Table of Contents
A couple weeks ago, we covered the $1.12B credit opportunity sitting in locked alpha that nobody had built yet, citing Siam Kidd's Alpha Collateral thesis as the structural answer. Forge, created by the Endure Network (SN 30), is the first protocol to ship.
Antonio Verrico
Forge is the first native lending market for TAO on Bittensor. You post your alpha as collateral, borrow TAO against it, and your subnet exposure stays intact. Lenders go the other way and deposit TAO into the pool, earning yield above what root staking pays, secured by alpha collateral whose risk parameters are set by a competing miner network. That miner network is Endure, a risk intelligence subnet where miners compete to produce the parameters Forge runs on.
Risk Becomes a Bittensor Output
Every lending market needs LTV ratios, interest curves, and liquidation thresholds. Most pick those numbers through a governance forum or an internal risk team and update them when the market forces a correction. Forge picks them through a competing miner network on Endure that submits new parameters every epoch and gets scored against what happened in the market that round.
When a miner submits a parameter set, validators score it against the previous epoch's outcomes. The question is whether the LTV held through alpha volatility, whether the interest curve attracted borrowing without underpricing the credit risk, and whether liquidations triggered early enough to keep lenders whole. The miners that get those answers right earn more emissions next epoch, the ones that get them wrong lose share, and there is no path to defending a model the team committed to last quarter because the team is not setting parameters in the first place.
Bart Hillerich
Therefore, the risk layer updates on the same timescale the market moves. Alpha volatility shifts across the network, and the parameter sets that priced it correctly take more share. Liquidity tightens in one subnet's alpha, and the miners that saw it coming gain ground. The model that wins on Tuesday is not the model that won on Monday.
This is the first Bittensor lending market where the risk model itself is a subnet output rather than a static configuration, and the rest of the design follows from that single choice.
The Third Yield Option for TAO Holders
If you hold TAO today, your yield options sit at the extremes. Root staking pays the network's baseline rate and carries almost no downside. The other option is staking your TAO into a subnet's alpha, where the return potential is higher, but you are now exposed to whatever happens to that alpha's price.
Forge gives you a third curve in the middle. You deposit TAO into the lending pool and earn yield from borrowers paying to access TAO liquidity against their alpha collateral. Your principal stays denominated in TAO, which means a 40% drawdown in one subnet's alpha price does not move your position. You are taking credit risk instead of alpha price risk, and if a borrower's collateral drops below the liquidation threshold, the protocol liquidates that position to make you whole.
That distinction matters more than the yield headline. An alpha-staker is concentrated in one subnet's outcomes and absorbs the full price swing in either direction. A Forge lender is exposed to system-wide borrowing demand across every subnet whose alpha gets posted as collateral, with liquidation thresholds the miner network keeps tuned to current conditions. The risk profile is diversified, and the yield premium over root is earned by carrying real credit risk rather than swinging on alpha volatility. The position unwinds when you want your TAO back.
The yield itself is set by the same miner network producing every other parameter. Interest curves move with utilization, so as borrowing demand climbs, lenders earn more, and when the pool sits underutilized, rates drop until borrowers come back. Nobody at Forge picks the rate. The number that wins is the one that brings in the most capital at the lowest sustainable risk.
— Endure Network (@EndureNet) June 8, 2026
Endure: Why DeFi Risk Intelligence Belongs on a Decentralized Network
Safety Alone Doesn't Win Emissions
Most lending protocols hold their risk team to one metric: don't blow up. Endure treats that as table stakes and scores its miners on something more useful.
The win condition is the equilibrium between TVL and revenue. Set the LTV too conservatively, and capital sits idle, borrowers leave, lenders earn nothing, and the protocol generates no fees. Push it too aggressively, and a single drawdown triggers liquidations that hit lenders directly. The miners earning the most emissions are the ones who keep finding the precise point where the protocol attracts the most capital at the lowest sustainable risk.
That trade-off is impossible to optimize from the outside. A salaried risk analyst gets the same paycheck whether the protocol generates $10M in interest or $100M, but Endure's miners are paid in emissions that scale with how well their parameters perform, which is the same number the protocol cares about.
60% of Forge Revenue Buys Back Alpha
The revenue model closes the loop. 60% of all Forge fees go straight to buying back the subnet's alpha token at the protocol level, with no team discretion in the middle.
When Forge's TVL grows, borrowing grows with it, interest revenue follows, and alpha buy pressure climbs on the same curve. There is no separate cash stream getting captured by an LLC and no carve-out for management. The same dollar that pays the lender funds the buyback.
For you as an alpha holder, the alignment is straightforward. Whatever grows the protocol's top line grows the price support under your token at the same rate, and the mechanism lives in the contract rather than in a roadmap.
The Lending Market Is a Data Source for Endure
The visible milestone is Forge launching. The strategic move underneath it is everything Endure does with the data Forge generates.
A lending market is the highest-stakes test environment for risk intelligence anywhere. Real capital gets liquidated when parameters miss, real revenue gets generated when they fit, and every epoch produces ground-truth feedback that no simulated stress test can replicate. The miners who score well on Forge are scoring well in an environment where their mistakes cost real lenders real money.
That feedback is what Endure plans to sell next. Funds running external capital are obvious buyers, because they need risk signals they did not produce themselves. Market makers carrying alpha collateral need liquidation thresholds that update with the assets on their books, not parameters fixed last quarter. Protocols outside Bittensor would pay for parameter feeds Forge is already generating internally, so they can read risk from a source that has skin in its own outputs.
The thesis is that a network producing risk intelligence against live capital outperforms a research team writing quarterly reports, and Forge is the live capital that gets the thesis tested. The longer Forge runs the more outcome data Endure accumulates, and the more defensible Endure becomes as the layer financial institutions read risk from. Every borrow on Forge is an epoch of training data for the next product Endure ships.
Risk Is a Subnet Now
Forge produces its risk through a competing miner network, scores those parameters against live outcomes every epoch, and converts 60% of protocol revenue into direct buy pressure on the alpha token that funds the network. The lending market is the visible product, and the intelligence layer underneath it is what Endure plans to scale.
You have three TAO yield options now: root staking for the baseline, alpha staking for the upside, and Forge for a yield curve produced by a market that keeps getting smarter. The proving ground is open.
Read the litepaper here.
Disclaimer: This article is for informational purposes only and does not constitute financial, investment, or trading advice. The information provided should not be interpreted as an endorsement of any digital asset, security, or investment strategy. Readers should conduct their own research and consult with a licensed financial professional before making any investment decisions. The publisher and its contributors are not responsible for any losses that may arise from reliance on the information presented.
