USD.AI: The Hidden Beneficiary of the AI Boom

As capital flowing into the artificial intelligence (AI) industry rapidly increases, the demand for investment in AI infrastructure such as data centers and GPUs is also rising. According to law firm Quinn Emanuel, the issuance of debt related to data centers is expected to reach approximately $182 billion by 2025.

However, the funding required for AI infrastructure is not evenly distributed among all companies. The current market is centered around high-credit big tech firms and hyperscalers. By 2025, the bond issuance by hyperscalers is projected to be about $121 billion, accounting for roughly two-thirds of the total debt issuance related to data centers.

In contrast, small and medium-sized data center operators and GPU cloud companies find it challenging to secure funding. There are two main reasons for this. Building AI infrastructure requires significant upfront investments in data center sites, GPUs, power, and cooling facilities. However, revenue only begins to materialize once the facilities are operational. Without long-term contracts or stable cash flows, it is difficult to secure external funding.

The rapid depreciation of GPUs is also a concern. Typically, GPUs lose 45-65% of their value within three years, and after five to six years, a significant portion of their economic lifespan is considered over. Banks apply high risk weights to such assets, and considerable time is needed for due diligence and collateral valuation. As a result, the speed of loan execution often fails to keep pace with the investment speed in the AI industry.

USD.AI is a protocol that addresses the funding issues faced by AI infrastructure companies through on-chain liquidity. When users deposit stablecoins, a fully collateralized synthetic dollar, USDai, is issued. By staking this, users can receive a yield-bearing token, sUSDai. The deposited funds are lent to AI infrastructure companies, secured by GPUs. The returns paid to sUSDai holders consist of the interest on loans borne by borrowing companies and the interest earned from managing idle funds in U.S. Treasury bills before loan execution. This allows for more flexible lending compared to traditional financial markets.

The core of USD.AI is its ability to connect the growth of the AI industry to actual cash flows. As computing demand increases, so does the demand for GPU investments and funding, leading to an expansion of loan assets and interest income. This model brings the revenue generated from AI infrastructure finance on-chain, rather than merely riding the AI narrative.

High yields in the DeFi market often rely on token incentives. In contrast, USD.AI has a structure that distributes actual interest generated from GPU-backed loans to depositors.

Currently, USD.AI's cumulative loan agreements amount to approximately $265 million. A total of 14 loans are in progress, of which 11 are active loans generating interest, totaling about $100 million. The average loan interest rate is 11.5% per annum. There are three loans in the escrow stage, totaling $153.8 million, with an average interest rate of 11.6% per annum.

Adding to this is a pipeline loan of $259.2 million, which has completed the term sheet and is awaiting execution, bringing the total loan amount to $513.8 million across 20 loans.

Loans serve as the basis for sUSDai yields. Currently, active loans contribute 6.79 percentage points to the sUSDai yield, while loans in the escrow stage add 1.24 percentage points. The executed loans alone generate an approximate yield of 8 percentage points. If all pipeline loans are executed, the yield will increase by an additional 1.91 percentage points. Thus, a significant portion of the revenue generated on-chain is based on actual cash flows from financial transactions.

Even when the fundamentals of the protocol improve, it is common in the crypto market for token prices not to reflect this. Continuous selling pressure from token unlocks and the structure where protocol revenue does not translate into token value are major causes. In contrast, USD.AI is currently showing changes in both token supply and value capture structure.

The total supply of $CHIP is 10 billion. It is allocated as follows: 27.5% for ecosystem bootstrapping, 29.6% for investors, 23.1% for core contributors, and 19.8% for reserves.

The new supply circulating in the market comes from the ecosystem bootstrapping allocation. Approximately 2 billion tokens will be unlocked during the first quarter immediately after the TGE, followed by about 750 million tokens being unlocked sequentially over the next two years and nine months. After the first quarter post-TGE, the scale of new supply will significantly decrease.

The $CHIP TGE took place on April 30, 2026. Currently, it has been about three months since the TGE. The vesting of investor, core contributor, and reserve allocations will begin one year after the TGE. There will be a period of approximately three quarters without significant new unlocks.

Based on price and trading volume trends, it can be interpreted that a portion of the initial circulating supply released after the TGE has been absorbed by the market. The FDV of CHIP briefly rose to $1.4 billion immediately after the TGE but fell to around $300 million, a decline of approximately 79%, within a month.

Additionally, based on data from Binance, Bybit, and Upbit, where CHIP trading volume is high, the cumulative net volume, calculated by subtracting the declining trading volume from the rising trading volume, recorded approximately -$12.9 billion, indicating that selling volume significantly outweighed buying volume. This suggests that the selling volume after the TGE greatly exceeded the buying volume, and a significant portion of the initial circulating supply has been released into the market.

There are also mechanisms in place to connect protocol revenue to token value. USD.AI repurchased 33.88 million $CHIP tokens last May, which corresponds to about 3.39% of the total supply. Including the repurchase of previously conducted airdrop tokens, the cumulative repurchase amount is approximately 3.71% of the total supply.

It is also important to note that the funds available for repurchase are increasing. USD.AI's total profit rose from $963,740 in Q1 2026 to $5.1 million in Q2 2026, expanding approximately 5.3 times in just one quarter.

The increase in Q2 performance was driven by the execution fees from GPU-backed loans. The Q2 loan execution fees amounted to $4.67 million, accounting for most of the overall profit increase. If annualized simply, the total profit for Q2 would be approximately $20.4 million. As loans expand, the profits secured by the protocol also increase.

As protocol profits increase, the funds available for repurchase also grow. This strengthens the structure that connects protocol growth to token value. However, it is worth considering that protocol profits may temporarily rise during periods when loan executions are concentrated.

Additionally, USD.AI has proposed a "Revenue Protocol" utilizing sCHIP in its future roadmap. While the specific distribution method needs further confirmation, if a structure is introduced that distributes a portion of the protocol's revenue to sCHIP holders, it could align the interests of token holders and the protocol more directly. If revenue distribution to holders is implemented in addition to the existing buyback, the value capture structure of $CHIP could be further strengthened.

In summary, $CHIP is at a point where three factors are simultaneously aligning: ▲ high growth of revenue based on actual interest income ▲ entering a supply gap period without new unlocks ▲ a buyback structure that connects revenue to token value.

Of course, there are risk factors as well. Due to the nature of the lending business, the borrower's repayment ability and the residual value of the collateral GPUs influence the protocol's soundness, and GPUs are assets that depreciate at a rate of about 20% per year. If the investment cycle for AI infrastructure slows down, it is necessary to keep in mind that both new loan demand and the soundness of existing loans may deteriorate simultaneously. Since there are only two quarters of revenue data available, it is also necessary to verify the sustainability of the growth trend.

Nevertheless, USD.AI stands out as one of the few crypto protocols generating actual cash flows amid the massive wave of the AI boom. Key points to watch in the future include the execution speed of the pipeline loan of $259.2 million, whether the growth of Q3 revenue continues, and the concretization of buyback and revenue share policies. If these three points are confirmed, USD.AI will prove its title as "the hidden beneficiary of the AI boom" with on-chain data.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.