Charts
DataOn-chain
VIP
Market Cap
API
Rankings
CoinOSNew
CoinClaw🦞
Language
  • 简体中文
  • 繁体中文
  • English
Leader in global market data applications, committed to providing valuable information more efficiently.

Features

  • Real-time Data
  • Special Features
  • AI Grid

Services

  • News
  • Open Data(API)
  • Institutional Services

Downloads

  • Desktop
  • Android
  • iOS

Contact Us

  • Chat Room
  • Business Email
  • Official Email
  • Official Verification

Join Community

  • Telegram
  • Twitter
  • Discord

© Copyright 2013-2026. All rights reserved.

简体繁體English
|Legacy

USD.AI (CHIP) Project Report

CN
Hotcoin
Follow
4 hours ago
AI summarizes in 5 seconds.

1. Project Overview

USD.AI is a permissionless decentralized lending protocol focused on providing financing services for AI infrastructure. The protocol allows GPU operators to tokenize their hardware assets as collateral and obtain financing instantly. The core token of the protocol is CHIP, with a token standard of ERC-20, deployed on the Arbitrum chain, while also supporting cross-chain circulation on Ethereum and Base chains through LayerZero's OFT standard. The total supply of CHIP is 10 billion tokens, with a circulating supply of approximately 2 billion tokens (20% of total supply) as of the time of writing. CHIP was launched for trading on April 21, 2026. Currently, CHIP is listed on several mainstream exchanges, including Binance, Coinbase, Upbit, Bybit, OKX, KuCoin, Gate, MEXC, and Robinhood, with perpetual contract trading available on platforms like Binance, OKX, Bitget, and Hyperliquid. The protocol is developed by the Permian Labs team, and the USD.AI Foundation was officially established as the off-chain management entity of the DAO in January 2026. Investors include well-known institutions such as Coinbase Ventures, DCG (Digital Currency Group), Dragonfly, and Framework Ventures, and the project was selected for the NVIDIA Inception program.

2. Project Introduction

USD.AI is positioned in the "InfraFi" track, with core narrative: The AI industry is undergoing an unprecedented capital expenditure cycle, and GPU hardware, as the core production material for AI infrastructure, produces real and sustainable cash flow, but the credit speed of the traditional financial system cannot match the lifecycle rhythm of GPU hardware. GPU hardware depreciates about 20% annually, with an effective lifespan typically around three years, while traditional banks have a credit approval period that can last 12 to 24 months; even flexible private credit funds require 6 to 12 months to complete underwriting. This structural mismatch means that by the time traditional financing is in place, the GPU hardware may have already passed a significant portion of its lifecycle.

USD.AI aims to address this issue. The protocol builds a bilateral credit market: borrowers (AI infrastructure operators, including new cloud computing companies, data center operators, etc.) obtain financing using GPU hardware and the cash flow it generates as collateral; depositors (capital providers) earn returns from the interest generated by real GPU loans by depositing stablecoins, rather than receiving token subsidies or transaction fees. Since its launch in 2025, the protocol has reportedly executed over $225 million in loans and approved financing limits exceeding $1.2 billion, with borrowers including newly listed cloud computing companies and institutions from North America, Europe, Australia, Latin America, and Asia. As of now, the protocol has total deposits of approximately $348 million, with over 74,000 active users and an active loan pipeline of about $236 million.

The project likens itself to the "Fannie Mae moment" of the GPU credit market – just as the housing market only achieved true scalability after tradable mortgages emerged, the AI infrastructure market also needs a liquid debt product to define its benchmark interest rate. USD.AI's goal is to become the on-chain benchmark interest rate for GPU-backed credit.

3. Products and Technology

The product system of USD.AI revolves around three core layers: deposit layer, lending layer, and governance layer.

In the deposit layer, the protocol offers two core assets. The first is USDai, a synthetic dollar fully backed by PYUSD (a compliant stablecoin issued by PayPal, with underlying assets in U.S. Treasury bonds and cash equivalents), pegged 1:1 to the dollar. Users can mint USDai by depositing PYUSD, and USDai itself does not generate income, serving as a stable unit of account and medium of exchange. The second is sUSDai, the staked version of USDai, which is the core profit tool of the protocol. Users receive sUSDai by staking USDai, with its yield coming from two parts: the interest paid by GPU infrastructure borrowers and the returns generated by idle reserves invested in U.S. Treasury bonds. Income is automatically accumulated through the increasing exchange rate between USDai and sUSDai. The current annualized return of the protocol is approximately 7.09%, with an expected annualized return of about 13.51%.

In the lending layer, AI infrastructure operators apply for financing through GPUloans.com. The loans are structured as non-recourse (the only recourse being the collateral itself, but there is flexible recourse against the borrower’s company entity in cases of fraud or malicious acts), with a loan-to-value ratio of 70%-80% and a repayment cycle of once every 30 days, with principal amortized over 3 years. Each loan's borrower operates through a bankruptcy-remote SPV incorporated in Delaware, legally isolating GPU assets from the overall balance sheet of the operator. Complete legal documents include loan and collateral agreements, UCC-1 financing statements, SPV agreements, pledge agreements, limited guarantees, and data center lien waivers. The loan process is divided into four stages: escrow establishment and funding injection (with Wilmington Trust acting as an independent escrow agent), OEM server construction and delivery, hardware installation and verification, and escrow release and fund routing.

In terms of technical architecture, USD.AI adopts a hybrid structure of on-chain and off-chain. The off-chain layer provides real-world legal enforceability: comprehensive lien rights, SPV structure, bank accounts, custodianship, and insurance. The on-chain layer offers transparency and programmability: loan NFTs (ERC-721) represent collateral positions, on-chain records serve as the authoritative lender registry, and smart contracts automate payments and reserve withdrawals. The two layers synchronize at each important event node. sUSDai is implemented based on ERC-4626 standard (deposit) and ERC-7540 standard (redemption), with redemption adopting a FIFO queue mechanism over a 30-day cycle. The protocol uses Chainlink as the price oracle and achieves cross-chain token transfers through LayerZero.

The protocol has also designed several innovative mechanisms. CALIBER (Collateral Architecture for Legally Interoperable, Bankruptcy-remote, Enforceable Rights) is its legal and tokenization framework, transforming physical GPU hardware into on-chain programmable collateral through a four-layer structure: physical asset layer (UCC Article 7 pledge agreement), SPV isolation layer, ERC-721 token layer, protocol integration layer. QEV (Queue Extractable Value) is its market-driven redemption mechanism, which uses a 30-day cycle and ZK privacy bidding auction to transform withdrawal pressure into an orderly, incentive-aligned process, eliminating bank run risks, with stayers earning extra returns from auction fees.

In terms of security, the protocol has completed multiple rounds of audits, including reports from Cantina (May 2025), ktl (May, October, November 2025, March 2026), and Quantstamp (February 2026), and maintains a continuous Bug Bounty program on the Cantina platform. Risk mitigation measures include: a 70%-80% LTV ratio providing equity buffers, a debt service reserve account (DSRA) covering about 3 months of peak repayment amounts, property insurance provided by Alliant, hardware value reinsurance provided by Barkr, real-time hardware monitoring provided by Aravolta, and the ability of ITAD partners to recover and resell hardware in the event of default.

4. Economic Model

The total supply of CHIP tokens is 10 billion, with 27.5% allocated for ecosystem bootstrapping to guide liquidity for the protocol, covering two core goals of yield initiation and capital formation. Of this allocation, the first 10% has already been distributed in the first season (The Allo Game), and the remaining portion will be used for airdrops, incentive programs, and other growth initiatives.

The reserve accounts for 19.5%, designated for future grants, partnerships, R&D, etc.

The allocation for core contributors is designated for builders and the operations team of Permian Labs. The lock-up structure is: 0% unlocked in the first 12 months, a one-time release of 33% in the 12th month, with the remaining 67% unlocking monthly in equal amounts over the subsequent 24 months.

The allocation for investors reflects the CHIP commitments of early supporters. The unlocking structure is consistent with core contributors: 0% unlocked in the first 12 months, 33% unlocked in the 12th month, and the remaining 67% unlocking monthly in equal amounts over the subsequent 24 months.

The functional positioning of CHIP includes two main dimensions: governance and staking. In terms of governance, CHIP holders vote on-chain to determine the core parameters of the protocol, including collateral standards (which GPU hardware qualifies, LTV ratio, depreciation schedule), interest rate parameters (benchmark interest rate, tiered rate adjustments), loan entry standards (minimum collateral coverage, acceptable purchase contract structures, geographical restrictions, borrower qualifications), protocol fee parameters (origination fees, net interest margins, QEV fees), and protocol upgrades and integrations. In terms of staking, CHIP can be staked as sCHIP, with the staking module assuming a backing function for the protocol – in the event of a loss, the staked CHIP may be used to cover the gap, making stakers proactive participants in the protocol's risk framework. Unstaking requires a cooling-off period set by governance.

The protocol's revenue model is based on loan origination fees and net interest margins.

5. Team and Investors

USD.AI is developed by Permian Labs. According to official public information, David Choi serves as the CEO of USD.AI, and Ivan Sergeev serves as the CTO of Permian Labs. The official documents mention that Ivan Sergeev worked at 21.co (the company later transformed under the leadership of Balaji Srinivasan and Lily Liu) and experienced the challenges of aligning early hardware financing with operational scalability. Permian Labs’ GitHub organization is called metastreet-labs, suggesting that the team may have connections or evolution relationships with the previous MetaStreet protocol (an NFT lending protocol).

The USD.AI Foundation was officially established on January 27, 2026, as the off-chain management entity of the USD.AI DAO, responsible for legal infrastructure, DAO treasury custody, and ecosystem coordination. The foundation's actions are guided by token holder governance and do not directly operate the protocol.

In terms of investment, the institutional supporters showcased on the official website include Coinbase Ventures, DCG (Digital Currency Group), Dragonfly Capital, and Framework Ventures. Additionally, the project was selected for the NVIDIA Inception program and established deep partnerships with PayPal (PYUSD), with PayPal providing an incentive program of up to $1 billion at an annualized rate of 4.5%. Disclosed borrower partners include QumulusAI, Sharon AI, and Quantum SKK, with approved financing limits exceeding $1.2 billion.

6. Roadmap

According to official public information, the development history and planning of USD.AI is roughly as follows:

In 2025, the protocol officially launched operations, completed the initiation and execution of the first GPU-backed loans, with a total transaction volume of sUSDai exceeding $7.7 billion for the year. In the same year, multiple rounds of smart contract audits from Cantina and ktl were completed, and the first season incentive program The Allo Game was initiated.

In January 2026, the USD.AI Foundation was officially established, and CHIP tokens were officially announced. In the first quarter of 2026, the ICO, airdrop, and TGE were completed, and the Quantstamp audit was finished. On April 21, 2026, CHIP officially launched for trading, concurrently starting Flatiron (Allo Game second season), with all rewards distributed through CHIP airdrops, running until October 14, 2026.

In terms of recent planning, the sCHIP staking function has been launched. The official stated that more practical features will be added to CHIP in the coming weeks. On the lending business side, the focus is on scaling execution – expanding borrower bases, deepening institutional partnerships, and growing real income from actual hardware. The direct minting and redeeming of USDai will be restricted to KYC-verified whitelisted market makers and institutions in the second quarter of 2026 (but holding, transferring, staking, and secondary market trading of USDai and sUSDai will always remain permissionless). The complete auction functionality of the QEV mechanism is also under development.

7. Risks and Opportunities

In terms of risks, first is the risk of GPU hardware depreciation and technological iteration. GPU hardware depreciates approximately 20% annually, and AI chip technology iterations occur very rapidly, with the release of new generations of hardware potentially leading to accelerated decreases in the value of older models as collateral, affecting the loan collateral coverage ratio. While the protocol mitigates this through Barkr's value reinsurance and conservative LTV ratios, a systemic decline in collateral value in extreme technological transition scenarios remains a risk to watch.

Secondly, there is liquidity mismatch risk. The yield of sUSDai comes from GPU loans with a 3-year amortization period, which are fundamentally illiquid assets. Although the QEV mechanism aims to alleviate redemption pressure, a 30-day redemption cycle might result in depositors facing longer wait times under extreme market conditions. The protocol explicitly states that it will not prematurely liquidate loans to meet withdrawal requests.

Thirdly, there is borrower concentration risk. Currently disclosed major borrowers include QumulusAI, Sharon AI, and Quantum SKK, with the $1.2 billion approved limit concentrated among a few institutions. If major borrowers encounter operational difficulties or defaults, it could significantly impact the protocol.

Fourthly, there is regulatory and compliance risk. The protocol involves the tokenization of real-world assets, cross-border lending, stablecoin issuance, and other regulatory-sensitive areas. While the protocol has adopted SPV structures, KYC/KYB verification, and other compliance measures, the regulatory policies across global jurisdictions remain rapidly evolving, presenting uncertainties.

Fifthly, there is smart contract and technology risk. Even though the protocol has completed multiple rounds of audits and maintains a Bug Bounty program, the complexity of the on-chain and off-chain hybrid architecture increases potential attack surfaces. Multi-signature operations, cross-chain bridging, oracle dependencies, and other components all carry technical risks.

Sixthly, there is risk in the token economic model. CHIP currently has a circulating supply of only 20% of the total, and the unlocking of tokens for core contributors and investors in the future (starting after 12 months) could lead to significant sell pressure. Furthermore, CHIP does not grant holders direct distribution rights to protocol revenues; its value capture logic primarily relies on governance rights and staking backstop functions, and long-term value support remains to be validated.

In terms of opportunities, first is the structural growth of the track. Capital expenditures for AI infrastructure are in the midst of a historic expansion cycle, with huge and ongoing demand for GPU financing. The traditional financial system has notable structural disadvantages in this area, providing an entry point for on-chain lending protocols.

Secondly, there is a narrative of real yields. Unlike most DeFi protocols that rely on token incentives, USD.AI's yield comes from real GPU loan interest and Treasury bond yields, which maintain a strong competitive advantage in the current market environment.

Thirdly, there is an institutional-level partner network. Support from top institutions such as Coinbase Ventures, DCG, and Dragonfly, combined with deep integration with PayPal PYUSD (including a $1 billion incentive program), as well as selection for the NVIDIA Inception program, provide strong resource backing and business expansion capability for the protocol.

Fourthly, there is broad DeFi ecological integration. USDai and sUSDai have been integrated into mainstream DeFi protocols such as Pendle, Euler, Morpho, Aave, Curve, Balancer, Fluid, and Gearbox, providing a solid basis for composability and liquidity.

Fifthly, there is a first-mover advantage. The on-chain credit market backed by GPUs is still in its early stages, and USD.AI has established certain barriers in legal framework, underwriting processes, and partner networks.

8. Summary

USD.AI (CHIP) is a decentralized lending protocol focused on financing for AI infrastructure, constructing a bilateral credit market connecting the financing needs of AI operators with the capital supply of DeFi by tokenizing GPU hardware into on-chain collateral. The protocol has a high level of completion in product design, integrating the legal framework of traditional structured finance (SPV, UCC liens, custodial agents) with the transparency and programmability of DeFi, forming a unique hybrid on-chain and off-chain architecture. CHIP, as a governance token, grants holders voting rights on core parameters of the protocol and the ability to participate in risk backing through staking but does not directly share protocol revenue. The project has garnered investment from multiple leading institutions and established substantial partnerships with PayPal, Chainlink, and others. Currently, the protocol operates real loan businesses and generates income, which is relatively rare in the RWA and DeFi lending tracks. However, factors such as GPU hardware depreciation risk, borrower concentration, liquidity mismatch, token unlocking pressure, and regulatory uncertainty remain core variables that require ongoing attention.

About Us

Hotcoin Research, as the core research institution of the Hotcoin Exchange, is committed to transforming professional analysis into your practical tools. We analyze market trends through “Weekly Insights” and “In-depth Research Reports”; leveraging our exclusive column “Hotcoin Selection” (AI + expert dual screening), we pinpoint potential assets and reduce trial-and-error costs. Each week, our researchers face you live through live broadcasts, interpreting hot topics and predicting trends. We believe that warm companionship and professional guidance can help more investors navigate through cycles and seize the value opportunities of Web3.

Risk Warning

The cryptocurrency market is highly volatile, and investments themselves carry risks. We strongly recommend that investors make investments based on a thorough understanding of these risks and under a strict risk management framework to ensure capital safety.

Website: https://www.hotcoin.com/zh_CN/learn/index/

X: x.com/Hotcoin_Academy

Mail: labs@hotcoin.com

免责声明:本文章仅代表作者个人观点,不代表本平台的立场和观点。本文章仅供信息分享,不构成对任何人的任何投资建议。用户与作者之间的任何争议,与本平台无关。如网页中刊载的文章或图片涉及侵权,请提供相关的权利证明和身份证明发送邮件到support@aicoin.com,本平台相关工作人员将会进行核查。

|
|
APP
Windows
Mac
Share To

X

Telegram

Facebook

Reddit

CopyLink

|
|
APP
Windows
Mac
Share To

X

Telegram

Facebook

Reddit

CopyLink

Selected Articles by Hotcoin

4 hours ago
Hotcoin Research | Shell Change or Evolution? The Reconstruction and Upgrading from MakerDAO to Sky
6 days ago
Hotcoin Research | Re-staking High Yield Bubble Bursting? Analyzing the Systemic Risks of LRT from the Kelp DAO Theft Incident
6 days ago
Asteroid Shiba (ASTEROID) Project Report
View More

Table of Contents

|
|
APP
Windows
Mac
Share To

X

Telegram

Facebook

Reddit

CopyLink

Related Articles

avatar
avatarHotcoin
4 hours ago
Hotcoin Research | Shell Change or Evolution? The Reconstruction and Upgrading from MakerDAO to Sky
avatar
avatarMGBX Global
1 day ago
From Top 63 to Top 40: The Rise of MGBX is More Than Just a Ranking Story.
avatar
avatarMatrixport
1 day ago
BIT Research Report | AI Investment Enters "Realization Period" Which Targets Are Creating Real Profits?
avatar
avatarCoinW研究院
3 days ago
CoinW Research Institute Weekly (April 20, 2026 - April 26, 2026)
APP
Windows
Mac

X

Telegram

Facebook

Reddit

CopyLink