Mmm.. I don't know, I think they are building out things as fast as they can, and nothing is sitting idle, other than inventory allocated for delivery to hardware buyers. Bear in mind to keep up with difficulty gains in the network, they need to be racking and stacking 60+ blockerupters per day. They need to do this 7 days a week, while also building out all the space, networking switches, power, and cooling to support over 3100 blockerupters they currently have deployed. (They have proven their ability to deploy at a peak rate of ~60 blockeruptor boards a day.)
Remember these numbers will only go as they try to maintain their share of an ever growing hashrate. Someone mentioned the law of large numbers earlier. I think an explanation might be in order.
"The first is the law of large numbers. As a company gets bigger, each percentage of incremental revenue suddenly represents a fundamentally larger number. As the base grows, the amount of new business needed to make a material difference in earnings also rises, increasing the pressure on sales to find new markets, new categories, and new geographies. In other words, the larger a company becomes, the more the entire engine has to work harder."
IE: It's going to become exceedingly harder and harder for AM to maintain their share of the network, and it seems at some point they might not be able to keep up. (When and if that happens only really FC could guess.)
-helixone
I'm familiar with the concept, and I don't think that there is any conflict whatsoever (right now) between growing total hashrate and the ability to keep cold spares. Think about it: the delay of deployment due to learning how to deploy these quickly has already been flushed out. The estimated rate of 60 boards per day is an estimate of actual deployment, and can't be used as a "peak" value of actual hardware racking - nobody but Friedcat's team knows the actual rack rate. Add to that the knowledge that the percentage of the network has not only kept up, but has INCREASED at the same time, and we can extrapolate that the deployment of hardware has accelerated beyond the initial few weeks' rate. They're learning how to deploy faster, so there's no reason to put a limit on their deployment efforts just because we don't see the effect on the network hashrate. That's how I came up with my theory... mind you, it is just a theory, but I did take into account the idea that the deployment must accelerate beyond the growth of the network in order to maintain the current percentages. Of course, it won't stand up perpetually, but for now, deployment is accelerating beyond what's necessary. And that makes me think that there's no reason they can't assembly-line this into a cold standby situation. That's how I think I'd do it, anyway, in order to avoid growing hashrate too quickly while leveraging the workforce putting the hardware into place at peak efficiency.
Another thing to consider. FC has only ever stated an upward bound on deployment of 50% of network. To get there they would have to build and deploy 2000 of the blockeruptor boards overnight. I'm fairly confident that no matter how fast they can build and deploy these things that there is any risk of them doing so at this point in the game. The network has to many hashes.
I stand my my theory that they are now deploying as fast as they can, and are no longer holding back. (Earlier on I would have agreed that what you were speculating might be a possibility.)
When I said peak rate, I didn't mean they have hit a peak, and they can't go higher, I meant that they have proven the ability to deploy 60 boards a day over a 10 day difficulty period, and that is the fastest rate they have been deploying hardware to date.
-helixone