If you hold $TAO, you have heard this critique a hundred times: subnets earn more from emissions than from real customers. Bittensor cofounder Const just answered it head on at Exploit Summit. His answer changes how you should read the revenue numbers. The critique, in his words, is fair He did not dodge it. He called it a reasonable critique. Then he zoomed out. Almost nobody in AI is profitable right now. He pointed to inference companies like Baseten and Together, which he says spend more serving models than they earn. The whole industry is running at a loss to grab market share, because demand keeps exploding. In his view, Bittensor's high emissions reflect that same demand, not a broken model. The part most people miss Here is the difference he highlighted. AI companies borrow, raise and burn cash. If the bubble pops, they can run out of money or drown in debt. Bittensor is a protocol, not a company. If compute prices rise, emissions simply buy less compute. There are no loans to default on and no bank account to drain. It bends with the market instead of breaking. His bottom line: revenue matters, but the bigger test is whether these networks can produce state of the art intelligence at lower cost. Three other things he revealed 1. Miners are paid to attack the system Bittensor lets a global swarm try to exploit every subnet, and builds around whatever they find. On Lium, validators now probe rented machines in real time and push software fixes the moment they spot a cheat. Some subnets ship multiple updates a day. 2. Bittensor has become a backdoor to scarce GPUs With export controls on top Nvidia chips, he says Lium, Targon and NodeExo are among the few places in the world where anyone can rent them. He also said Huawei's Ascend chips are set to launch on Lium this year. 3. Anyone can mine, not just PhDs His advice to the average person: run an agent like OpenClaw or Hermes in the background and point it at subnets. In his words, PhDs were never really the winners on Bittensor. People with grit were. The risk he flagged Const believes the biggest threat is not other AI labs. It is governments. He warned that as regulation tightens, open networks that fund AI outside the big labs could become a target. For TAO holders, that is a real risk to weigh, not just a slogan. My read on all of this His answer does not erase the gap between emissions and revenue. That gap is real, and I will keep tracking it. But it reframes the question. The bear case asks why Bittensor is not profitable yet. The better question is which survives if the AI money machine slows down: companies carrying debt, or a protocol that adjusts to the price of compute. Which side of that do you think wins? https://t.co/2B52G82Gze