??!! Alpha says you gotta pay that to recieve product on top of other taxes/fees? It’s late January of 2014 and LTC difficulty just dropped to 3130, which means profitability is reportably going “up” somehow even though the price of LTC has dropped to $20 per coin. There’s another expected dip coming on the difficulty soon, which will further drive down the price of LTC at least until alpha comes out with their ASICs with script for LTC.

  3. All of the above assumes some kind of predictable difficulty increase that completely ignores the fact that the initial push into Litecoin and the dramatic jumps in difficulty were a result of the huge media coverage regarding it’s price jump and the shortage of ATI video cards.

    As there are no ASIC machines available, the media coverage for Litecoin has shifted somewhat to other altcoins such as Dogecoin and Vertcoin, and the initial rush of people thinking they could capitalize on the price jump by building ATI rigs posted by media outlets has slowed, it should be apparent that those kind of linear or predictable difficulty adjustments can’t be applied. Look at the past month of Litecoin difficulty as well as the price drop.

    As the initial rush has passed, it’s increasingly unlikely that we’ll see that same kind of artificially inflated network growth prior to ASICs being released for Scrypt coins, especially since the entire network of people currently mining would have to grow tenfold in the next six months to hit the numbers you’re talking about. Given the current supply of ATI cards, the difficulty plateaus and drops, and the price drops for Litecoin, people that actually get a hold of one of the first Scrypt ASIC machines are going to be shaking their head at this article all the way to the bank.

Leave a Reply

Your email address will not be published. Required fields are marked *