Oh yeah, forgot that the electricity is charged annually at the 2nd location, that gives us an advantage because we can get for the electricity price an average of 6% interest over the year, rather safely too!
Tho, likely we would save the electricity funds in this proportion: 25% BTC, 25% Stock/Investments (Rather liquid, but has an inherent risk), 25% Liquid Funds & Bonds, 25% cash or further HW investments expected to gain 100%+ in cashflow before electricity is due (January of each year) (100%+ received on HW does not mean it's been paid off).
Topic: Mining resources hosting/leasing service, worth it? - page 3. (Read 4452 times)
There are three major concerns.
1) Trust. Best prices are for contract up front. Collect a lot of contracts and disapear.
2) Non-linear cost. Are you going to have insurance, electronic friendly fire protection, dedicated warehouse space, 24 hour security? As the operation scales the risks scale too and without mitigation those risks are simply passed to customers.
3) Electricity is the most important geo dependent variable. If this did make sense it would make sense in a sub 8 cent per kWh area. Otherwise it is more risk and cost for a less efficient operation.
1) Trust. Best prices are for contract up front. Collect a lot of contracts and disapear.
2) Non-linear cost. Are you going to have insurance, electronic friendly fire protection, dedicated warehouse space, 24 hour security? As the operation scales the risks scale too and without mitigation those risks are simply passed to customers.
3) Electricity is the most important geo dependent variable. If this did make sense it would make sense in a sub 8 cent per kWh area. Otherwise it is more risk and cost for a less efficient operation.
1) Trust is non-issue. We are a finnish registered company, and doing fraud like that i would be personally liable in the court. Our VAT ID is: FI22551954
We got thousands of customers, so any trust issues are extremely bad for us. We've acted even on hints of trust issues within 24hrs in the past, hints caused by people who dislike our success.
2) Already got insurance up to 1million euro. Electronic friendly fire protection not right now, we got fire protection, just not electronics friendly. Good point.
Dedicated warehouse space: That 150m2 location
More is available fast, or with some searching cheap.24/7 security: Camera surveillance, including online. Access very limited, "off the beaten path". Cops visit occasionally the parking lot, along with security company which visits couple times a night.
3) Unfortunately that is impossible pricing at Finland, alone the electricity added tax is is about 0.03$/kWh. Like said, upside is no need for AC.
If 100% gain from no need for AC, that already drops effective price to ~0.13$/kWh.
AC is certainly not needed 10months of a year. Infact, problem might be we still might need some HEATING during 3-4months of a year! :O
electricity is rather expensive for us - 0.18$/kWh in primary location for GPUs and 0.16$/kWh
ya, that's going to be a problem. probably even a show-stopper level problem.
On the upside we don't need AC, and even tho office has AC it's not calced in because it would have this in any case and i could put the FPGAs on another room to get the load off the AC.
Summer fans + evap should be more than sufficient for the GPU farm.
Assume 100% gain on that, it's 30% cost savings.
Floor space is also extremely cheap. Basic nodes are extremely cheap. UPS costs a bit if done, but not that bad.
A lot of things are very cheap infact, but power is not one of them.
Also on another upside, some people are not capable to run their own miners, skillwise or space wise. Also with us you'd have less upfront costs. The setup fees need to be adjusted tho, or 2 levels offered, roughly 50% of HW cost, 100% of HW cost.
You also get "insurance" in the sense that HW will be replaced free of charge if failed. The monthly fees cover this.
If a particular GPU is unavailable it will be upgraded to similar price & W/Mhash free of charge, or the setup fee considered for 12-24months "value" and prorata credit towards next device.
Grown enough, delivery times will be also completely different than waiting for your own hardware etc.
There are three major concerns.
1) Trust. Best prices are for contract up front. Collect a lot of contracts and disapear.
2) Non-linear cost. Are you going to have insurance, electronic friendly fire protection, dedicated warehouse space, 24 hour security? As the operation scales the risks scale too and without mitigation those risks are simply passed to customers.
3) Electricity is the most important geo dependent variable. If this did make sense it would make sense in a sub 8 cent per kWh area. Otherwise it is more risk and cost for a less efficient operation.
1) Trust. Best prices are for contract up front. Collect a lot of contracts and disapear.
2) Non-linear cost. Are you going to have insurance, electronic friendly fire protection, dedicated warehouse space, 24 hour security? As the operation scales the risks scale too and without mitigation those risks are simply passed to customers.
3) Electricity is the most important geo dependent variable. If this did make sense it would make sense in a sub 8 cent per kWh area. Otherwise it is more risk and cost for a less efficient operation.
electricity is rather expensive for us - 0.18$/kWh in primary location for GPUs and 0.16$/kWh
Ya, that's going to be a problem. Probably even a show-stopper level problem.
Pricing Edited 14/03/2012
Starting to near final production costs + prices IMHO.
Options for +~50% setup fee of hardware value for lower monthly will be provided if there is demand.
Setup fees: 195€ for billing terms monthly & quarterly, 75€ for longer cycles (6months, 12months).
RATES:
BFL Single: 49.95€
Icarus/X6500: 37.95€
Z-Tex: 29.95€
7970: 42.95€
6770: 19.95€
5830: 33.95€ ** TO BE VERIFIED -> Unknown power consumption per GPU
Quarterly rate(3 months): Monthly rate * 3
Semi-annual rate(6 months): Monthly rate * 6
Annual rate(12 months): Monthly rate * 12
PRICES IN USD: RATE * 1.34 (Xe.com 27th of March)
Which results in Mhash per € costs:
BFL Single: 16.61Mhash / €
Icarus/X6500: 10.01Mhash / €
Z-Tex: 7.01Mhash / €
7970: 12.80Mhash / €
6770: 10.02Mhash / €
5830: 9.43Mhash / €
This means that at 6month payment cycle (no setup fee later on), your effective 1st year costs would be, including overheads, electricts etc etc, only the single payment, no hidden charges:
BFL Single: 599.4€
Icarus: 455.4€
Z-Tex: 359.€
7970: 515.4€
6770: 239.4€
5830: 407.4€
Expect 2-4% lower price on 12month payment cycle.
All prices exclude VAT 23%
Reseller options will be featured. Bulk discounts will be offered. Please contact me for additional information.
The important questions are ....
How can you offer below HW costs?
Economies of scale, long term goals. The HW payoff is set far into the 2nd year, only thing that matters is positive constant cash flow.
Standardized setups will allow us to spend a lot of R&D to push through more efficiency.
For europeans just acquiring the HW will cost more than the annual price! Slight difference, but not much.
This kind of pricing will allow bigger operators to scale faster if supplies last.
Why, oh why? Couldn't you just mine and earn more?
Likely, yes. But this often yields revenue upfront vs. mining which is constant small flow. Also there will be some HW to hash for ourselves, but that is mostly the items which is waiting to be provisioned/burn-in test, old HW, testing setups etc. which might yield for us at some points of time even 10Ghash/s+ in burn-in testing!
Also, this will yield something which i appreciate: Known, stable return on investment. BTC value goes up & down, and we would at times need to wait months before we could sell BTC, making scaling efforts really slow at times. Personally, i prefer to hold on to my BTC anyway!
When?
Proof of concepts will be built by end of April, launch shortly there after. Some beta accounts will be available.
Full sales should begin no later than end of August, targeting 1st of June.
What we need to do is:
What if no one is interested?
Doesn't matter, we'll just mine bitcoins for ourselves then
Sure then a lot of hours will be wasted into research and development which would have not otherwise been done, but overally i consider the opportunity costs to be rather fair.
Who can get an beta account?
Those who can give us productive feedback, only a few will be accepted
How are you able to finance this?
We are an established hosting company with some options for self financing few ghash growth per month on average without affecting other operations at all.
If demand is great we might accept a limited amount of preordering, preorders would be limited to something which we can handle to provide within a ~month, not 4-6months
For FPGAs the queue might be longer due to the nature of FPGAs arriving in batches, and there is no constant supply.
Shipments from US (ie. BFL Single) happens only once every 1-2months as well because it makes no sense to have small shipments from our US warehouse.
Do i get HW access, so can i OC the GPUs?
NO! OC'ng will make the lifespan of the GPU far shorter, in extreme cases into just weeks instead of years, nevermind the increased wattage and lower Mhash/W ratio. Further, there is group of people who doesn't give a shit about others and will kill the GPUs in a month if it yields them even 0.1€ more revenue, and would cause expenses to skyrocket with constant RMAs and replacement HW purchases (You can RMA single GPU only so many times ...). Further crash of single GPU causes whole node to crash, likely causing downtime for other customers on the same node as well.
That being said, we are targeting high performance, for example 600-650Mhash for 7970, if not more. Final testing will show.
FPGAs will run at the rates they are delivered, or in case of X6500 what every unit can run with combined few % safety margin, likely 190-200Mhz range. Icarus will yield ~380Mhash, BFL Single ~830.
Scale maximum?
2012 Fall: Current locations will be scaled to a total of approximately: 20Ghash/s in GPUs and 20Ghash/s for FPGAs.
2012 Winter: 60Ghash/s in GPUs utilizing current locations (Requires new electric circuits). FPGAs can be temporarily be hosted where GPUs are.
2013: Likely new location will be acquired, scouting will take a little bit of time. New location will add a minimum of ~20kW capacity at the start (80 to 200Ghash/s), but might require AC for cooling. We have a potential backup location with 6kW capacity available. We will acquire similar locations on as-needed basis.
Starting to near final production costs + prices IMHO.
Options for +~50% setup fee of hardware value for lower monthly will be provided if there is demand.
Setup fees: 195€ for billing terms monthly & quarterly, 75€ for longer cycles (6months, 12months).
RATES:
BFL Single: 49.95€
Icarus/X6500: 37.95€
Z-Tex: 29.95€
7970: 42.95€
6770: 19.95€
5830: 33.95€ ** TO BE VERIFIED -> Unknown power consumption per GPU
Quarterly rate(3 months): Monthly rate * 3
Semi-annual rate(6 months): Monthly rate * 6
Annual rate(12 months): Monthly rate * 12
PRICES IN USD: RATE * 1.34 (Xe.com 27th of March)
Which results in Mhash per € costs:
BFL Single: 16.61Mhash / €
Icarus/X6500: 10.01Mhash / €
Z-Tex: 7.01Mhash / €
7970: 12.80Mhash / €
6770: 10.02Mhash / €
5830: 9.43Mhash / €
This means that at 6month payment cycle (no setup fee later on), your effective 1st year costs would be, including overheads, electricts etc etc, only the single payment, no hidden charges:
BFL Single: 599.4€
Icarus: 455.4€
Z-Tex: 359.€
7970: 515.4€
6770: 239.4€
5830: 407.4€
Expect 2-4% lower price on 12month payment cycle.
All prices exclude VAT 23%
Reseller options will be featured. Bulk discounts will be offered. Please contact me for additional information.
The important questions are ....
How can you offer below HW costs?
Economies of scale, long term goals. The HW payoff is set far into the 2nd year, only thing that matters is positive constant cash flow.
Standardized setups will allow us to spend a lot of R&D to push through more efficiency.
For europeans just acquiring the HW will cost more than the annual price! Slight difference, but not much.
This kind of pricing will allow bigger operators to scale faster if supplies last.
Why, oh why? Couldn't you just mine and earn more?
Likely, yes. But this often yields revenue upfront vs. mining which is constant small flow. Also there will be some HW to hash for ourselves, but that is mostly the items which is waiting to be provisioned/burn-in test, old HW, testing setups etc. which might yield for us at some points of time even 10Ghash/s+ in burn-in testing!
Also, this will yield something which i appreciate: Known, stable return on investment. BTC value goes up & down, and we would at times need to wait months before we could sell BTC, making scaling efforts really slow at times. Personally, i prefer to hold on to my BTC anyway!
When?
Proof of concepts will be built by end of April, launch shortly there after. Some beta accounts will be available.
Full sales should begin no later than end of August, targeting 1st of June.
What we need to do is:
- Web interface
- Interface pool setup
- Lots and Lots of statistics and graphs
- HW testing, finalizing setups and verifying the average GPU can work at the clocks we want to
- Setup GPU farm location ready to accept nodes: building shelves, wiring, ensure stability etc.
- API for advanced users
- Website + billing (graphics being the most time consuming)
- Setup other usage options for GPGPU, ie. Folding@Home
What if no one is interested?
Doesn't matter, we'll just mine bitcoins for ourselves then
Sure then a lot of hours will be wasted into research and development which would have not otherwise been done, but overally i consider the opportunity costs to be rather fair.Who can get an beta account?
Those who can give us productive feedback, only a few will be accepted
How are you able to finance this?
We are an established hosting company with some options for self financing few ghash growth per month on average without affecting other operations at all.
If demand is great we might accept a limited amount of preordering, preorders would be limited to something which we can handle to provide within a ~month, not 4-6months
For FPGAs the queue might be longer due to the nature of FPGAs arriving in batches, and there is no constant supply.
Shipments from US (ie. BFL Single) happens only once every 1-2months as well because it makes no sense to have small shipments from our US warehouse.
Do i get HW access, so can i OC the GPUs?
NO! OC'ng will make the lifespan of the GPU far shorter, in extreme cases into just weeks instead of years, nevermind the increased wattage and lower Mhash/W ratio. Further, there is group of people who doesn't give a shit about others and will kill the GPUs in a month if it yields them even 0.1€ more revenue, and would cause expenses to skyrocket with constant RMAs and replacement HW purchases (You can RMA single GPU only so many times ...). Further crash of single GPU causes whole node to crash, likely causing downtime for other customers on the same node as well.
That being said, we are targeting high performance, for example 600-650Mhash for 7970, if not more. Final testing will show.
FPGAs will run at the rates they are delivered, or in case of X6500 what every unit can run with combined few % safety margin, likely 190-200Mhz range. Icarus will yield ~380Mhash, BFL Single ~830.
Scale maximum?
2012 Fall: Current locations will be scaled to a total of approximately: 20Ghash/s in GPUs and 20Ghash/s for FPGAs.
2012 Winter: 60Ghash/s in GPUs utilizing current locations (Requires new electric circuits). FPGAs can be temporarily be hosted where GPUs are.
2013: Likely new location will be acquired, scouting will take a little bit of time. New location will add a minimum of ~20kW capacity at the start (80 to 200Ghash/s), but might require AC for cooling. We have a potential backup location with 6kW capacity available. We will acquire similar locations on as-needed basis.
I've been thinking about this since last summer, but it brought up again lately and realized this might make a lot of sense for some people.
My idea is to offer mining hardware for hosted lease, with some customization options, and a range of products to choose, from GPUs to FPGAs.
Pricing would be set according to real world pricing and cost to operate and acquire the hardware - not by the potential BTC mining results.
You would directly pay for the mining hardware, not the infrastructure etc. you wouldn't need to care yourself about the physical setup etc.
Price would include hardware replacements, service level agreement, and all the overheads needed to operate something like this.
You wouldn't not need to worry about anything else than "What pools do i want to use?".
You would get access to RRD graphs of YOUR cluster, with each device listed and total graphs.
At later point we will also enable you to change pool configurations on the fly for yourself, and for the whole cluster - maybe also some other configuration options to a limit.
The performance would be managed by us, meaning we first optimize it for best MH/W combination.
The biggest drawback is that electricity is rather expensive for us - 0.18$/kWh in primary location for GPUs and 0.16$/kWh where a small cluster of FPGAs will be set up (office). The upside is that we don't need AC being located near to the arctic circle
Office has AC so FPGAs will run cool, with low rejects and high speeds.
The SLA would cover:
* Network uptime
* Hardware uptime
* Hardware replacements
Support is 24/7 for basic questions, hardware technical (replacements) is ~12/7, commercial is available ~12/5 (business days)
Capacity to grow offsite location is up to around 100kW (cooling limited), office 7kW.
Offsite location has 24/7 recording camera surveillance as well.
GPUs already available or en-route:
All current FPGA options will be made available, with longer delivery times:
Some GPUs will be upgraded with better after market coolers if the temperatures raise too high. Target is to keep them below 70C.
Offsite location will likely also feature UPS backup at a later date.
Office will likely NOT have UPS for the FPGAs
My idea is to offer mining hardware for hosted lease, with some customization options, and a range of products to choose, from GPUs to FPGAs.
Pricing would be set according to real world pricing and cost to operate and acquire the hardware - not by the potential BTC mining results.
You would directly pay for the mining hardware, not the infrastructure etc. you wouldn't need to care yourself about the physical setup etc.
Price would include hardware replacements, service level agreement, and all the overheads needed to operate something like this.
You wouldn't not need to worry about anything else than "What pools do i want to use?".
You would get access to RRD graphs of YOUR cluster, with each device listed and total graphs.
At later point we will also enable you to change pool configurations on the fly for yourself, and for the whole cluster - maybe also some other configuration options to a limit.
The performance would be managed by us, meaning we first optimize it for best MH/W combination.
The biggest drawback is that electricity is rather expensive for us - 0.18$/kWh in primary location for GPUs and 0.16$/kWh where a small cluster of FPGAs will be set up (office). The upside is that we don't need AC being located near to the arctic circle
Office has AC so FPGAs will run cool, with low rejects and high speeds.
The SLA would cover:
* Network uptime
* Hardware uptime
* Hardware replacements
Support is 24/7 for basic questions, hardware technical (replacements) is ~12/7, commercial is available ~12/5 (business days)
Capacity to grow offsite location is up to around 100kW (cooling limited), office 7kW.
Offsite location has 24/7 recording camera surveillance as well.
GPUs already available or en-route:
- 5830
- 5850
- 5870
- 6950
- 6770
- 7970
All current FPGA options will be made available, with longer delivery times:
- BFL Single
- Icarus
- X6500
- Z-Tex
Some GPUs will be upgraded with better after market coolers if the temperatures raise too high. Target is to keep them below 70C.
Offsite location will likely also feature UPS backup at a later date.
Office will likely NOT have UPS for the FPGAs
Jump to: