The Bittensor Subnet Using Incentive Competition to Screen Drugs, Not Train Models
Subnet 68 (NOVA) points Bittensor's usual compute-and-incentive playbook at molecule screening instead of language models — and says it's now validating hits in a physical lab.
BitExplorer · Jul 26, 2026
Most of Bittensor's compute-heavy subnets point their incentive competition at training or serving AI models. Metanova Labs' Subnet 68, NOVA, points it somewhere else: screening molecules against disease targets.
In its H1 2026 recap, posted July 8, the team frames the milestone less as "decentralized drug discovery is possible" — which it says the network has already shown — and more as closing the loop from network-driven competition into durable incentives, and, notably, physical lab validation of what the network actually finds. In other words: the point isn't just that miners can screen candidate molecules faster through competitive incentives, it's that what they find is now being checked on a lab bench, not just on a leaderboard.
It's a useful reminder that "compute subnet" on Bittensor doesn't only mean GPUs training the next model — the same incentive structure is also being pointed at problems well outside AI, with real-world verification built into the loop.