Update

Clarification of Unsold Shares
Now ASICMINER controls 163,962/400,000 of the whole identity, while Bitfountain controls 236,038/400,000. The former shares will not be diluted, while the latter will. Whenever we need to sell shares to raise funds or giving potential core employees shares/options, only our (Bitfountain's) shares will be reduced.

We consider it to be very fair to investors, because while either letting the investors break even first or protecting the investors from dilution are seen everywhere, having both at the same time is not that common, especially when the portion at the investors' side is already large.

Device Selling
After careful consideration, we decided to address it as a very important decision ahead. There are two factors that serve as the motivation of selling our devices earlier than what we planned before:

1. Vacuum of Deployment. We are now experiencing one. The unconventionally high power density makes our devices hard to be deployed to professional data centers, which also have very high rent charges as their extra downside. So we had and will have to build them of our own. It's a difficult task in a sense that besides the long cycle, unexpected incidents will create a period that we have a lot of devices lying there waiting to generate hashes.

2. Quantity in the Future. The price negotiation of the full mask is almost finished, so if there are no surprises, we will have virtually no third-party limitations on how many chips we could produce in this year. If the exchange rate of BTC against USD further shoots up in the next few months, we could even produce hundreds of wafers each month. Deploying all of them in time is out of our reach unless we expand significantly, which might be pre-mature and inflexible.

The most significant factor in selling devices is pricing. Devices are extremely illiquid because the trading and transportation are great ways of wasting the precious hash-time. Therefore no one knows how they should be priced yet. We would like to do the price discovery with auction by small quantity. The total number used for discovering the price will be very small compared to the whole batch, and the slow deployment rate before the construction of the whole farm infrastructure of the rest and arriving devices provides another motivation.

Other Info
We met a temperal black out this morning. It's caused by the capacity expansion (installing cables and switches). The PCB of the main board and the power module have been finished. The assembling will start as soon as the chips are packaged and passed the automatic tester. Our use of ~0.5MW of power in the new place is also approved. The network bottleneck is not in the internet connection side, but in our local Ethernet. It could be solved by optimizing the firmware and the switch structure, but a simple hot-fix is to add more internet connection accounts and routers, which is what we are doing now. The slow pace of the infrastructure building is also a hard limit, and it forced us to scatter a small portion of our devices to temperal places. They should be put back as soon as the permanent place is OK.

^This.  AM should have planned out exactly which data centers to deploy at ahead of time.

The data centers should be selected on the basis of cheap power and proximity to their pools' hosts.

Scrambling around building their own is a big loss of time and focus from their core competency of designing, producing, and running ASICs.

wow, ASICMINER is rallying hard on all the PTs!  0.898 hit on bitfunder