Topic: BFL competitors days numbered? (Read 1524 times)

Who are you?

-D

#5 reserved.

Heheh

Also would be interesting if one of the ASIC manufacturers turned out to be a scam.

Thanks for that post. I think I almost agree with just about everything you said.

I have been watching the price rise over the last few days and I am hoping for a steady rise in price. I know price drives hash rate but I also believe if the price were to tank and the difficulty to skyrocket there will be a erosion of the community. That would be bad for BTC. It is nice to see some increases as we approach the block halving.

But I'm getting off topic now.

I don't know much about making chips but if some one has a design for an ASIC at 110nm for example, how hard is it to print the same design on a smaller chip with out having to redesign the whole thing? If it is simple then hardware manufacturers may have some new improved chips sooner than later in order to stay competitive. Of course if difficulty stays at~250Th/s all year the power use differences  probably wont be an issue, but by not being an issue it becomes and issue!?!

I prefer to just shoot down other peoples ideas because then it's harder to throw back in my face :p

But seriously, it is quite difficult to guess with change in tech and whatnot, so I will say that as a tentative estimate, I'm foreseeing a fast run up to 250TH, by end of Jan, then a steep decline in the rate of new power coming online. The next 250TH I'm guessing will come online over the course of 10-12 months depending on a few factors.

1) Price (up or down?); up boosts the curve, especially if we see an increase RIGHT after 25 halving. Even $15/coin would shoot up the rate at which ASICs are acquired

2) How fast ASIC 2.0 is released and/or if there are price drops on ASIC 1.0 in this time frame.

3) How quickly supply can meet demand divided amongst the suppliers

4) If suppy > demand by a significant amount, if a supplier decides that mining excess unsold hardware makes sense.

5) Reserved

BUT, assuming reasonable price level close to current ones, ASIC 1.0 being the only game in town for 2013, a reasonable rollout of supply, and no malicious suppliers, my best bet would that we hit 500TH around the end of 2013, if not later. For why, as I said, the initial burst of pre-orders will be the main thrust into ASICs, which will explode difficulty, late comers will be eyeballing price/difficulty/reliabilty, etc. and take a slower tack to buying. 500TH is roughly 8333 singles or ~10.25Million USD, close to the current amount of GPU hardware, which I think will take current users time to liquidate their GPU rigs, and decide if it's worth going into ASICs or buying coins.

@bcpokey

Good follow up to my statement. Those are very good points.

Absolutely the speed of reinvestment will diminish assuming BTC revenue is reinvested.
But there maybe new investment money coming online.
If you assume the same last six months of difficulty increase percentage wise ( 0.2% to 0.6% per day) I still think 6-12 months post pre delivery shipment completion(8-12weeks to get those things out the door) we will see 500THs. So say 7-15months after first delivery. That's where my numbers come from.

If you would wager a guess at when we would hit 500THs what would it be?

I suspect the difference in efficiency between BFL's 65nm process and their nearest competitor's process(90nm) will turn out to be not all that great. Down the road the relatively small difference in efficiency between products may determine who's gear gets unplugged and who's gear keeps hashing, but at this point getting products in the hands of end users is of much greater import IMO.

Well you cut out the most important part of that sentence, but ok.

As to being halfway there, that is anyone's guess. We've gotten figures of 20,000 7.5GH/Sec ASICs from BFL (less actually since jalapenos are 4.5GH/sec), 800 54GH/sec Pre-orders for Tom, and 300 66 GH/sec units  for Avalon (in Jan), which is still  < 250TH (approx 200TH). This includes all demand built up over the last 6 months, with the excitement of hashing 50Coin blocks at 3.3M difficulty (again less when this started).

To think that you will see the same demand at 33M difficulty, 25coin blocks and whatever price bitcoin will be as the first 250TH ($5Million) sell sell sell to repay their investments is... somewhat unreasonable.

Will we hit 500TH by the end of 2013? It's certainly possible. Will we hit it significantly before the end of 2013? It's not very likely.
Why? Well just think of the demand vs. difficulty curves we have seen thusfar. Then extrapolate to the upcoming situation, @25TH we have 3.3M difficulty and a $1299 60GH/sec machine takes 1 week to pay off, @ 250TH we have 33M difficulty, it takes 18weeks (4.5 months) to pay off. @ 350TH we have 46M difficulty, 25weeks (6.25 months) to pay off, @450TH 59M diff 32 weeks (8 months). As difficulty increases, ROI decreases and people become increasing skittish, slowing down demand, and that is only difficulty increasing, no other factors.

After the ASIC pre-orders get shipped we should be more than halfway there.
Hitting 500TH could take 6-12 more months. IMHO

If I was selling ASICs I would be looking into making my next batch match or better the current best ASICs.

It's all theory until these things hit the street and we know what the actual performance numbers are.

Yes and no. Process size matters, and who can deliver what volume in what time frame matters. Pricing also matters, speed matters, efficiency matters, it all matters.

Are the competitors doomed simply because they are not on 65nm? Not really. Will they need to work to remain competitive? Of course. If there are 2 products, with equal prices, and equal hashing power, I will choose the one with the lower fixed cost to run, no question. It's foolish to think that no one has any plans for changes in design and pricing however and everything will remain stagnant.

500TH in a year is certainly possible, however do keep in mind the following:

500TH = 500,000 GH. Prices are currently on the order of $20 / GH, so that's roughly $10,000,000 of ASICs at current pricing/performance. At current pricing after block halving the total amount of coins generated per month will be worth roughly $1.25Million. That means at 500TH, it will be 8+ months selling 100% of the bitcoins generated, to break even (if somehow the price could hold that up). So this is for a product with as you said a potential 10month ROI (or more), and no other possible use except bitcoin. Not as sexy as the "GPUs that pay themselves off". I certainly expect to see 500TH, but I imagine it will be a slower run-up than many people think, As shown by Organofcorti, bitcoin miners are typically fairly risk-averse, despite appearances to the contrary.

Process size doesn't mean shit, the only thing that matters is who is actually delivering units the fastest.

Who thinks BFL's competitors are due to wither away if they don't quickly go with a 65 or 45nm process?

500 TH+ competitors will have already started to experience sharply declining sales due to 10+ mo ROI.

Difficulty could reach these levels in a year. 


Is investing with a BFL competitor more risky without having a roadmap/plan/stated intentions from the competitor?  Will vendor paranoia persist leaving the community/customers in the dark regarding future technology plans?