Topic: Experienced GPU miner thinking about starting an ASIC farm; what should I know? - page 2. (Read 405 times)

This day-ahead index is a specific agreement with the energy seller, not oncor, right?

I 100% agree, of course GPU mining will live on, but right now isn't a good time to enter because of the threat of PoS. I definitely will get back into GPU mining when equipment is cheaper. My strategy is more long term.

The problem with the AI/rendering compute services is: (1) there are few customers (2) it requires reconfiguring the rig, which will cost 30% more in equipment + 30% more in power per card.

I was thinking that I could buy them in order to get to 100 kW as quickly as possible in order to secure lower per-kWh rates to prepare for the upcoming GPUs or newer ASICs, and especially to attract outside investors. I would sell the S9's for scrap after a year after they serve that purpose. I also think scaling to 100 kW will give me some experience with the construction/HVAC part.

Should I move up to a slightly higher tier, like the L3 series, Innosilicon ZMaster or T15?

I wanted to go with the day-ahead index, where I always know what the rate will be in 24h and it's not subject to instant spikes like the real-time LMP. I'm not worried about the spikes because I will write custom, automated software to cut off the Internet to the ASICs to power them down. Maybe other farms will want the software.

I wrote code to analyze historical index prices on ERCOT, and it looks like I can go from 5.3¢ --> 4.5¢ for energy while only losing 5% uptime. That will barely affect the demand charge.

That's a very good point. This is the hard part. I need to find a place with heavy power at a reasonable rent. Thanks for reminding me I can't just expect more power and I always have to keep a reserve of capacity. I would rather spend that $40k on replacing the S9's with more efficient ASICs than on ONCOR, which is a waste if I leave that location.

I've definitely seen warehouses that have their own transformers, and they tend to have much heavier capacity. The rates are cheaper too. I need to look for those in particular. I think I also have to find buildings that are wider rather than longer, so that hot air exits right out the back.

The point is to demonstrate a working, profitable farm to investors before going bigger. At that point, I would try to lease land and install metal sheds with my own transformer equipment.

I don't think GPU mining is guna "die" any time soon, eth or not there will always be something to mine with them, not to mention they are versatile and you can run all sorts of other algorithms (AI, chemical modelling, etc) as well if you ever need to pivot.

I'm curious why you want to go with s9's? they are quite old at this point, might get better ROI with something newer.

It looks like you are trying to go with index power rate? you gotta be real careful with that, if prices spike and you don't manage the miners in time (can happen overnight), your energy bill will basically ruin a years worth of profitability

Does your location already have power available? If you need to have oncor install more power, they are guna charge you potentially a lot ($40k+) and they *might* refund you a minimum of 4 years later.

I've been GPU mining since early 2013 on and off, starting with a single Radeon HD 7850 video card, then multiple R9 270's during the Dogecoin boom, then with 200 GPUs (RX 570's and GTX 1060's) during the 2017-2018 ETH gold rush with help from outside investors. My business went bankrupt in late 2018 because a big investor pulled out after ETH started falling, we only got 200 cards instead of 1000+ so we defaulted on the lease, and I scaled back down to a single rig in 2019-2021.

But now, with my regular job (software development) allowing me to work fully remote, I'm really interested in moving to somewhere with cheap-ish power and opening a small warehouse again. Since I'm currently in NJ, I thought about upstate NY first, but I don't like the BitLicense crap and state income tax, plus it's the middle of nowhere. I also considered NH/MT because they have no sales tax, but I'm interested in having a real life outside mining. OR/WA have cheap power but are too far away timezone-wise, as my job is located in EST.

My primary destination would be the Dallas TX area, unless one of you can recommend a better state. It seems to be an up-and-coming destination for mining farms and Bitcoin people (should be easier to find investors), plus the regulators are nice. The heat could be a problem, but at least the climate is dry. The best power cost I can get on ONCOR seems to be 7.0¢ in total (5.2¢ energy, 1.8¢ demand). I'm hoping to get day-ahead pricing and lower the cost to 6.0-6.5¢ by shutting down during price spikes.

My initial budget is $60k: $50k for the equipment and $10k for the lease deposit/construction costs. I want to add $6k of equipment per month, for a total of $100k invested by EOY. Warehouse rent will cost $1400-$2000/month and I think I can get a 3PH 600A 240v supply for 200 kW of capacity (600 x 240 x 1.73 x 80%). I have the financial stuff, like how to avoid income tax & sales tax, under control thanks to my accountant.

However, with the uncertainty about ETH PoS, I don't want to buy GPU rigs again. I'm thinking of starting out with ASICs. I did some research and I think I can start out with 100 Antminer S9's for $380 each. With custom firmware, I could run them at 13.5 Th @ 1100w. Mining calculators are telling me each one could make $1.40/day of profit, which is a ~270 day payback period. Of course, BTC difficulty will keep growing and that margin will be squeezed and that doesn't account for other costs.

But I'm new to ASIC mining, so I'm worried about the following stuff:

• 1: What would the failure rate be on these old S9's? I heard that ASICs die quicker than GPUs.
• 2: Is 7¢ too high of a rate to get into ASIC mining, especially with Antminer S9s? If BTC price falls below $30k, the profit margin looks like it will be so low it will be unviable for me.
• 3: How can I get insurance for this business that has the highest chance of paying out if there's a fire?
• 4: Can I get away with installing 240v 6-20R outlets then plugging the PSUs directly into them to avoid the cost of PDUs? I used to have 30A L6-30R outlets.
• 5: Is it worth mining with ASICBoost? Why isn't everybody doing it?
• 6: Should I go with the Antminer L3 series instead? It seems they're selling for 1-year low prices and less of the revenue is spent on power.

Once ETH goes PoS, and I locked in a cheap power rate due to demonstrating that 100 kW of demand to the suppliers thanks to the Antminers, I would stop investing in new ASICs and start buying up cheap GPUs. I think that could be a great time to get into large scale GPU mining with older cards like the GTX 1000's / RX 500's, which should crash the most in price after the n00b GPU miners panic. At 6-7¢/kWh, it's probably a better long-term choice than ASIC mining.

Any recommendations from the seasoned ASIC operators? Should I just wait for PoS and stick to GPU mining because it's what I know best?