Garrett and Keith Walks Through TAO’s New Emissions Reality

Garrett and Keith covered one of the more consequential protocol changes of the past month on Carrot and the Stick’s episode 14.

The discussion focused on how TAO Flow is now calculated, with chain buys, miner registrations, and root sell pressure all being repriced into the inflow-outflow math that determines subnet emissions.

15 Key Points From the Conversation

a. TAO FLOW IS THE METRIC THAT DETERMINES SUBNET EMISSIONS. It measures the difference between $TAO flowing into a subnet versus $TAO flowing out, and it doubles as a signal to investors about whether a subnet is net-extractive or net-attractive.

b. 3,600 $TAO PER DAY GOES INTO SUBNET POOLS, that’s half a $TAO every 12 seconds. The chain decides whether to inject that as standard emissions (TAO plus matching subnet ‘$ALPHA’ token, keeping price stable) or as a chain buy ($TAO only, propping up the price).

c. CHAIN BUYS ARE SELECTIVE. The protocol does not buy into every subnet. Chain buys are deployed when there is a delta between a subnet’s emissions and its price, where activity warrants a higher price but the market is not delivering it.

d. THE CHAIN BUY MECHANIC JUST CHANGED. Previously, when the chain bought $ALPHA from a subnet pool, the $ALPHA was effectively recycled or deleted with no holder. Now, that $ALPHA accumulates in a protocol-controlled storage map.

e. THE PROTOCOL CAN NOW BE PAID OUT AT DEREGISTRATION. If a subnet deregisters, the protocol itself receives a pro-rata share of the pool’s $TAO, just like any other $ALPHA holder. Previously, all of that value would have gone to alpha holders alone.

f. CHAIN BUYS ARE NOW COUNTED AS $TAO INFLOW. Previously, a subnet receiving aggressive chain buys could see its net TAO Flow go negative even though the chain was effectively investing. The recalibration corrects that distortion.

g. MINER REGISTRATIONS ARE ALSO NOW COUNTED AS TAO INFLOW. Miners buy $TAO and swap it for $ALPHA to purchase a registration slot, and that swap is now recognized as inflow into the subnet.

h. ROOT SELL PRESSURE IS NOW COUNTED AS TAO OUTFLOW. Root validators auto-sell the $ALPHA emissions they earn for $TAO to pay their stakers, which is structurally a sell operation against the subnet.

i. NEW SUBNETS START WITH 100% ROOT PROP. Root proportion takes roughly 100 days to drop to 50%, which means new subnets carry significant structural outflow weight against their TauFlow calculation for their first months.

j. THE FLYWHEEL IS HARDER TO START BUT MORE DEFENSIBLE ONCE RUNNING. New subnet owners now have to actively build positive $TAO inflow to counterbalance root cell pressure, which raises the bar for what gets through the first months.

k. DEREGISTRATION IS WORKING AS DESIGNED. Painful for affected miners, $ALPHA holders, and operators, but structurally healthy. The hosts framed complaints about deregistration as a signal that the mechanism is functioning correctly.

l. SUBNET INVESTING IS STRUCTURALLY ANGEL INVESTING, NOT STOCK INVESTING. Most subnets will fail. The hit rate looks more like venture capital than public markets, but the winners can be 100x or more.

m. CONST’S “FUTURE OF WORK” FRAMING ANCHORED THE BACK HALF OF THE EPISODE. Anything that can be well-defined and measured digitally will eventually be produced by agents using digital commodities. Bittensor is the economic substrate that lets those agents self-optimize.

n. THE AGENT ECONOMY NEEDS AN INCENTIVE LAYER. Today’s agents spend tokens against credit card allowances with no economic feedback loop. Bittensor’s subnet model gives agents an environment where they can produce digital commodities, consume each other’s outputs, and operate inside a self-optimizing ecosystem.

o. THE HOSTS’ CLOSING TAKE. Embrace AI rather than fight it. The traditional jobs being automated are not the ones humans actually want. The opportunity Bittensor offers is to channel that automation into a structure that produces value for participants rather than concentrating it in centralized platforms.