Choosing a subnet

Subnet choice determines your ceiling. This guide covers how to narrow the field with on-chain data, and where you have to stop trusting numbers and go read code.

Get the order right

The most common mistake is starting from the emission leaderboard. The subnet paying the most attracts the strongest competitors and the highest registration cost, which is precisely why it pays the most.

A more reliable order:

1
Start from what you can actually build
Filter to subnets whose task overlaps with skills you already have. A mid-tier subnet where you understand the problem beats a top subnet where you do not.
2
Check that entry is realistic
Confirm registrations are open, look at free UID slots, and read the current burn cost. A closed or full subnet changes the calculation entirely.
3
Only then compare economics
Among subnets that pass the first two filters, compare miner pool size and per-miner averages to decide where your effort is best spent.

Building a shortlist

The subnet explorer includes miner-oriented filters, each sorting on a value read from the chain:

Filter	What it sorts on	Use it when
Open UID slots	Registrations allowed and at least one free slot.	You want to enter without evicting an existing miner.
Lowest registration cost	The current recycled-TAO burn.	Your budget is the binding constraint.
Largest miner pool	Daily TAO emission allocated to miners.	You are confident you can rank competitively.
Highest per miner	Miner pool divided by active miner count.	You want reward relative to the size of the field.

Add candidates to the comparison tool in miner mode to see registration cost, miner pool, free slots, and immunity period side by side.

Watch the registration cost before you commit

The burn rises with each registration and decays when nobody registers. TaoScope charts the costs it has observed on each subnet page, so you can tell whether today is expensive or cheap relative to recent history.

The numbers that matter

Slot utilisation

Occupied UIDs against the ceiling. Below the ceiling you can register into empty space. At the ceiling, every registration evicts someone, and you will be the newest and lowest-ranked neuron in a field that has already survived that test.

Registration cost

Your entry price, paid in TAO and not refundable. Compare it against what a realistic ranking would earn per day — not against what the top miner earns.

Miner emission pool

The subnet's daily emission multiplied by the 41% miner share. This is the total prize that all miners on the subnet divide between them.

Measured reward distribution

TaoScope reads the subnet metagraph and reports the top miner's share, the median earning miner's share, and how concentrated the top ten are. This is the most useful signal on the page: a subnet where the top ten hold almost all incentive is a subnet where mid-table placement earns very little.

Immunity period

How long you are protected from eviction after registering. Short immunity demands that your miner works correctly on the first attempt.

Working out whether it pays

Base the estimate on a ranking you can plausibly reach, not on the top slot. Using the median earning miner is a more honest starting point for a newcomer.

Break-even estimate

daily_earnings = miner_pool_per_day × your_expected_incentive_share
days_to_break_even = registration_cost / daily_earnings

Every input except your expected share is shown on the subnet page. For the share, the measured median gives you a defensible figure for “a miner in the middle of this field”.

Then add the costs that are not on chain

Break-even on the registration burn alone is not break-even. Add compute, bandwidth, electricity, and your own time. A payback figure that looks acceptable at 41 days may not survive a GPU rental bill.

These are snapshots, not forecasts

Every figure reflects current conditions. Emission share moves, competitors arrive, validators reweight, and alpha prices change. Nothing here predicts future earnings.

What the data cannot tell you

On-chain data narrows the field. It cannot make the decision, because the things that most determine whether you succeed are not on chain:

What the task actually is, and whether you can do it well.
What hardware is genuinely required, as opposed to what a README claimed a year ago.
How the validator scoring function works — the specification you are graded against.
Whether the subnet is actively maintained and whether its incentive mechanism is about to change.
How responsive the operator community is when your miner behaves unexpectedly.

For each of those, go to the subnet's repository and Discord, both linked from its TaoScope page. TaoScope deliberately does not publish task categories, hardware tiers, or difficulty ratings, because we cannot source them reliably and a confident-looking wrong answer is worse than none.

Tip

Once you have a candidate, work through Start mining and rehearse on the test network before spending anything.

Last updated August 7, 2026