Sai: The Core Perpetual DEX Built on Nibiru

Key Insights

• Sai is a perpetual decentralized exchange (perp DEX) built on Nibiru, a high-throughput Layer-1 with a unified EVM and Wasm execution layer.
• The perp DEX has proven to be the dominant instrument for crypto exposure, enabling leverage trading of assets. As of Dec. 4, 2025, cumulative volume on perp DEXs has grown 653.3% year-to-date (YTD) to $11.3 trillion.
• Sai utilizes decentralized oracles to power its markets and uses Sai Liquidity Positions (SLP) vaults as a GLP-style liquidity basket, where a single, multi-asset pool backs multiple perp markets at oracle-derived prices rather than constant-ratio AMM curves. This design enables the listing of both crypto and non-crypto assets, provided reliable external price feeds are available.
• Sai’s roadmap focuses on expanding real-world asset markets, integrating more DeFi deployments, and delivering a CEX-like account experience with gasless trades, automated strategies, a crypto-based savings account, multichain USDC deposits that allow users to fund directly from networks like Ethereum and Base, a native mobile application, and improved fiat and cross-chain funding.
• Maker capital (liquidity provided by depositors) is pooled into single-asset SLP vaults that earn a share of trading fees and take the other side of trader profit and loss (P&L). By choosing which SLPs to fund, makers effectively decide which markets receive liquidity and are able to launch and scale on Sai.

Primer

Sai is a perpetual decentralized exchange (perp DEX) that aims to simplify, predict, and make leveraged trading more accessible. Users connect to the application with standard EVM wallets, deposit collateral such as USDC or stNIBI, and trade perp markets through an interface that resembles a centralized futures exchange. They can open long or short positions, set up to 100x leverage on certain assets, and use familiar order types, including market, limit, stop, and conditional (stop-loss/take-profit) orders.

Sai was created by a team with experience in distributed systems, infrastructure, and financial protocols. The Sai team is led by Nibiru founder and CEO Unique Divine, who has a background in applied mathematics and machine learning, as well as prior experience at IBM and Sommelier Protocol.

Sai is built on Nibiru Chain, a high-throughput Layer-1 with a unified EVM and Wasm execution environment. Nibiru combines an EVM-equivalent runtime (Nibiru EVM) and a Wasm environment in a single state machine, allowing Solidity and Rust contracts to coexist, share accounts and gas, and call each other through built-in system contracts that bridge the two VMs. Its consensus layer, Nibiru BFT, is an evolution of CometBFT, delivering fast finality and sub-two-second settlements. Alongside core modules for staking, governance, IBC, and a native oracle, this architecture provides Sai with the low-latency execution, deterministic settlement, and oracle support necessary to offer a CEX-like onchain derivatives platform. Sai’s defining characteristic is its approach to pricing and risk: execution is anchored to decentralized oracles and backed by pooled vaults called Sai Liquidity Positions (SLPs), emphasizing consistent behavior across different market regimes.

Why Sai exists?

Perpetual decentralized exchanges (Perp DEXs) have emerged as the dominant instrument for crypto market participants. Their rise has coincided with a structural shift away from centralized exchanges, driven by events in 2022 during which FTX, Celsius, and Voyager lost user funds. As of Dec. 4, 2025, cumulative volume on perp DEXs has risen 653.3% year-to-date (YTD) from $1.5 trillion to $11.3 trillion.

However, the user experience has not evolved at the same pace as volume. New users often encounter steep learning curves related to margin, funding, and liquidation. Casual traders often discover leverage through interfaces that expose them to complex mechanics without clear guardrails. Even experienced traders can face execution failures when volatility spikes.

Traditional AMM curves and thin orderbooks are inherently sensitive to local liquidity conditions. When markets are stressed, execution quality deteriorates, as users require reliability. Sai exists to address this structure gap: to provide a perpetual venue where execution is anchored to external markets, risk behavior is more predictable, and idle capital is optimized for capital efficiency.

Oracle-Based Execution vs. Traditional AMMs and Orderbooks

Most perp DEXs rely on one of two execution models. Orderbook-based venues maintain a central limit order book (CLOB) where traders post bids and asks, and trades execute when orders cross. Execution quality depends on resting depth at each price level and can deteriorate quickly when liquidity dries up or is concentrated on a few participants. AMM-based perp DEXs instead derive prices from pool balances and bonding curves. While this can simplify liquidity provision, it often produces large price moves when big trades hit the pool and can diverge significantly from external prices during volatility.

Sai’s execution model utilizes oracle-based pricing with AMM-backed liquidity. For each market, Sai maintains an index price sourced from decentralized oracles that aggregate data from major exchanges. When a trade is executed, the protocol starts from this oracle price and then applies a price impact function that depends on the current open interest (OI) and the size of the trade relative to the configured market depth.

The price impact function tracks OI over time and by direction (long or short), along with a depth parameter for each pair, known as One Percent Depth, which represents the trade size required to move the price by 1%. Price impact is calculated as:

Price Impact = (Start OI + Trade Size / 2) / One Percent Depth

Where open interest and trade size are measured on the relevant side of the market. The resulting percentage adjusts the oracle price upward for long positions and downward for short positions.

Sai’s execution model offers several advantages. By anchoring to external prices, Sai reduces the influence of local manipulation and thin-orderbook dislocations. For traders, the practical effect is that fills and liquidations tend to reflect global market behavior more than the momentary quirks of a single venue’s orderbook or pool.

Features of Sai

Built-in Protection for Traders

Sai is designed around the concept of “protection by default.” It does not remove the inherent risks of leverage, but it seeks to make those risks clearer and less tied to edge-case microstructure failures.

Risk management is built on three pillars:

• Oracle-anchored pricing: Execution and liquidation reference the same oracle-linked index price, reducing the chance that a temporary local deviation triggers unwanted liquidations.
• Transparent margin and liquidation logic: Each position is governed by an initial margin requirement and a maintenance margin threshold. A position’s equity is defined as collateral plus unrealized profit & loss (P&L), net of borrowing fees and estimated closing costs. When equity falls below the maintenance level, the position becomes eligible for liquidation. A keeper, an external automation agent, can then call the TriggerTrade function to close the position at the current oracle price adjusted by price impact.
• Borrowing fees that respond to imbalances: borrowing fees (Sai’s funding mechanism) accrue continuously based on time, open interest, and directional skew. Markets are organized into borrowing groups that share parameters such as base fee per block, maximum open interest, and an exponent that accelerates fees as long/short imbalances increase. Only the side with higher net open interest pays borrowing, which nudges positioning back toward neutrality.

In practice, this means traders see an estimated liquidation price in the interface, understand how their borrowing costs behave, and can rely on liquidations being triggered by sustained movements in the broader market.

If a trader’s collateral is insufficient to cover realized losses and all applicable fees, the protocol still records the fees owed in its internal ledgers for governance, referrers, and trigger service providers, but no additional funds are taken from the trader. The remaining loss is absorbed by the protocol vault, a fee-funded reserve that backstops the system up to configured limits.

Yield on Idle Capital

Most trading platforms leave idle margin unused. Sai instead allows collateral that is not actively deployed in positions to earn yield through the Sai Savings Account (SSA). Idle stablecoins on Sai can be deposited into SSA to earn yield sourced from protocol activity and associated strategies, and can be redeposited as trading collateral at any time.

This makes capital on Sai more efficient. Active traders can keep accounts funded without fully sacrificing return on idle balances, while yield-focused users can earn passively and remain one step away from deploying that capital into leveraged positions.

Sai Liquidity Positions (SLPs)

Sai’s liquidity is provided by Sai Liquidity Positions (SLPs), which are single-asset vaults that aggregate collateral and underwrite all markets denominated in that asset. If BTC, ETH, and SOL perpetuals all use USDC as collateral, they draw liquidity from the same USDC SLP vault. Depositors who provide USDC to that vault receive SLP share tokens that represent a pro-rata claim on the vault’s assets and accrued fees. These shares can be redeemed subject to the vault’s withdrawal mechanics and risk parameters.

SLPs are effectively the counterparty to all trader P&L. When traders profit, the vault pays out their gains. When traders lose, those losses are accrued to the vault and increase the value of the remaining SLP shares. In addition to net trader P&L, SLPs receive a portion of the trading and liquidation fees generated on the platform. Over time, this allows SLP providers to earn from both flow and mispriced risk.

Liquidity providers effectively decide which pairs can support deeper liquidity and higher open interest. This makes SLPs a gating resource for new markets, including future synthetic equities, commodities, and other real-world assets.

Asset Coverage Beyond Crypto

Sai’s oracle-based architecture can extend beyond crypto-native assets. As long as a robust and reliable price feed exists, the protocol can, in principle, support a perp market for that asset. Over time, this allows Sai to offer more than just crypto major pairs and can also include markets for equities, indices, commodities, and other real-world assets.

Supporting more assets and utilizing reliable price feeds has two important implications. First, it allows existing perp users to consolidate more of their trading into a single onchain venue, rather than hopping between platforms for different asset classes. Second, it provides traders with an onchain entry point to traditional markets without needing to learn entirely new pricing behavior.

Liquidation Cascades and Sai’s Design

On Oct. 10, 2025, crypto markets experienced one of the sharpest deleveraging events in their history, a textbook liquidation cascade. A single macro shock, a surprise 100% tariff announcement from President Donald Trump, triggered heavy selling and over $19 billion of liquidations from leveraged positions. As prices fell, exchanges began closing positions to protect their own solvency. Market makers cut risk and withdrew liquidity, which thinned out orderbooks, and forced selling to push prices lower again. In some venues, internal spot markets were used to value collateral, and wrapped assets depegged, driving further liquidations for users who had never touched those markets. Auto-deleveraging systems then started closing other traders’ positions to plug bad debt. The result was a domino reaction driven by excess leverage, collateral design, and microstructure rather than by a single price move.

Sai is designed to reduce these specific failure modes. All positions are governed by explicit initial and maintenance margin requirements, and the interface surfaces an estimated liquidation price for each trade. Liquidations are triggered only when equity, defined as collateral plus unrealized profit and loss (P&L) net of borrowing and expected closing fees, falls below the maintenance level. When that threshold is breached, a keeper can call the TriggerTrade function to close the position to the oracle-linked index price adjusted by the protocol’s price impact function, not at a potentially distorted local orderbook level. Collateral is valued via external price feeds rather than thin in-venue spot books, and Sai does not use auto-deleveraging to socialize losses to profitable traders. If losses and fees exceed a trader’s collateral, the remaining shortfall is absorbed by the protocol vault, a fee-funded reserve with predefined limits.

Other Key Features on the Roadmap

In addition, Sai’s 2026 roadmap includes several initiatives designed to expand access, improve UX, and deepen integrations:

• Refined trading interface: Ongoing improvements to keep the UI simple for new users while providing fast, efficient execution for active traders.
• DeFi integrations: Connecting with other DeFi applications for swaps, routing, and using SLP positions as collateral through an earn-focused interface.
• Automated strategies: Vaults that run predefined or user-managed strategies on Sai so users can gain exposure with a single click.
• Accounts: A CEX-like account experience with gasless trades, multichain balances, and support for fiat deposits.
• Multichain USDC deposits: Users can fund accounts directly with USDC from Ethereum, Base, and other supported networks without manual bridging or asset swaps. This ensures collateral arrives in the intended denomination.
• Gas-abstracted execution: Sai will not require users to hold NIBI to submit trades or manage positions.
• Integrated fiat on-ramps: Native fiat on-ramps allow users to fund accounts via card or bank transfer, reducing reliance on external exchanges and lowering onboarding friction for first-time leveraged traders.
• Mobile experience: A dedicated mobile application that includes core trading flows and portfolio management.
• Comprehensive tooling: Robust APIs and developer tools for building applications and integrations on top of Sai.
• Data platform: Access to backtesting, historical data, custom order types, and strategy automation tooling for more systematic users.

Closing Summary

Sai is being built for a segment of the market that is already large and continues to evolve. Perp DEXs have reached $11.3 trillion in cumulative volume as of Dec. 4, 2025, but most venues still rely on execution models that are highly sensitive to local liquidity conditions. Sai offers an alternative approach: it anchors pricing to decentralized oracles, applies a transparent price impact function tied to open interest and depth, and routes trades through SLP vaults that systematically underwrite P&L. Combined with explicit margin rules, responsive borrowing fees, and a savings layer for idle capital, the platform is designed to make perp behavior more predictable across market regimes while remaining fully onchain and DeFi-native.

Sai's opportunity is to translate this architecture into a durable market share. Success will depend on its ability to grow deep SLP liquidity, responsibly extend asset coverage into equities, commodities, and RWAs, and execute on a roadmap that includes gasless accounts, automated strategies, savings products, and mobile access. If Sai can attract and retain maker capital, sustain organic trading flow, and maintain disciplined risk controls through volatility, it is positioned to become a meaningful venue in the perp DEX landscape and a core piece of infrastructure for onchain leverage across both crypto-native and traditional assets.