Quote
So for an update on shipping times... As you might imagine, we get this question a lot. It seems like it would be a simple question to answer, but it's not. Let me describe the process so people can better understand why.
If you look at the current FPGA board in a Single or MiniRig, you'll see lots of capacitors, resistors, etc... about 350 little tiny parts attached to the board. Contrary to some of the conspiracy theories out there, all our boards are completely custom made, they aren't purchased from another manufacturer, etc... they are designed by us and made for us and us alone. As such, we are required to volume source every single part that goes on the board.
The ASICs are similar in so far as they also have nearly 350 components on each board. With the FPGAs, we sourced parts in the hundreds or low thousands at a time. For some of the ASIC parts, we are sourcing hundreds of thousands at a time which requires direct ordering & lead time dependancy from the respective manufacturers. However, for this first batch, we're mostly able to depend on available distribution stock from places like Mouser and DigiKey. Shortly after we get the first batch of everything in, we'll have our larger mega stockpiles arrive from other vendors and/or direct from the distributors, it's just the first batch that's going to be rough.
So, we've got he myriad distributors shipping thousands of little pieces to us, the PCB manufacturer sending us the bare PCBs, the HSF manufacturer sending the HSFs to us, the PSU manufacturer sending the PSU's to us, the case manufacturer sending the cases to us and most importantly, the fab sending the ASIC chips to us. All of these must arrive on time and as expected for everything to go off without a hitch. So far, so good.
When we made our announcement for shipping dates, we padded in some extra weeks in case of delays, and as we try to herd all these cats into one corral, our padding is slowly eaten up with mostly minor problems, but they all add up. With the bump in specs, we spent some time ensuring the power subsystem is over powered to accommodate the new and future power requirements - our chips are capable of higher speeds than what we initially intended to send out in the first batch, and they still have quite a bit of headroom; We decided to go ahead design the power subsystem to handle the maximum theoretical load of the chips. This means we can now crank the board up with some minor tweaking. Each chip is theoretically capable of operating at 1 GHz, we are running them at 500 MHz with the new specs... we will likely never see 1 GHz operations, simply because of heat density issues and a few other factors, but we have at least another 25% of headroom we can play with, if not more. Again, we built in a lot of padding into the specs, just in case something went wrong. We have basically padded everything we could in terms of estimates and that padding is what has allowed us to bump specs on short notice and keep our shipping times in line even in the face of delays.
Ok, so we have the cats herded, the specs staked out, now we have to actually build these things. As many of you know, we've purchased SMT machines to allow us to manufacture our own boards - and I have mentioned this before, but many have not heard it - we will not be using the SMT equipment to process our first batch of boards; we will be using the same house that did the pick and place for our previous generation products, which means we're still at the mercy of someone else for our first batch shipments. There has been some delays at that stage, but we have the padding, so it's not been a critical issue. There has also been some delays at the foundry, but again, we have padding, so it's not been a critical issue. We are also paying for an expedited run at the foundry (which does not come cheap) to keep our timeline up. All these things have to work out perfectly and our timeline is still looking good. However, if something does not work out perfectly, our timeline is going to slip, plain and simple. We've used up most of our padding at this point and we are still ironing out a few little wrinkles here and there. This has been a long explanation for a simple answer: I would like to tell you we are still on time or pretty close to it, because we are. However, I would also like to tell you that we are going to slip a couple weeks or so if anything goes wrong, and given the complexity of the issues facing us, I would say it's almost inevitable something will crop up between now and the beginning of November that we are not expecting; What that is, I don't know yet, but I would rather error on the side of caution, say the timeline is going to slip a little bit and then surprise everyone with an early delivery than promise an early delivery and not meet that promise. So that's what I'm doing and there's your answer. When I have more information, I'll let people know as soon as I can.
If you look at the current FPGA board in a Single or MiniRig, you'll see lots of capacitors, resistors, etc... about 350 little tiny parts attached to the board. Contrary to some of the conspiracy theories out there, all our boards are completely custom made, they aren't purchased from another manufacturer, etc... they are designed by us and made for us and us alone. As such, we are required to volume source every single part that goes on the board.
The ASICs are similar in so far as they also have nearly 350 components on each board. With the FPGAs, we sourced parts in the hundreds or low thousands at a time. For some of the ASIC parts, we are sourcing hundreds of thousands at a time which requires direct ordering & lead time dependancy from the respective manufacturers. However, for this first batch, we're mostly able to depend on available distribution stock from places like Mouser and DigiKey. Shortly after we get the first batch of everything in, we'll have our larger mega stockpiles arrive from other vendors and/or direct from the distributors, it's just the first batch that's going to be rough.
So, we've got he myriad distributors shipping thousands of little pieces to us, the PCB manufacturer sending us the bare PCBs, the HSF manufacturer sending the HSFs to us, the PSU manufacturer sending the PSU's to us, the case manufacturer sending the cases to us and most importantly, the fab sending the ASIC chips to us. All of these must arrive on time and as expected for everything to go off without a hitch. So far, so good.
When we made our announcement for shipping dates, we padded in some extra weeks in case of delays, and as we try to herd all these cats into one corral, our padding is slowly eaten up with mostly minor problems, but they all add up. With the bump in specs, we spent some time ensuring the power subsystem is over powered to accommodate the new and future power requirements - our chips are capable of higher speeds than what we initially intended to send out in the first batch, and they still have quite a bit of headroom; We decided to go ahead design the power subsystem to handle the maximum theoretical load of the chips. This means we can now crank the board up with some minor tweaking. Each chip is theoretically capable of operating at 1 GHz, we are running them at 500 MHz with the new specs... we will likely never see 1 GHz operations, simply because of heat density issues and a few other factors, but we have at least another 25% of headroom we can play with, if not more. Again, we built in a lot of padding into the specs, just in case something went wrong. We have basically padded everything we could in terms of estimates and that padding is what has allowed us to bump specs on short notice and keep our shipping times in line even in the face of delays.
Ok, so we have the cats herded, the specs staked out, now we have to actually build these things. As many of you know, we've purchased SMT machines to allow us to manufacture our own boards - and I have mentioned this before, but many have not heard it - we will not be using the SMT equipment to process our first batch of boards; we will be using the same house that did the pick and place for our previous generation products, which means we're still at the mercy of someone else for our first batch shipments. There has been some delays at that stage, but we have the padding, so it's not been a critical issue. There has also been some delays at the foundry, but again, we have padding, so it's not been a critical issue. We are also paying for an expedited run at the foundry (which does not come cheap) to keep our timeline up. All these things have to work out perfectly and our timeline is still looking good. However, if something does not work out perfectly, our timeline is going to slip, plain and simple. We've used up most of our padding at this point and we are still ironing out a few little wrinkles here and there. This has been a long explanation for a simple answer: I would like to tell you we are still on time or pretty close to it, because we are. However, I would also like to tell you that we are going to slip a couple weeks or so if anything goes wrong, and given the complexity of the issues facing us, I would say it's almost inevitable something will crop up between now and the beginning of November that we are not expecting; What that is, I don't know yet, but I would rather error on the side of caution, say the timeline is going to slip a little bit and then surprise everyone with an early delivery than promise an early delivery and not meet that promise. So that's what I'm doing and there's your answer. When I have more information, I'll let people know as soon as I can.
