Avalon 백서

Avalon

원문

bitpro가 원문에서 글자만 모은 내용이에요. 표·그림·수식은 빠지거나 깨질 수 있어요. 전체 8쪽 중 2쪽

ins?</summary> Yes, sUSDa can be bridged to other chains. Use our [Bridge Page](https://usda.avalonfinance.xyz/bridge/) </details> ## DeFi Lending <details> <summary>How Much Is the Liquidation Penalty?</summary> The liquidation penalty (or bonus for the liquidator) varies depending on the asset used as collateral.&#x20; Example Scenarios: • Example 1: Bob deposits 10 BTC and borrows the equivalent of 5 BTC in USDT. If Bob’s health factor drops below 1, his loan is eligible for liquidation. A liquidator can repay 2.5 BTC worth of USDT (50% of the borrowed amount). In return, the liquidator receives 2.5 BTC plus a 5% bonus of 0.125 BTC, totaling 2.625 BTC for repaying the debt. • Example 2: Bob deposits 5 BTC and 4 BTC worth of ETH, borrowing 5 BTC worth of USDT. If Bob’s health factor drops below 1, his loan is eligible for liquidation. A liquidator can repay 2.5 BTC worth of USDT (50% of the borrowed amount). Since the liquidation bonus for BTC is 15% and ETH is 5%, the liquidator chooses to claim BTC. The liquidator receives 2.5 BTC plus a 15% bonus of 0.375 BTC, totaling 2.875 BTC for repaying the debt. </details> <details> <summary>How to Avoid Liquidation</summary> To avoid liquidation, users should aim to keep their health factor high. Here are two ways to do this: 1. Deposit More Collateral: Adding more collateral to your position increases your health factor, lowering the risk of liquidation. 2. Repay Part of Your Loan: Making partial repayments increases your health factor more significantly than additional deposits. Regular repayments also help reduce the overall debt. To avoid liquidation entirely, keep your health factor above 2. This gives you a significant safety margin. Additionally, keep an eye on the price fluctuations of stablecoins like USDC, which may affect your health factor due to market conditions. For example, if USDC is priced at $0.95 instead of $1.00, it can cause your health factor to decrease, triggering liquidation risk. </details> <details> <summary>Can I Participate in the Liquidation Ecosystem?</summary> Yes, liquidations are open to anyone who wishes to participate, but the competition can be fierce. Most liquidators develop sophisticated bots or tools to get ahead and claim liquidation bonuses quickly. If you have the right strategy and tools, you can profit from liquidating undercollateralized loans. </details> <details> <summary>Why Use Avalon DeFi Lending?</summary> Avalon’s DeFi lending protocol stands out by offering a highly secure, efficient, and scalable lending platform. With our isolated pools, deep liquidity, and multichain deployment, we provide the best conditions for both borrowers and lenders. Our innovative approach to risk management ensures that your assets are safe, while our liquidation process offers opportunities for liquidators to participate in the ecosystem. Whether you’re borrowing BTC or BTC LSD tokens or looking to capitalize on lending opportunities, Avalon has the right solutions to meet your needs. </details> # A New Era of AI and Real-World Assets On-Chain Institutional-Grade Structured Token Mechanism for AI Compute Access Avalon Labs AI Marketplace is designed as an **open platform for AI model developers and GPU hardware providers**, offering a new paradigm for accessing compute resources. Traditional GPU rental models focus exclusively on leasing raw hardware capacity. By contrast, Avalon Labs introduces the next evolution: **AI-Model-as-a-Service (AI-MaaS)**. AI-MaaS delivers **ready-to-use** AI Models. The first model we introduce is **reinforcement-learning–optimized models** built on GPU infrastructure, allowing enterprises and developers to tap into **pre-trained, performance-optimized AI systems** without the complexity of hardware procurement, infrastructure management, or model training pipelines. This approach unlocks a new level of efficiency, scalability, and accessibility for AI adoption. ## Commercial Rights Tokenization (CRT) In parallel with AI innovation, Avalon Labs introduces **Commercial Rights Tokenization (CRT)**—a pioneering framework for bringing **real-world assets (RWAs)** on-chain. Unlike traditional tokenization models that focus on securitization, CRT isolates and tokenizes **commercial rights of access and service**. This enables commodities, services, and goods to be legally structured, accessed, exchanged, and monetized through blockchain infrastructure in a regulatory-compliant manner. CRT creates a scalable path for bringing tangible, utility-driven assets on-chain, establishing a **new standard for institutional-grade RWA integration**. ## Structured Marketplace Design Under the Avalon Labs AI Marketplace, we have developed a **Structured Marketplace** model that addresses volatility in compute demand by separating allocation into distinct service tiers: * **Senior Allocation (sAI)** – Provides **prioritized access** to AI-model rental time slots, offering more predictable availability and reduced exposure to volatility. Senior allocation ensures stability and service continuity for users with mission-critical compute needs. * **Junior Allocation (jAI)** – Offers **flexible access** to residual rental capacity once senior demand is met. While subject to greater variability, junior allocation provides the opportunity for cost advantages and greater upside during periods of strong market demand. This structured approach creates a **transparent, enforceable, and risk-adjusted framework** for accessing AI infrastructure, combining the **efficiency of DeFi architecture** with the **rigor of institutional finance**. # How does the Structured Marketplace Work? **Objective** To create a **regulated, transparent, and economically resilient token framework** that enables institutional and DeFi-native participants to access, trade, and stake rights to the AI reinforcement learning (RL) model rental time, via two structured layers: **Senior Allocation (sAI)** and **Junior Allocation (jAI)**. This mechanism builds on **commercial rights tokenization (CRT)** and **smart contract–enforced access**, while managing liquidity and risk with sophisticated marketplace structure design. ## **Senior Allocation (sAI) — Institutional-Grade Access Token** **Instrument Classification** * **Digital Receipt Token (DRT)** under **UCC Article 7**, representing enforceable rights to AI-model rental time * Structured as an **electronic document of title**, prioritized in allocation and redemption * Secondary classification under **Article 12 (CER)** for transfer control and enforceability #### **Token Features** <table><thead><tr><th width="188.95703125">Component</th><th>Specification</th></tr></thead><tbody><tr><td><strong>Token Type</strong></td><td>ERC-20 compliant receipt token (<code>sAI</code>)</td></tr><tr><td><strong>Minting Conditions</strong></td><td>KYC/KYB/AML verified participants Mintable via smart contract, subject to <strong>dynamic cap</strong> to ensure liquidity and value stability</td></tr><tr><td><strong>Liquidity Access</strong></td><td>Integrated with DEXs, with seed liquidity provisioned at launch</td></tr><tr><td><strong>Target Rate</strong></td><td>Anchored at <strong>$1 USD</strong> — soft peg enforced via VWAP-based liquidity management</td></tr><tr><td><strong>Redemption Policy</strong></td><td>Redemption processed <strong>monthly</strong>, subject to <strong>predefined redemption window</strong> and <strong>dynamic cap</strong></td></tr><tr><td><strong>Staking Option</strong></td><td>Users may stake <code>sAI</code> to receive <code>Staked sAI</code></td></tr><tr><td><strong>Staking Benefit</strong></td><td>Derived from <strong>net rental market rates</strong> generated by enterprise-level renters of AI models</td></tr><tr><td><strong>Unstaking Policy</strong></td><td>7-day cool-down period to prevent gaming and support redemption queue integrity</td></tr><tr><td><strong>Redemption Waterfall</strong></td><td>Seniority: <strong>first in line</strong> in all redemption scenarios, including liquidation</td></tr><tr><td><strong>Use Cases</strong></td><td>Enterprise-grade compute access, capital preservation, short-term liquidity needs</td></tr></tbody></table> #### **sAI Liquidity & Price Stability Management** <table><thead><tr><th width="160.03125">Market Phase</th><th width="164.8125">Trigger</th><th>Protocol Action</th></tr></thead><tbody><tr><td><strong>Expansion</strong></td><td>VWAP > $1</td><td>Surplus USDC from time slot rental is <strong>minted to sAI token holders</strong>, proportionally</td></tr><tr><td><strong>Contraction</strong></td><td>VWAP &#x3C; $1</td><td>USDC cash flow used to <strong>repurchase sAI from DEX</strong>, restoring peg before any new mints</td></tr><tr><td><strong>Partial Liquidation</strong></td><td>VWAP &#x3C; $0.95 (6-month average) + Model Owner Trigger</td><td>Underlying GPU/models assets partially liquidated to fulfill redemptions at $1 face value jAI layer absorbs NAV shortfall</td></tr></tbody></table> ## ## **Junior Allocation (jAI) — Volatility Absorber & High Market Rate Generator** **Instrument Classification** * **Residual receipt token** with exposure to **rental demand volatility**, serving as the system’s **risk absorption layer** * Defined as a **Controllable Electronic Record (CER)** under **UCC Article 12** * Structured to capture **surplus market rates**, while **subordinating claims** in redemption and liquidation **Token Features** <table><thead><tr><th width="245.26171875">Component</th><th>Specification</th></tr></thead><tbody><tr><td><strong>Token Type</strong></td><td>ERC-20 compatible <code>jAI</code> (transferable)</td></tr><tr><td><strong>Minting Conditions</strong></td><td>KYC/KYB/AML verified Dynamic cap linked to TVL of sAI and risk parameters</td></tr><tr><td><strong>Price Discovery</strong></td><td>Updated <strong>monthly</strong>, based on junior allocation and cumulative rent collection</td></tr><tr><td><strong>Redemption Policy</strong></td><td>Monthly redemption windows with dynamic pricing; subject to available surplus</td></tr><tr><td><strong>Liquidity Access</strong></td><td>No DEX liquidity Native <strong>OTC desk</strong> with <strong>limit order support</strong> via Avalon Labs</td></tr><tr><td><strong>Market Rates Mechanism</strong></td><td>Participates in residual rates after all senior obligations are met, higher potential market rates in peak demand periods</td></tr><tr><td><strong>Risk Absorption</strong></td><td>Absorbs all downside in rental demand shocks, hardware devaluation, or partial liquidation events</td></tr><tr><td><strong>Redemption Waterfall</strong></td><td>Second in priority; claims honored only after sAI fulfillment</td></tr><tr><td><strong>Use Cases</strong></td><td>venture-style exposure to AI infrastructure</td></tr></tbody></table> # Operation Overview ## **One-Stop Hub for AI Model Rental in DeFi** Artificial Intelligence (AI), one of the most transformative technologies of this century, is reshaping industries and redefining how humans work and interact. At the core of AI’s progress lies **GPUs and computing power**—the essential infrastructure that enables large language models (LLMs) and other advanced AI systems to operate at scale. Traditionally, the market has focused on **GPU rentals**, where enterprises rent raw compute resources. However, the next evolution is emerging: the rental of **ready-to-use AI models** built on top of GPU infrastructure. This approach allows users and enterprises to access optimized, pre-trained models that deliver sophisticated outcomes—without the need to manage underlying hardware or training pipelines directly. At Avalon Labs, we are committed to bridging advanced AI infrastructure and blockchain technology, creating innovative ways to integrate real-world AI and compute into on-chain markets. ## **Structured Marketplace Innovation** The structured marketplace is an **open platform** for all types of AI models and GPU hardware. Any AI model developers and GPU owners can access the Avalon Labs AI Marketplace after receiving approval. **As a core competitive advantage,** Avalon Labs collaborates with industry leaders to integrate the reinforcement learning (RL) model and other advanced machine learning techniques as the first launching product. Rather than presenting the RL model itself as a product, we deploy RL- based optimization to enhance the performance of models (including LLM fine-tuning) and maximize GPU utilization efficiency. This transforms the GPU hardware rental into **AI-Model-as-a-Service (AI-MaaS)**: * Enterprises and developers gain access to fine-tuned models and compute capabilities. * RL optimization improves accuracy, efficiency, and outcomes across LLM and other AI workloads. Avalon Labs will leverage blockchain infrastructure to tokenize this new approach, delivering it via the **Avalon Labs AI Marketplace**. To address the inherent volatility of GPU and AI-model rental markets, we introduce the **Structured Marketplace**, a system designed to reallocate risk across different levels of participants. ## **Allocation Framework** The marketplace operates with two allocation categories: * **Senior Allocation** – Prioritized access to AI-model rental time slots, with more predictable access and lower exposure to volatility. Users benefit from stability in compute/model access. Utility token: **sAI**. * **Junior Allocation** – Flexible access, activated after senior allocation demand is met. This allocation carries higher variability but offers greater upside potential. Utility token: **jAI**. ## **Operational Principles** The Avalon Labs AI Structured Marketplace follows the principles of GPU rental markets while integrating blockchain-native features for transparency, compliance, and safety. Operations follow three steps: 1. **AI-Model Rental Time Allocation & Market-Based Pricing** * Allocation priority defined by senior vs junior access tiers * Dynamic market-based rental fee rates in USDC 2. **Risk Management** * Allocation tiers absorb volatility differently (senior prioritization vs junior flexibility) * Transparent rules embedded in smart contracts 3. **Liquidity & Redemption** * Users can enter/exit allocations with predictable redemption logic * Market-driven mechanisms ensure healthy liquidity # AI-Model Rental Time Allocation & Market-Based Pricing Avalon Labs AI Marketplace introduces a next-generation marketplace model that transcends traditional GPU rental by integrating **reinforcement learning (RL)-enhanced AI models**. Working with leading industry partners, Avalon Labs has developed in-house RL optimization frameworks that can be deployed on GPU infrastructure. This integration transforms raw GPU rental into **AI-Model-as-a-Service (AI-MaaS)**, enabling users to access models that are not only compute-efficient but also deliver higher accuracy and performance for large language models (LLMs) and other advanced AI workloads. This approach marks a paradigm shift: instead of merely providing access to hardware, the marketplace offers **ready-to-deploy, RL-optimized models** that reduce training costs, improve inference efficiency, and create outsized impact across industries. ## **Commercialization Tokenization Framework** Unlike legacy real-world asset tokenization, which often relies on **securitization before tokenization**, Avalon Labs AI Marketplace introduces **Commercial Rights Tokenization (CRT)**. CRT brings AI-Models and compute power directly on-chain as enforceable digital property rights, enabling a dynamic marketplace for decentralized participants. Our framework draws upon well-established commercial law principles to ensure that ERC-721 NFTs represent the commercial right to administer senior or junior pools. The sAI and jAI tokens representing senior and junior allocations of AI-model rental time slots are legally enforceable and compatible with traditional financial and commercial systems. ## **UCC Article 7 – Documents of Title** Provides legal enforceability for custody and transfer, comparable to warehouse receipts or bills of lading. ERC-721 NFTs represent the commercial right to administer senior or junior pools. **sAI** and **jAI** tokens, which represent rights to compute/model rental time slots allocations, can be structured as **electronic documents of title** under Article 7, ensuring that token holders’ rights are recognized in both on-chain and off-chain legal frameworks. ## **UCC Article 12 – Controllable Electronic Records (CERs)** Specifically designed for digital assets, NFTs, and tokens, Article 12 defines the concepts of **“control” and priority rights**. This enables **sAI** and **jAI** holders to have legally enforceable ownership of compute hours and AI-model rental rights, even in contested or bankruptcy scenarios. ## **UCC Article 9 – Secured Transactions** If **sAI** or **jAI** tokens are pledged as collateral, Avalon Labs AI Marketplace can **perfect security interests** under Article 9. This provides a regulated pathway for credit markets to form around tokenized compute and AI-model rentals, opening opportunities for secured lending and secondary liquidity. By grounding ERC-721 NFTs, sAI and jAI in UCC Articles 7, 12, and 9, Avalon Labs AI Marketplace establishes a **commercially enforceable pathway** for tokenized compute and AI-Models. This framework not only strengthens legal certainty but also sets a new benchmark for **real-world asset (RWA) tokenization** in the AI and blockchain industries. ## **Structured Marketplace Allocation** AvalonLabs.AI introduces a **Structured Marketplace** to address volatility and supply-demand imbalance in GPU and AI-Model rental markets. The marketplace separates participants into two allocation tiers, redistributing risks and ensuring efficient utilization: * **Senior Allocation (sAI)** – Holders of **sAI** receive **priority access** to rental time slots, with secured and predictable compute availability. This allocation is designed for enterprises and institutions requiring stability and guaranteed service levels. * **Junior Allocation (jAI)** – Holders of **jAI** participate flexibly, absorbing market volatility, uncertainties, and idle capacity. While carrying greater risk, the junior allocation benefits from potentially lower entry costs and higher upside when demand surges. This dual-structured approach introduces financial engineering principles into compute markets, without compromising compliance, allowing risk-adjusted access to AI infrastructure. ## **Key Participants** * **RL Model Owner**: Avalon Labs and its AI partners, developing and deploying reinforcement learning–optimized AI models. * **Renters**: AI startups, enterprises, research institutions, and universities that require high-performance AI models or compute capacity. * **Data Centers**: Custodians of GPU infrastructure, responsible for power, cooling, connectivity, and security. * **Senior Allocation Participants (sAI Holders)**: Compute/model access buyers seeking stable and prioritized service levels. * **Junior Allocation Participants (jAI Holders)**: Compute/model access buyers seeking flexible, market-driven access with greater volatility. * **Commercialization Tokenization Agent**: The legal entity that operationalizes the CRT process, tokenizes rental rights, and ensures enforceability of compute/model access on blockchain networks. ## **Flow of Funds** **Commercial Rights Tokenization (CRT)** To ensure legal and operational independence, ownership of the AI models and GPU hardware will be vested in a **bankruptcy-remote Special Purpose Vehicle (SPV)**. This SPV is a wholly owned subsidiary, managed by an independent manager, whose sole function is to maintain bankruptcy remoteness and safeguard the underlying assets. When buyers purchase **Senior Allocation (sAI)** or **Junior All
프로젝트로 돌아가기