TAO shed 12.70% on Wednesday, dropping to $221.07, as elevated derivatives activity underscored event-driven volatility across the ecosystem. The slide came the same day the Conviction v2 staking upgrade reached devnet-ready status, introducing decaying locks designed to push for longer-term commitment without permanent token bondage.
Price Rout and Derivative Action
The June 3 selloff was notable for the burst of derivatives trading that accompanied it. Open interest surged while the spot market absorbed the hit, a pattern that points to leveraged positions being unwound rather than a simple panic dump. The move extended a stretch of pressure on the token that started in April when Covenant AI, a major subnet operator, exited by selling roughly 37,000 TAO of α tokens. That single event sparked a sharp selloff and reignited governance urgency across the network, with the community still debating how to handle large unlock events.
Conviction v2 Hits Devnet
While the market reeled, the technical side of the network took a clear step forward. Developers merged pull requests #2687 and #2696, moving Conviction v2 to a devnet-ready state. The new system introduces decaying locks — meaning stakers commit tokens for a rolling period but the lock gradually weakens over time, set at about 648,000 blocks, roughly a 60-day half-life. That's meant to replace the current all-or-nothing lockup model with something that rewards patient holders without permanently tying up capital.
Mainnet PR #2643, which would actually deploy the upgrade on the live chain, remained open but blocked as of late May. The holdup wasn't spelled out, but the gap between devnet and mainnet suggests there's still internal testing or community approval to get through before the new staking logic goes live.
What the New Staking Model Aims to Fix
A SubnetRadar snapshot taken around the same time captured the scale of the existing system. About 4.58 million α tokens were locked, with roughly 4.14 million of those counted as conviction. Only 16 active lockers held those positions, and the top convict — subnet 79, known as MVTRX — controlled 1.27 million α alone. That concentration has been a concern: a handful of big wallets can dominate governance and staking rewards, making the network vulnerable to coordinated exits.
The decaying-lock design tries to broaden participation by lowering the mental cost of committing tokens. Instead of a permanent bond, users can stake with a known decay schedule. The idea is that subnet operators, the groups that actually run the AI and compute subnets, will be less likely to flip their tokens after a short burst of rewards. The upgrade's backers hope it dampens what they call "mercenary churn" — participants who stake just long enough to harvest incentives and then bail.
The Broader Demand Question
But staking mechanics alone won't fix the network's deeper challenge. The core question hanging over TAO is whether subnets can generate durable, paid demand — for inference, data feeds, compute routing — that feeds value back to the native token beyond the short-lived hype cycles that have historically driven AI rotations. Covenant AI's departure rattled confidence precisely because it showed how fast a major demand source can vanish.
The upgrade doesn't directly address that demand side. It smooths the supply schedule by making staking less rigid and discouraging quick flips, but it doesn't create new buyers or subnet customers. If the subnets themselves fail to attract paying users who need actual computation or AI inference, the token's price will remain sensitive to every derivatives event and big unlock.
For now, the market is watching two things: whether the mainnet Conviction v2 PR gets unblocked and passes, and whether any of the top subnets can land real revenue contracts. The next few weeks will tell if the network can turn technical progress into something that actually stabilizes the token price.
