Author: 0xJeff
Translated by: Tim, PANews
Stablecoins are the most important infrastructure in the crypto field. Without stablecoins, we would not have a stable unit of account for investors to temporarily store assets (establishing centralized exchanges, decentralized exchanges, perpetual contracts, money markets, and all other vertical fields would become extremely challenging).
Stablecoins are experiencing rapid popularization, with total supply, trading volume, and turnover speed (indicating the frequency of stablecoin exchanges) showing a surge between 2023 and 2025, particularly in areas such as payments and cross-border transactions.
Moreover, we are seeing clearer regulatory frameworks and further institutional adoption of stablecoins: for example, payment company Stripe launched stablecoin financial accounts in 101 countries; Société Générale plans to issue a USD stablecoin; major banks like Bank of America, JPMorgan, Citigroup, and Wells Fargo are planning to jointly issue stablecoins; many large enterprises are exploring using stablecoins for payments to reduce transaction fees from Visa and Mastercard; and more related developments.
Circle's recent successful initial public offering has also sparked a new wave of stablecoin enthusiasm, attracting more stakeholders.
With the deep application of traditional finance, stablecoin innovation in the AI field is also advancing, aimed at addressing challenges faced by service providers and users in the Web3 AI ecosystem.
First Challenge
Although AI teams typically design AI tokens as the core component of the AI ecosystem (covering payments, governance, utility, etc.), they are more willing to focus resources on AI product R&D rather than the decentralized finance field.
For example:
- Virtuals uses their VIRTUAL and AGENT liquidity pools, which helps accumulate value for their VIRTUAL token, but simultaneously makes it difficult for the Agent team and LPs to provide liquidity (due to Impermanent Loss).
- Aethir uses $ATH as a payment method for computing services, which can create a strong flywheel effect for the token but also brings the problem of payment unit price volatility.
- Bittensor pays miners, validators, and subnet owners in dTAO (subnet token) ➔ Participants must sell subnet tokens for stablecoins to maintain operating funds.
While these two examples can be good flywheels for AI tokens, they also hinder some key participants from participating because the design itself causes market volatility.
(By the way, the three cases mentioned just now are excellent examples, and many AI teams' token designs are currently very poor, especially those with "fair launches")
Token quantity growth coupled with less-than-ideal design leads to thin market liquidity, making it difficult to build DeFi ecosystem applications on top of it.
Projects Addressing This Challenge
Maitrix: AI Stablecoin Layer
Maitrix launches over-collateralized AI-native stablecoins (AI USD) for various ecosystems, transforming the originally volatile (but high-yield) AI economy into a predictable, composable, and vibrant economic system through AI-native stablecoins.
Maitrix's Key Components
Users mint and burn AI USD by depositing AI tokens and their derivative assets (liquidity staking or staked AI tokens) through CDP.
- Stablecoin Launching Platform: AI projects can create their own AI stablecoins using native tokens and derivatives
- Curve ve(3,3) Mode Incentives: MAITRIX tokens implement governance through voting escrow mechanism, transfer token emission direction, and adopt a bribery system similar to the ve(3,3) mechanism.
- Stablecoin Automated Market-Making Model: Supports exchanges between various AI USD tokens
Supported AI USD Stablecoin Assets (to date)
- Aethir USD (AUSD): Stable Payment Solution for Computing Services
- Vana USD (VanaUSD): Data-Supported Stablecoin
- Virtual USD (vUSD)
- ai16z USD (ai16zUSD)
- 0G USD (0USD)
- Nillion USD
More participants can be discussed later.
Currently, there are not many detailed documents about the use cases of each AI stablecoin.
I will explain the technical details in detail when the technical white paper comes out, but for now, Maitrix is the only team building this financial infrastructure layer for AI projects, and has established key partnerships with multiple top AI ecosystems.
Maitrix's testnet has received a warm response, and the mainnet launch is coming soon.
Second Challenge
As AI continues to develop and become widely applied, market demand for computing resources is increasing day by day. Data centers and cloud service operators need to plan resource expansion strategies in advance to seize future market opportunities.
Enterprise-level GPUs (such as NVIDIA H100/H200) are often expensive and require large-scale financing. Traditional financing methods (such as bank loans or equity financing) are lengthy and complex, making it difficult for data centers to quickly expand capacity to meet market demand.
This is precisely why Gaib and USDAI have emerged.
Projects Addressing This Challenge
Gaib: First AI and Computing Resource Economic Layer
Gaib helps data centers efficiently finance by tokenizing future GPU cash flows, while providing investors with yield-generating assets backed by real-world assets (GPUs).
Its working principle is roughly as follows:
- Cloud service providers and data centers package future GPU revenue cash flows as securitized financial products
- These cash flows are tokenized for 6 to 12 months.
- Investors purchase these tokens and start claiming periodic rewards.
They call this AI synthetic dollar "AID".
Each AID token is backed by a portfolio of graphics card financing transactions and configured with government bonds or other liquid assets as reserves.
The estimated floating yield is around 40% annually. This highly depends on the ratio of debt financing to equity transactions in the GPU investment portfolio (equity yields are 60-80%+ annually, debt yields are 10-20% annually).
So far, the project has accumulated about $22 million in total locked value, incentivizing deposits through "Spice" points, which will make investors eligible for future airdrop rewards.
Earlier this year, Gaib collaborated with Aethir on the first GPU tokenization pilot project.
This pilot project is only a GPU tokenization and fragmentation solution, part of their roadmap, which will be extended to the GPU-backed stablecoin "AID".
If you are curious about the protocol's operating principles and anchoring mechanism, it is recommended to check their Gitbook documentation.
USDAI (Built by Permian Labs): Yield-Generating Synthetic Stablecoin Backed by Real-World Assets
USDAI is similar to Gaib in some ways but different, it is a stablecoin backed by hardware assets as loan collateral (including graphics cards, telecommunications equipment, solar panels, etc.).
What is discussed here is a pure debt financing transaction: borrowers (asset holders) obtain loans from USDAI and pay interest, and the generated revenue from these interests will belong to USDAI token holders.
Permian Labs is the team behind the Metastreet project, a top-tier structured credit market platform that provides innovative financial services including Non-Fungible Token collateral loans, structured credit products for illiquid assets, real-world assets (such as luxury watches and artworks), and NFT yield token split solutions like Pendle PT YT.
USDAI has not yet been launched, but its target is to set an annual yield of 15-25%, with asset allocation to be gradually adjusted in three stages: initially 100% invested in US Treasury bonds, ultimately transitioning to 100% allocation of physical hardware assets. (Note: Based on domain knowledge, "hardware" in the crypto field may refer to physical assets like mining machines, which needs to be confirmed by the project's whitepaper)
USDAI uses CALIBER technology, which simplifies loan and issuance processes while complying with legal standards, ultimately enabling GPU on-chain operations.
Here is more information from Delphi Digital about USDAI
Okay, let's clarify the differences: USDAI focuses on debt assets, providing more diverse asset types. With its CALIBER model, they can cover a wide range of application scenarios (wherever the demand is), while Gaib focuses more on equity assets, offering higher expected yields.
USDAI's private beta has been officially launched, with an initial locked value of $10 million.
If you are interested in being among the first participants, you can fill out a form (they will provide additional incentives to early participants).
Other AI-related Stablecoin Products
Almanak recently launched alUSD (based on the ERC-7540 standard, an extension of ERC-4626). This is a tokenized AI yield optimization strategy that maximizes risk-adjusted returns by investing stablecoins on platforms like Aave, Compound, Curve, and Yearn.
The Almanak team is about to launch a points activity to introduce initial liquidity, while continuously promoting DeFi composability, allowing users to use alUSD as collateral or engage in circular lending to maximize returns in the future.
AixFi, interstable: This is a DeFi protocol vault that automatically deploys USDC assets. Initially, it will adopt a rule-driven mode, gradually introducing AI decision-making mechanisms, and plans to officially launch on the Virtuals platform this month.
What's Next?
We could very well witness the rise of another protocol similar to Ethena, focusing on creating high yields for users' stablecoins by utilizing GPU computing power. But more critically, how the protocol maintains its anchoring exchange rate mechanism, ensuring the price can quickly return to the $1 value benchmark during a crisis.
In the future, we may see more tokenized AI strategies. We have already witnessed AI's excellent ability to optimize yields, dynamically calculating gas fees, rebalancing fees, slippage, and other variables (as demonstrated by Giza's ARMA strategy). If these strategies are tokenized into highly composable vaults that can be used as collateral or generate yields through 5-10x leverage, what imagination space would that open up?
We will see enhanced liquidity in the Web3 AI ecosystem. As participants like Maitrix are building stablecoin infrastructure for top AI ecosystems, more value created by AI will become more composable and accelerate into the DeFi domain, thereby enhancing the entire Web3 ecosystem's value capture ability.
Although these teams are very interesting, in the stablecoin field, risk, anchoring management, redemption, and liquidation mechanisms are crucial. Be sure to fully read and understand all risk factors before deciding to invest.
