Contributor: School of Crypto
in the past couple of weeks i’ve delved deep into the bittensor miner analytics – particularly in how much each miner makes and which subnets are the most profitable to mine. i posted a tweet a little while ago showing the top earning miners in one day on bittensor. people seem to be interested in this material, so i decided to look a bit closer at the overall mining landscape in subnets.
the following are metrics for one day earnings for the top 50 subnets that are not currently conducting a 100% miner burn:
top 20 subnet reward daily earnings (for top 10 miners)
compute subnets are dominating miner emissions in the top 10
as you can see, compute/inference subnets, led by chutes are completely dominating the miner rewards in the top 10. lium, targon, gradients and compute horde also have divvied up some sizeable rewards to miners.
it’s notable to see ridges and affine as ai coding agent/reinforcement learning agents in 2nd and 3rd place. these subnets are consistently showing very high rewards to miners on a daily basis.
top 21-50 subnet reward daily earnings (for top 10 miners)
the 21-50 subnets tend to have miners earning small amounts of rewards or just 1 to 2 miners that are earning all the rewards. i look forward to digging deeper into each of these incentive mechanisms in order to find out exactly why there are either a small amount of miners earnings most of the rewards, or lots of miners earning small amounts of emissions.
categorization of top rewards per miner
most miners make less than 1 TAO daily, with a small percentage earning 10 or more a day
out of the 288 miners tallied in this report, more than two thirds of miners are making less than 1 TAO a day. subnets like hippius, ready ai, vidaio, data universe, zeus, hermes, tao private network, and alphacore have miners making less than .2 TAO or less on their subnets.
chutes dominates the high tier with all top 10 miners making 10 TAO or more. lium had 9 miners all making 11.4 TAO each. affine has 3. targon had it’s top 3 miners making 35.23, 27.79, and 27.09 TAO. ridges has 1 (which is the top earning miner in the land)
the medium tier is spread around between the top miner earning subnets and lower ones on the list that have 1 or 2 that make 1-10 TAO with the rest making little to no earnings.
ridges still leads the way with it’s winner take all incentive mechanism for it’s ai coding agent subnet
ridges led all subnets in individual miner daily earnings w/ a whopping 107.65 TAO earned. affine came in second w/ 76.88 TAO pay day. 5 of the top 10 earners are compute subnets led by chutes and targon. it pays to be a top miner in a top subnet!
takeaway
it appears that there are a handful of subnets in the bittensor ecosystem that pay extremely well and then there’s a pretty big drop off after the top 5. it’s notable that the compute subnets are the top earners. however, it’s important to take into consideration the expenses of these miners in order to earn such rewards.
it would be interesting to see how much take home pay these top compute (chutes, targon, lium, gradients) miners make. on the other hand, if you are making a top coding or reinforcement learning agent on ridges and affine, you are doing quite well for yourself and blowing all the other miners out of the water.
most of the other subnets have 1 or 2 medium sized earners in the 1-10 TAO range with the rest in the decimal TAO rewards.
this report is literally just taking one day’s numbers into account. i’d like to sum up a weekly and monthly report to see if this small granular look is consistent in a longer time horizon. what are your thoughts?
You’re right to wonder about the expenses of the compute miners, but it’s also the upfront stake required. I looked at Chutes where people are apparently mining with consumer grade GPUs, but to be competitive I believe you need tens of thousands of TAO staked. That’s a very high barrier to entry.