USD.ai STEP application

DAO Programs & Initiatives STEP Applications & Updates
1

Due Diligence Questionnaire Response – USD.ai

All figures are current as of March 20, 2025, unless stated otherwise. This application recommends a cap of $250,000 or the minimum required for STEP candidacy, whichever is higher.

Applicant information

Name: USD.ai

Address (Headquarters): MetaStreet Foundation, Clara Formations (Cayman) Limited, PO Box 10061, George Town Financial Center, 90 Fort Street, Suite 306

City, State, Postal Code: Grand Cayman, KY1-1001, Cayman Islands.

Country: Cayman Islands

Website: https://usd.ai

Primary contact name: Glenn Kennedy

Title: Foundation Director

Country: Cayman Islands

Email, Telegram, Forum, & other methods of contact

Expected yield

All of the yield generated by the Current Plans will be through M^0 as the Treasury Bill yield provider. Their separate application can be read here: M^0 - STEP Application

We recommend the STEP Committee to first underwrite the M^0 application as the yield generated by USDai will be $M until it evolves into Stage 2 (where it diversifies into other crypto-native yield, through loans against hardware) in the later future.

Current plans:

  • sUSDai is the staked version of USDai. It will earn a yield based on a staking ratio, the base rate is currently 4.15% APR. We conservatively estimate a staking ratio of about 50%, thus the sUSDai yield is estimated to be about 8.30% APR.
  • This is reinformed with points attributed to the depositors (1x for sUSDai and 5x for USDai) which amplifies the yield generated from the same stablecoin as non-yield generating positions will receive points in lieu of yield generated by their capital.

Future plans

  • The yield derives from a two-stage collateral structure:
  • Stage 1 (Initial phase): 100% M^0 Treasury Bills (T-bill) exposure.
  • Stage 2 (Post-Q2 2025): Diversify into onchain loans against DePIN assets targeting 20-30% utilization. The remaining collateral (70-80%) will remain in TBills via M^0.

1593×1191 111 KB

Expected maturity: N/A

Underlying asset

  • Stage 1: Fully collateralized by M^0 Treasury Bill assets.

  • Stage 2: Transition to hardware-backed lending (AI GPUs, DePIN assets), target of 20-30% diversification.

Minimum/Maximum transaction size

No minimum transaction size. No maximum transaction size.

Current AUM for products

  • USD.ai is currently based on two protocols with AUMs of $282m of collateral.

  • M^0 has $152m of collateral

  • MetaStreet Protocol has $130M of collateral

  • USDai has lined up a minimum of $30m of new deposits into its launch in late April.

Current AUM for issuer

  • M^0 has $158m of issued collateral
  • MetaStreet Protocol has $130M of collateral

Volume of transactions LTM

  • M^0 has had $200m of LTM volume
  • MetaStreet has had $221M of LTM volume (including LP volume)

Source of first-loss capital

Stage 1: M^0 has a 3% equity buffer on account of maximum 97% ratio of minted $M to collateral

Stage 2: 42% overcollateralization buffer for risk mitigation (70% LTV). Insurance provider: Evertas (insures 20% of Bitcoin hash rate globally).

3. Basics and Background

1. How will this investment improve Arbitrum’s RWA ecosystem?

USD.ai is a yield-bearing synthetic dollar designed to initially scale Arbitrum DeFi, and then AI infrastructure RWAs and DePIN, uniquely positioned to integrate Arbitrum’s DeFi ecosystem with crypto-native RWAs. Unlike traditional RWAs that focus on tokenizing passive income streams, USD.ai actively generates yield through both Treasury Bills and broader DePIN economic activity. Its architecture aligns with Arbitrum’s vision of scaling DeFi by bringing real, sustainable financial flows to the network. The Arbitrum Foundation has already recognized this potential by awarding USD.ai a foundation grant, reinforcing the alignment between both ecosystems.

Stage 1 of its deployment focuses on building a strong TVL base through M^0 utilization, ensuring that the system has the capital required to scale, diversify, and onboard new protocol participants. Users deposit assets to mint USD.ai, which is initially used to generate T-Bill yield. In the future (months later), this liquidity is then utilized against loans backed by DePIN infrastructure and services, regardless of the chain they are on. This creates an organic demand loop where DePIN providers borrow USD.ai to fund operations while liquidity providers earn yield from the revenue generated by various forms of decentralized infrastructure. The process mirrors the way traditional financial markets finance large-scale infrastructure development, but instead of relying on slow-moving institutional capital, it leverages DeFi’s agility to create a high-velocity growth engine.

This model directly benefits Arbitrum by introducing a new category of real-world yield generation that is deeply integrated into the protocol’s deep DeFi offering, like a Wall Street for crypto activity. DePIN operators use USD.ai as working capital, deploying it into networks that provide services such as decentralized storage, connectivity, and energy distribution, all of which generate revenue that cycles back into Arbitrum. This positions Arbitrum as a key financial hub for DePIN by first facilitating native DeFi participation through lower-risk T-Bill-backed stability in Stage 1 and then expanding into diversified lending against productive on-chain and off-chain assets in Stage 2. By embedding DePIN financing into its ecosystem, Arbitrum strengthens its position as the leader in RWA integration while ensuring sustainable liquidity flows into its DeFi infrastructure.

Beyond its initial liquidity bootstrapping, USD.ai serves as a direct onboarding mechanism for DePIN operators, bridging the gap between decentralized infrastructure and DeFi. Many DePIN projects operate in isolated ecosystems with fragmented liquidity, often reliant on inefficient funding mechanisms. USD.ai changes this by bringing DePIN operators into Arbitrum’s DeFi ecosystem, exposing them to its lending protocols, liquidity markets, and yield opportunities. This ensures that Arbitrum does not just passively benefit from DePIN growth but actively becomes its financial backbone. Unlike any other blockchain initiative, this is a structured and intentional onboarding pipeline that connects real-world infrastructure providers with DeFi participants, creating a reinforcing cycle of growth. Tether followed a similar model as they themselves provided loans to Bitcoin miners as they expanded their capital base.

The alignment between USD.ai and Arbitrum is clear: both are designed to scale high-value economic activity while maintaining capital efficiency. The Arbitrum Foundation’s support underscores this shared vision, recognizing that infrastructure-backed assets represent a natural evolution of the RWA narrative. With USD.ai’s initial phase focused on building deep liquidity through M^0, followed by the gradual expansion into multi-sector DePIN integration, Arbitrum is positioned as the central hub for this emerging sector. This investment is not just about another synthetic asset—it is about establishing Arbitrum as the premier blockchain for real-yield infrastructure financing, unlocking an entirely new class of financial primitives that will drive long-term growth across both ecosystems.

In summary:

  1. USD.ai establishes on-chain credit markets for AI and DePIN by using liquid collateralized credit tokens (LCTs).
  2. Stage 1 ensures full stability with 100% T-bills via M^0 contracts.
  3. Stage 2 brings scalable DePIN & AI infrastructure-backed credit.
  4. This strengthens Arbitrum’s RWA ecosystem by introducing yield-bearing, capital-efficient stable assets.

2. Identify key management personnel and individual experience. Also include third parties utilized for managing assets and their qualifications.

David Choi - Co-founder, CEO

David Choi has a background in MEV, DeFi investing, and traditional finance, as a seed investor in Arbitrum-based projects such as XAI and Maverick, along with other decentralized finance protocols, spending over 8 years in the crypto industry. He is an ETHGlobal Hackathon winner, with experience in smart contract development and on-chain trading strategies. Prior to his work in crypto, he worked in investment banking at Deutsche Bank. He holds a degree from the University of Southern California (USC).

Conor Moore - Co-founder, COO

Conor Moore has experience in private equity, real estate finance, and investment banking, having worked at Rockpoint, a real estate-focused private equity firm, and Eastdil Secured, a real estate investment bank. He also worked in investment banking at Deutsche Bank, where he focused on capital markets and structured finance. His background includes evaluating institutional investments and executing large structured deals. He graduated from University of Maryland

Ivan Sergeevi - Co-founder, CTO

Ivan Sergeev has a background in quantitative finance, algorithmic trading, and blockchain infrastructure. He previously worked at DRW, where he focused on high-frequency trading, and at 21.co, a company originally specializing in ASIC mining technology, which was later acquired by Coinbase. His expertise includes smart contract architecture, financial modeling, and hardware-based mining solutions, culminating into 10 years in the crypto industry. He holds a Bachelor & Masters degree in Electrical Engineering from MIT.

Other experience by team members: Synthetix, Balancer, Nansen, P2P validator, Magic Eden, and others.

Third-Party Participants:

  • M^0: T-bill collateral manager
  • (Stage 2 only) Evertas: Insurance provider, insures 20% of Bitcoin hash rate globally.
  • (Stage 2 only) MetaStreet Protocol: Structured finance layer for DeFi lending

3. Describe any previous work by the entity or its officers/key contributors similar to that requested. References are encouraged.

  • MetaStreet Protocol: Engineered DeFi lending models for RWAs and other low liquidity models.
  • NodeFi: $200M+ in TVL at the peak, first major DePIN lending platform.
  • Tactical Compute: Co-founded $40M AI compute infrastructure fund regulated by Abu Dhabi Global Markets with Aethir, a >$1bn FDV Arbitrum-native compute protocol.

4. Has your entity or its officers/key contributors been subject to an enforcement action, criminal action, or defaulted on legal or financial obligations?

None.

5. Describe any conflicts of interest for your entity and key personnel.

None.

6. Insurance coverages, guarantees, and backstops

  • Insurer: Evertas (institutional risk underwriting, insures 20% of Bitcoin mining hash rate globally). This only applies to Stage 2.

7. Historical tracking error in your proposed product, or similar to that being proposed Product 2024 2023 2022 2021

  • N/A

8. Please describe any experience your firm has in working with decentralized organizational structures

USD.ai is a decentralized protocol and has had extensive amount of governance done through Snapshot voting.

9. What is your entity’s current assets under management, assets held in trust, total value locked, or equivalent metric for your legal structuring?

  • M^0 has $158m of issued collateral by Minters
  • MetaStreet Protocol has $130M of collateral across various types (RWAs, Arbitrum-native Nodes, NFTs).

10. How many of these assets held are present on Arbitrum One, if any?

The Permian Labs team pioneered the first-ever onchain DePIN financing deal, setting a precedent for how real-world infrastructure assets can be integrated into DeFi lending markets. This was executed on the MetaStreet protocol within the Arbitrum One ecosystem, marking a milestone in RWA-backed lending.

The deal involved 60 RK-3855 mobile chip servers housed in a Malaysian data center, where these tokenized assets served as collateral for an onchain loan. These loans, originally issued against $890k in collateralized assets, continue to be repaid every 30 days and have successfully rolled over more than six times, demonstrating the sustainability of DePIN-backed lending.

The financial structure of this deal further validates the model that USD.ai will expand upon. The deposit value stands at $621k, supporting a $3.3M contract value, with all hardware fully deployed and generating revenue through Aethir, a leading decentralized GPU network.

Key counterparties include Tactical Compute (Abu Dhabi Global Markets) as the borrower and Metaversal Asset Management (NYC-based) as the primary depositor. This transaction not only proved the viability of onchain DePIN financing but also provided a working framework that USD.ai will leverage to expand Arbitrum’s RWA ecosystem at scale. By integrating similar real-world infrastructure assets into its lending model, USD.ai builds upon this precedent, bringing structured, scalable, and capital-efficient DePIN financing to Arbitrum.

1600×1196 156 KB

4. Plan Design

1. Please describe your proposed product, including a description of the underlying assets and, if more than one asset, the proposed allocation among assets and general investment guidelines.

There are two products: $USDai and $sUSDai.

  • $USDai is the unstaked, non-yielding coin that converts stablecoin deposits into $M
  • $sUSDai is the staked, yield bearing coin that passes the yield generated by both $USDai and $sUSDai holders.

This proposal recommends a maximum of a $250k initial allocation toward the $sUSDai product, especially as it benefits from a staking ratio to amplify yields but through the same yield source (money market funds / US treasury bills).

Do investors have any shareholder, investor, creditor, or similar rights?

N/A

1. Describe the legal and contractual structuring for your product, including regulatory bodies overseeing your business and the product and identifying all legal jurisdictions interacting with your product.

As USD.ai minted assets will be fully allocated to $M in Stage 1, we recommend reading through the M^0 proposal to examine relevant regulatory bodies.

2. Would Arbitrum’s assets be bankruptcy remote from your own entity and its officers/key contributors?

As USD.ai minted assets will be fully allocated to $M in Stage 1, we recommend reading through the M^0 proposal to examine their bankruptcy remote processes.

How are Arbitrum’s assets protected vis-a-vis the bankruptcy of the brokerage or applicable financial institution (e.g., bank deposit insurance, securities insurance, etc.)?

N/A

Does the Issuer issue more than one asset? If so, what is the priority relationship between different asset classes?

USD.ai issues two assets: USDai and sUSDai. USDai is the non-yielding base asset, fully backed by M^0’s Treasury Bills, while sUSDai is the staked version that accrues all yield over time.

In Stage 1 (Current Phase), both USDai and sUSDai hold equal redemption priority. Since all underlying assets are fully liquid T-Bills, users can unstake sUSDai and redeem USDai at any time without a queue.

In Stage 2 (Future Phase), the system will introduce diversified collateral with varying maturity structures. At this point, USDai will take redemption priority over sUSDai, ensuring that non-yield-bearing deposits remain liquid while yield-bearing deposits reflect their longer-term collateral exposure.

This proposal focuses on Stage 1 only.

1. Provide a detailed cash flow diagram that shows the flow of funds from ARB/Fiat conversion, investment in underlying asset, payment of expenses, sale of underlying asset, and repayment (Fiat/ARB conversion), including the counterparties and legal jurisdictions involved.

As USDai is a fully onchain product, all counterparties and legal jurisdictions can be read through the M^0 STEP application. The flows below illustrate the minting of sUSDai and redeeming of sUSDai on a smart contract basis.

USD.ai operates as a fully on-chain product, with counterparties and legal jurisdictions documented in the M^0 STEP application. The fund flow process is as follows:

  1. Minting sUSDai
  • Users deposit $ARB or $USDC
  • Funds are swapped into $M, M^0’s on-chain Treasury Bill token
  • $M is escrowed within the USDai Vault
  • Users mint USDai, which can be staked for sUSDai
  • Passive capital in the USDai vault generates yield from M^0’s T-Bills, distributed to sUSDai holders
  1. Redeeming sUSDai (Liquidation Flow)
  • Users unstake sUSDai to revert to USDai (instantaneous in Stage 1)
  • USDai is redeemable for $M, triggering the T+1 off-ramp process
  • Once $M is converted back into USDC or ARB, funds are available for withdrawal

This structure ensures that all capital is continuously backed by real-world assets while maintaining immediate liquidity for redemptions.

1600×911 134 KB

1600×892 129 KB

2. Describe anticipated tax consequences (if any) in transacting on the underlying and/or receipt of yield.

All users are responsible for their own tax reporting based on local regulations depending on the classification of how yield-bearing tokens.

USDai accrues value over time, meaning holders may or may not be subject to capital gains tax based on appreciation. Unlike traditional interest-bearing accounts, sUSDai increases in value rather than distributing yield directly.

  • Example: A user stakes 1 sUSDai at $1.00. After one year, it becomes redeemable for $1.0415, reflecting the 4.15% APR yield.
  • The tax classification will depend on local jurisdiction, but in most cases, this would be treated as a capital gain rather than income.

3. Describe the process and expected timeline for liquidation of assets, if given instructions to do so by Arbitrum governance.

If instructed by Arbitrum governance, asset liquidation follows a structured process:

  • Unstaking from $sUSDai to $USDai is instantaneous since collateral is held in liquid Treasury Bills.
  • Redeeming $USDai for $M follows T+1 settlement, as M^0 processes off-chain conversions.
  • Since $USDai and $sUSDai are non-custodial, Arbitrum governance would maintain full ownership of any underlying assets held in vaults.

This ensures that funds remain accessible while adhering to M^0’s standardized liquidation process

4. What amount of first-loss equity will Sponsor provide to ensure over-collateralization, how is the first-loss equity denominated, and what is the source of capital?

For Stage 1, there is no first-loss capital risk for sUSDai holders. M^0 provides a 3% equity buffer, meaning USDai is always overcollateralized with liquid assets.

For Stage 2, when diversification into DePIN loans begins, risk mitigation will include:

  • 42% overcollateralization (70% LTV on loans)

  • Evertas insurance (covering 20% of Bitcoin mining hash rate globally)

  • A risk-adjusted approach to ensure capital buffers remain intact

At no point will sUSDai holders be exposed to uncollateralized risk in Stage 1.

5. Describe the liquidity and stability of the proposed underlying assets, including anticipated settlement times from the sale of the underlying to the repayment of ARB.

USDai and sUSDai are designed to maintain deep liquidity and stability:

  • Treasury Bills (Stage 1) provide 100% liquidity coverage with minimal volatility.
  • Redemption settlement times are dictated by M^0’s T+1 Treasury Bill process.
  • Stage 2 will introduce yield diversification, but with a 20-30% allocation into DePIN loans, ensuring stability remains paramount.

This ensures that Arbitrum governance can confidently allocate funds into USD.ai without exposure to unexpected liquidity shocks.

6. If relying on the blockchain for any of the transactional flows, please describe any blockchain derived risks and mitigations.

USD.ai is built with LayerZero’s OFT standard, ensuring seamless cross-chain interactions while maintaining security on Arbitrum. All smart contract transactions related to yield generation occur natively on Arbitrum One, reducing reliance on external chains.

Key risk mitigations include:

  • Immutable smart contracts for yield routing (no admin override)
  • Strict pool whitelisting (deposits can only enter approved M^0 pools)\

7. Does the product rely on any derivative product (swaps,OTC agreements?)

N/A.

8. List all the third party counterparties linked to your assets including and not restricted to prime broker if any, custodian, reporting agent, banks for derivatives or loans and provide primary contact details for the third party counterparties

USD.ai’s only counterparty in Stage 1 is the M^0 Protocol, which manages T-Bill exposure. All transactions occur fully on-chain, eliminating reliance on centralized entities.

9. Can you explain how is risk management (inv and operational) being done? Can you provide a copy of your risk management policy?

USDai is fully onchain and in Stage 1 will be collateralized by $M assets. M^0 has an Adopted Guidance report, as provided in their STEP application.

Performance reporting

1. What are your proposed performance benchmarks? If this is substantially different from the underlying assets, please explain why.

The primary benchmark for USD.ai is total yield generated, as yield accrual is the core function of the protocol. Since Stage 1 yield derives exclusively from M^0-backed Treasury Bills, performance will closely track M^0’s yield generation from U.S. government securities. The expected APR benchmark for sUSDai will be approximately 8.30%, derived from a 4.15% base rate with an estimated 50% staking ratio.

2. Describe the content, format, preparation process, and cadence of performance reports. This should include proof of reserves, if appropriate. Please include a sample report.

Performance reports will be generated from on-chain data and made available in multiple formats:

  • Live dashboards: USD.ai’s native app and Dune Analytics will provide real-time tracking of TVL, staking ratios, yield distribution, and redemption activity.
  • Snapshot reports: Key metrics will be compiled into monthly and quarterly reports and published on the Arbitrum forum for transparency. These will include:
    • Total USD.ai supply and staking ratio
    • Total yield distributed to sUSDai holders
    • Underlying collateral composition (100% M^0-backed T-Bills in Stage 1)
    • Redemption trends and liquidity buffer analysis
    • Protocol fee breakdown, if applicable

The following is one of multiple reports providable, configurable and provable as all actions are completed on-chain with regards to the USDai contracts.

1600×1000 93.9 KB

3. Who provides the performance reports in respect of the underlying assets?

Performance data will be sourced directly from on-chain records and made accessible via the USD.ai dashboard, Dune, and other analytics tools. Reports will be maintained by Permian Labs, USD.ai’s technical partner, ensuring accurate and up-to-date tracking of all yield and collateral data.

Since all assets are fully on-chain, proof of reserves can be verified in real-time through:

  • M^0’s collateral tracking tools (showing T-Bill-backed reserves)
  • USD.ai’s smart contract balances (ensuring that minted USD.ai remains 1:1 backed by M^0’s issued assets)
  • Independent dashboards (Dune, DeBank, etc.)

4. Describe any formal audit process and timing of such audits.

As all data is sourced from onchain data, 3rd parties can provide direct information, such as Dune, DeBank, and other onchain analytics providers. This approach ensures that all yield data is verifiable, transparent, and auditable at any given time.

Smart Contract/Architecture

1. How many audits have you had and name of auditors? Please provide a copy of reports.

USD.ai and its dependencies have undergone multiple audits by top-tier firms and security collectives.

MetaStreet & M^0 contracts audited separately

Both M^0 and MetaStreet are also live on Immunefi for continuous monitoring and bounty-driven disclosures.

2. Is the project permissioned? If so how are you managing user identities? Any blacklisting/whitelisting features?

USD.ai is a non-custodial, permissionless protocol on the smart contract level. However:

M^0’s Minter role is permissioned, and subject to internal governance and compliance frameworks.

Admin controls are managed by a multisig, with routing logic only allowing funds into predefined, whitelisted M^0 pools.

Borrowers (Stage 2 only) are permissioned via offchain onboarding and KYC, given hardware must be tokenized and verified via bailment contracts.

3. Is the product present on several chains? Are there any cross-chain interactions?

USD.ai uses the OFT (Omnichain Fungible Token) standard via LayerZero, enabling seamless bridging while preserving supply constraints across chains.

  • Primary execution environment: Arbitrum One

  • Yield generation and lending logic: On Arbitrum

Integration planned with USDT0 and routing logic for paired liquidity provisioning on Arbitrum AMMs.

4. Are the RWA tokens being used in any other protocols? Please describe the various components of the ecosystem.

In Stage 1, USDai will integrate with the diverse DeFi ecosystem in Arbitrum, such as money markets, oracles, AMMs and yield tokenization platforms.

In Stage 2, USDai will integrate across various DePIN economies across various blockchains, with the yield being generated and shared with the Arbitrum DeFi ecosystem.

5. How are trusted roles/admins managed in the system? Which aspects of the solution require trust from users?

The USD.ai smart contracts enforce strict limits on how capital can be allocated. Admins can only route deposits to M^0 pools that are pre-approved and whitelisted.

  • There is no discretionary investment decision-making by admins.

  • Users can verify the pool configurations and yield routing logic on-chain.

  • Admin multisig roles are secured and include signers from MetaStreet Foundation, Permian Labs, and third-party observers.

6. Is there any custom logic required for your RWA token? If so, please give any details.

Yes, primarily in sUSDai, which integrates:

  • ERC-4626 base vault functionality

  • Future compliance with ERC-7540 for DeFi-native interest-bearing compatibility

Redemption in Stage 1 is instantaneous (due to fully liquid $M).