Ethereum co-founder Vitalik Buterin published a research proposal on June 1 to replace collateralized debt positions — the architecture underpinning approximately $31.5 billion in DeFi lending deposits across Aave and MakerDAO alone — with an options-based framework that eliminates forced liquida...
"What if we use options as the base of DeFi, instead of CDPs and liquidations?" — Vitalik Buterin, Ethereum Co-Founder, EthResearch post, June 1, 2026
Ethereum co-founder Vitalik Buterin published a research proposal on June 1 to replace collateralized debt positions — the architecture underpinning approximately $31.5 billion in DeFi lending deposits across Aave and MakerDAO alone — with an options-based framework that eliminates forced liquidations entirely. Within 10 days, at least two independent implementations reached testnet, including one physically settled prototype on Base with verified contract addresses.
The proposal arrives in a market where CoinGlass data recorded $150 billion in crypto liquidations during 2025, with a single two-day cascade in October 2025 exceeding $19 billion. Oracle manipulation accounted for 13% of DeFi exploits in 2025, according to multiple security auditors. Buterin's design removes both failure points — real-time oracle dependency and liquidation cascades — by construction, though unresolved questions around rebalancing costs and oracle manipulation resistance remain.
The speed of implementation signals market demand for structural alternatives. Whether the options-based model can scale beyond testnet to absorb meaningful TVL remains the open question for DeFi lending in H2 2026.
Collateralized debt positions have served as DeFi lending's foundational primitive since MakerDAO launched DAI in December 2017. The mechanism is straightforward: borrowers lock collateral, mint or borrow assets against it, and face forced liquidation if collateral value drops below a threshold. The system requires real-time price feeds and immediate execution — two properties that create systemic fragility.
The cost of this architecture is measurable. According to CoinGlass's 2025 annual report, crypto derivatives markets processed approximately $150 billion in forced liquidations during 2025, averaging $400–500 million per day. The October 10–11, 2025 deleveraging event — the largest single liquidation cascade in crypto history — saw volumes exceed $19 billion in reported liquidations within 24 hours. CoinGlass estimates the actual figure approached $30–40 billion when accounting for disclosure timing gaps across platforms.
On the DeFi lending side specifically, Aave absorbed $196 million in bad debt following the KelpDAO rsETH exploit in April 2026, which triggered a broader $13 billion decline in DeFi TVL over two days, according to CoinDesk reporting. The Terra/Luna collapse of May 2022 — the most severe CDP-adjacent failure — erased approximately $42 billion when the algorithmic stablecoin UST lost its peg, triggering a cascade that contributed to the broader crypto market declining from $2.9 trillion to $798 billion by year-end 2022.
The pattern is consistent: CDP systems amplify volatility rather than absorbing it. When prices fall, liquidations generate selling pressure, which drives further price declines, which triggers additional liquidations. The mechanism that is supposed to protect protocol solvency instead becomes a transmission vector for systemic risk.
The June 1 EthResearch post, titled "Building index-tracking assets on top of options instead of debt," proposes a structural alternative. Instead of collateral-backed debt, the system splits one unit of a base asset (e.g., 1 ETH) into two paired option claims — designated P and N — that always sum back to the original asset. Because the payoffs are complementary and fully backed, there is no collateral ratio to maintain, no threshold to breach, and therefore no liquidation to trigger.
The P claim tracks upside exposure to a reference index (denominated as a variable T, which could represent USD/ETH, CPI, or other measures). The N claim captures the residual. At maturity, settlement requires only a single price determination — not a continuous feed — which allows the system to operate on slow, dispute-friendly oracles similar to those used by prediction markets.
The design is mathematically straightforward. For any price outcome, the sum of P and N equals the original deposited asset. This holds regardless of price direction or magnitude. There is no scenario in which the system becomes undercollateralized because the collateral is never borrowed against — it is split into complementary claims.
This structure enables synthetic index-tracking assets (effectively stablecoins or basket-tracking tokens) without the liquidation infrastructure that current protocols require. Buterin framed the proposal as an extension of his September 2025 blog post on "low-risk DeFi," arguing that such frameworks are necessary for Ethereum's sustainable economic infrastructure.
The proposal's most notable characteristic is the speed at which it moved from research post to running code. By June 11, Buterin posted on X: "Looks like the options thing is happening already," pointing to multiple teams building implementations in the EthResearch forum thread.
Cleave launched as a testnet options exchange, positioning itself as "DeFi's missing third pillar, alongside Uniswap for spot and Hyperliquid for perpetuals." Cleave's implementation splits any asset into upside exposure and a cash floor — fully backed, with no margin, no funding rates, and nothing to liquidate. The protocol operates as a direct expression of Buterin's no-liquidation architecture.
A separate developer deployed a physically settled version of the P/N construction on Base — Coinbase's Layer 2 network — running the complete lifecycle: minting paired claims from WETH, transferring them, exercising against a USDC strike, and settling the vault. The prototype used physical settlement (moving underlying assets) rather than price-based settlement, which removes the oracle from the critical path entirely. The deployment included verifiable contract addresses and transaction hashes.
A third team posted a perpetual, liquidity-backed variant that was also operational.
Buterin responded to the pace of development with a caution: "I do strongly urge that if any of these get on mainnet quickly, we formally verify it first." He tagged the Vyper team and formal-verification developers, consistent with his broader advocacy for AI-assisted formal verification as a standard for deploying production smart contracts.
A significant portion of the proposal's value proposition targets oracle infrastructure — the system of price feeds that DeFi protocols depend on to determine collateral values and trigger liquidations in real time.
Oracle manipulation ranked as the second most common DeFi attack vector in 2024, accounting for $52 million in losses across 37 incidents, according to security firm data. In 2025, oracle-based attacks comprised 13% of DeFi exploits, with over 31% of early-2025 DeFi losses attributed to oracle manipulation.
The vulnerability was illustrated by the Polymarket weather incident in April 2026, in which a trader allegedly manipulated a Paris weather sensor with a hair dryer to generate a $34,000 gain — demonstrating single-data-source vulnerability in prediction markets that also applies to DeFi oracle architecture.
Buterin's design addresses this by eliminating real-time oracle dependency. CDP-based liquidations require fast price feeds reporting current asset prices accurately and instantly, with no margin for error and no dispute window. Every failure in this chain — whether from latency, manipulation, or outage — can trigger incorrect liquidations or create arbitrage opportunities for MEV bots.
The options-based model settles at maturity using a single price determination, which can incorporate dispute resolution windows similar to optimistic oracle designs. Trading speed for security, the system becomes substantially harder to exploit via flash loan attacks or transient price manipulation.
The proposal is not without identified limitations, several of which Buterin acknowledged.
Rebalancing costs. Options-based positions tied to indices require periodic rebalancing, and every rebalance executes a trade. On Ethereum mainnet, gas costs and slippage could erode gains, particularly for smaller positions. This friction may limit the model's applicability to retail users unless deployed primarily on Layer 2 networks where transaction costs are lower.
Oracle manipulation at settlement. While the design removes continuous oracle dependency, the single settlement price determination at maturity creates a concentrated attack surface. If an attacker can manipulate the price feed at the moment of settlement, the impact is magnified relative to a system that samples prices continuously. The design of manipulation-resistant slow oracles remains, by Buterin's own acknowledgment, an open problem.
Liquidity fragmentation. Splitting assets into P and N claims creates new trading pairs that require independent liquidity. Whether sufficient market depth will develop for these derivative claims — particularly for less liquid base assets — is uncertain.
Behavioral adoption. The CDP model, for all its risks, is conceptually simple: deposit collateral, borrow assets, manage your ratio. The options-based model requires users to understand paired claim mechanics, maturity dates, and rebalancing requirements. Protocol interfaces would need to abstract this complexity significantly for mainstream adoption.
Existing protocol inertia. Aave ($23.5 billion TVL as of early 2026), MakerDAO ($8.02 billion TVL), and Morpho ($11 billion in deposits following a recent $175 million funding round at $2 billion valuation) have substantial infrastructure, integrations, and governance structures built around CDP architecture. Migration paths are unclear.
The proposal arrives at a specific moment in DeFi's development. Total DeFi TVL stood at approximately $128.6 billion in early 2026, down from peaks above $180 billion. The KelpDAO exploit in April 2026 triggered $13 billion in TVL outflows within 48 hours, demonstrating the system's continued fragility.
Existing on-chain options protocols — Panoptic, Opyn, and Premia — have operated for years but have not achieved the TVL or usage of lending protocols. Panoptic's "streamia" model (streaming premiums based on Uniswap v3/v4 liquidity activity) and oracle-free design share conceptual DNA with Buterin's proposal, but target derivatives trading rather than foundational lending infrastructure.
The distinction matters. Buterin's proposal does not position options as a trading product but as a base-layer primitive for constructing stablecoins, index-tracking assets, and other synthetic instruments that currently depend on CDP architecture. If successful, it would represent not an incremental improvement to DeFi lending but a replacement of its structural foundation.
Whether Cleave or similar implementations can attract meaningful TVL on mainnet — and whether formal verification can proceed at the speed of development — will determine whether the options-based model remains a testnet experiment or becomes a viable alternative to the $150 billion liquidation cycle.
The options-based DeFi proposal represents the most structurally significant challenge to CDP architecture since MakerDAO's 2017 launch. The design eliminates two of DeFi lending's most persistent failure modes — liquidation cascades and real-time oracle dependency — through mathematical construction rather than parameter tuning.
The rapid testnet implementations demonstrate both developer interest and the relative simplicity of the core mechanism. The harder questions — rebalancing economics, settlement oracle security, and the migration path from established protocols — remain open.
For the DeFi lending market, the next 6–12 months will test whether options-based architecture can move from formal verification to mainnet deployment with meaningful TVL. The $150 billion annual liquidation cycle provides the economic incentive. Whether the technical and adoption challenges can be resolved at comparable speed will determine the outcome.