Kamino AUTO — tokenised auto loans and a leveraged loop
Data as of 27 Sep 2026.
Verdict: AUTO itself is usable with limits; a max-leverage loop is a bet on a single issuer markdown. Kamino's 24.77% banner can't be reproduced from public data: AUTO itself accrues ≈8.0% a year (share price in Kamino's oracle: 7.95% over 83 days, 8.12% over 60). Looping to the max gets ≈16–17.6%; at launch leverage (2.9×), ≈12–13%. The rest of the banner is the launch rate (9.07%) and/or rewards not visible in the data. The core risk is a price that hasn't dropped a single day: loan defaults will only show up when the issuer marks the pool down — in one step.
What it is
AUTO is a Hastra token on Solana representing a share of a pool of "near-prime" consumer auto loans (below prime, above subprime) originated by Agora Data (Texas) and structured by Figure through Figure Forge / Democratized Prime. Launched 29 Jul 2026. Kamino runs a dedicated "AUTO Market": AUTO is collateral (LTV 0.78, liquidation threshold 0.80), and borrowers can take USDC, PYUSD and wYLDS (Figure's T-bill token). Since 21 Sep 2026 there's "Fixed Rate Multiply": borrow wYLDS at a fixed 5.3% for 30 days, auto-rolled "subject to available liquidity".
AUTO supply — Solana RPC getTokenSupply, 35.44M; held in the Kamino reserve — 31.84M (89.8%), Kamino API, 27 Sep 2026.
Findings
1. The oracle price is a model accrual, not a market 🔴
AUTO's oracle price didn't fall a single day from 6 Jul to 27 Sep (84 data points, +0.006–0.034% a day). Loan defaults will only reach the price when the issuer writes them down — in one step. Historical parallel: in 2007 CDO NAVs sat flat while the ABX index was already marking them down; the gap was the signal. For a loop this is what matters: at 4.5× liquidation hits on a 2.5% drop in AUTO's price, at 2.9× on 18%.
2. Almost all AUTO sits inside Kamino 🔴
35.44M AUTO issued, 31.84M (89.8%) in the Kamino reserve. The only market outside Kamino is one Orca AUTO/PRIME pool: $3.58m liquidity, $1.37m daily volume (DexScreener, 27 Sep). In a week Kamino deposits grew from $9.4m to $32.5m, and the reserve cap was lifted from $10m to $60m.
3. The fixed rate is already sold out
The fixed-rate wYLDS reserve (flat curve at 5.16%) is 100% borrowed: $10.01m of $10.01m, $98 free. New entries go into the variable-rate wYLDS reserve at 5.75% — $3.0m free there (that's the "$3.02m capacity" on the banner). Rolling the fixed rate "subject to available liquidity" means: no liquidity — the loan goes variable; not enough of that either — Kamino repays the debt itself by selling collateral (Solana Compass, 21 Sep 2026).
4. Yield is below the pitch
9.07% at launch (Solana Compass, 29 Jul), "~8.6% at start" (Cointelegraph), realised 7.95–8.12%. Hastra charges a 0.5% platform fee. Stated collateralisation is ≈2.4× cash-flow coverage (Cointelegraph), but no delinquency or default report for the pool was found in public sources.
5. Exit — "no price impact", on paper
Kamino advertises entry and exit "without price impact" because the vault is wired to Figure's mint and redemption (Solana Compass). Figure Forge: a token holder can redeem "whenever they want" — into loans, fiat or stablecoins (Crowdfund Insider, 21 Mar 2026). No source gives redemption timelines or queues.
The AUTO loop: what it pays and when it liquidates
Loop yield = AUTO yield + (leverage − 1) × (AUTO yield − borrow rate). AUTO yield = realised 8.0%.
| Leverage | Borrow at 5.30% (fixed, now sold out) | Borrow at 5.75% (variable, $3.0m free) | Debt ÷ collateral | Liquidation on AUTO drop | |
|---|---|---|---|---|---|
| No loop | 1× | 8.0% | 8.0% | 0 | — |
| Launch setting | 2.9× | 13.1% | 12.3% | 0.655 | −18.1% |
| 4.0× | 16.1% | 14.8% | 0.750 | −6.3% | |
| Max at LTV 0.78 | 4.5× | 17.6% | 16.0% | 0.780 | −2.5% |
Reserve parameters — Kamino API, 27 Sep 2026. 24.77% doesn't come out even at 9.07% AUTO yield and 4.5× (≈22.4%).
Loop entry and exit costs aren't established. Routed through the Orca pool at ≈0.2% per side, the cost applies to the full leveraged size, so it grows with leverage; through Figure redemption with no price impact, as claimed, close to zero (self-reported, unverified).
Funding the loop with a loan against other collateral means two liquidations stacked on top of each other and a five-layer stack: outside collateral → stablecoin loan → AUTO loop → Hastra token → Democratized Prime pool → Agora Data loans. The bottleneck is the variable rate on the first loan: stablecoin borrow rates on lending markets jump sharply once reserve utilisation nears 100%, and the loop's 2–3 pp spread can flip negative within days. Every layer adds its own failure point; four or more layers is a common line past which a DeFi structure is considered a no-go.
What would prove this wrong
- A one-step pool markdown. Any down day in AUTO's price in Kamino's reserve history means the "model accrual" read is broken, and loops above 2.9× become a bet on a single issuer move.
- An Orca–oracle gap. AUTO/PRIME on Orca trading more than 1% below the oracle two days running — the market sees what the NAV doesn't (the ABX mechanism).
- Forecast for 27 Oct 2026: AUTO's oracle price doesn't drop a single day, and 30-day realised growth stays within 7.5–8.5% annualised.
- Forecast for 27 Oct 2026: the fixed-rate wYLDS reserve in the AUTO market stays ≥ 95% utilised. A miss means fixed-rate capacity isn't the bottleneck.
What we don't know
- Pool reporting (delinquencies, defaults, coverage) — not found in public sources.
- Redemption timelines and queues for AUTO via Figure.
- Kamino's loop fees, and where the 24.77% comes from — rewards or a different rate.
- How near-prime auto loans in this specific pool behave in a recession: it has two months of history.
Who it suits and when it breaks
- Plain AUTO, no leverage — ≈8% on private credit with a redemption mechanism that's clean on paper but no public default reporting. Breaks on a pool markdown or a halt in Figure redemptions.
- A loop up to 2.9× — ≈12–13% with 18% of room before liquidation; for private credit with a smooth price, a real but not unlimited buffer.
- A 4–4.5× loop — 15–17.6%, but liquidation at −2.5…−6.3%: one pool markdown wipes the position.
Research for information only. Not investment advice.