段永平建倉Circle，他在賭什麼？

Original source: Fintech Blueprint

Compiled and organized by BitpushNews

Yesterday, the US SEC disclosed the latest quarterly 13F holdings report. Duan Yongping, known as the "Chinese Warren Buffett," made a significant portfolio adjustment through his family wealth and charitable fund account, H&H International Investment LLC, which manages over $20 billion. For the first time, he broke ground by establishing a position in the compliant stablecoin giant Circle (US ticker: CRCL), with a market value of $19.08 million.

As a staunch value investor, Duan Yongping became famous for heavily investing in Apple and Kweichow Moutai. His investment philosophy has always been "don't invest in what you don't understand." This move into Circle not only signifies the formal acceptance of Web3 compliant assets by traditional veteran capital. This article will deeply analyze Circle's Q1 performance and latest product layout, examining whether this stablecoin giant can complete a business model pivot from "interest-driven" to "infrastructure" through a fundamental restructuring of its underlying architecture.

Below is the main text:

Circle has had a busy week.

Alongside the release of its Q1 2026 earnings—where total revenue and reserve interest income approached $700 million (up 20% year-over-year), USDC circulation reached $77 billion, and on-chain transaction volume hit $21.5 trillion—the company also made two major product announcements and completed a $222 million token pre-sale.

Changing the "Interest Rate Coupon" Label

For a long time, Circle has been labeled an "interest rate agency tool": 99% of its 2024 revenue came from interest earned on USDC reserve assets.

This makes the business extremely sensitive to interest rate cycles, leaving equity investors with little basis for valuation pricing beyond net interest income and USDC issuance growth. Arc (its Layer-1 blockchain), Circle Agent Stack (agent technology stack), and the Payments Network represent Circle's concentrated efforts to change this narrative—aiming to diversify revenue and revalue the stock's valuation logic from "yield multiple" to "infrastructure multiple."

Perhaps most unusual: Circle, a publicly traded company with a traditional equity structure, raised $222 million through a token pre-sale for its new stablecoin-focused Layer-1 blockchain, achieving a fully diluted valuation (FDV) of $30 billion.

In finance, some instruments enter the standard cap table, while others are tokens specific to a protocol. It's worth noting that Coinbase's Ethereum L2 network, Base, has yet to issue a token. A publicly traded company worth tens of billions of dollars being able to complete such a token raise means that token assets have officially landed on Wall Street.

This funding round was led by Andreessen Horowitz (a16z), committing $75 million, with participation from BlackRock and Apollo. The pre-sale includes multi-year lock-up periods; investors also have repayment rights if Arc fails to reach key network milestones.

Circle holds 25% of the initial 10 billion token supply, with 60% allocated to network participants and 15% reserved as a long-term reserve. The Arc mainnet is expected to launch in the summer of 2026, and as of early May, its testnet has processed 244 million transactions.

Currently, the utility of the ARC token is still in the exploration phase. This means you can still raise over $200 million today without designing a sophisticated tokenomics model. Moreover, if we look closely, building a Layer-1 blockchain doesn't actually require $200 million.

Alongside the launch of Arc, Circle also unveiled the Circle Agent Stack—a toolkit for developers to build AI agents that transact using USDC, including wallets, a marketplace, and a nanopayments layer capable of supporting transfers as low as $0.000001.

With this, the company joins Stripe, Coinbase, Visa, Mastercard, Shopify, Fiserv, and Brex in the race to "provide banking services for robots."

Arc is a Defensive Move

Today, USDC runs on dozens of public chains and wallets like Ethereum and Solana. Circle can earn interest income from all these reserve assets. But the issue is how much of that revenue it can actually keep for itself.

According to the "Cooperation Agreement" signed with Coinbase in 2023 (agreed upon when the Centre consortium dissolved, giving Coinbase, as Circle's largest distribution channel, significant bargaining power), the distribution of reserve interest income is divided into three steps:

• Circle first takes a small issuer fee at the top.
• Then, both parties receive respective reserve interest income based on the proportion of USDC held in their respective custody products.
• As for all remaining profits—Coinbase directly takes 50%.

The result is that Coinbase can extract a portion of reserve interest income even from USDC that has no custodial relationship with it.

In 2024, out of Circle's total revenue of $1.68 billion, a whopping $908 million was ceded to Coinbase. This agreement automatically renews every three years, and Circle has no unilateral right to exit. Therefore, Arc is, to some extent, Circle's effort to build an underlying infrastructure that it fully controls and from which it can directly earn fees.

To reiterate: Coinbase enjoys a 50% "net clipping right" on virtually all of Circle's revenue, and Circle has no way out except to find a clever "back door."

Arc's customer acquisition logic is straightforward: a Layer-1 blockchain purpose-built for stablecoin-native finance. It uses USDC as its gas token, features sub-second transaction finality, optional privacy, EVM compatibility, and a quantum-resistant architecture. For institutions whose core business is moving money, this represents a new generation of settlement infrastructure and an alternative to ACH, SWIFT, and correspondent banking.

Launched as a testnet in October 2025, it has already attracted over 100 institutional participants, including BlackRock, Goldman Sachs, Visa, and State Street, and has processed 244 million transactions.

To be fair, similar institutions have previously joined Tempo and various AI payment and agent protocols we've covered. This indicates that the industry is diversifying in its approach to restructuring payment rails.

In contrast, the $30 billion FDV attached to the pre-sale seems somewhat hard to justify. The functionality of the ARC token is still in the exploration phase. What investors are currently betting on is essentially the option value of Circle owning the "stablecoin settlement mother chain"—thereby closing the vertical ecosystem loop and plugging the current value leakage to third parties. Whether this option is worth $30 billion depends on future transaction volumes. Specifically, it depends on whether Circle can migrate a sufficient share of the current $77 billion in circulation onto Arc to generate the service fee revenue needed to support that valuation.

Simultaneously, the regulatory backdrop intensifies this urgency.

The GENIUS Act, signed into law in July 2025, explicitly paves the way for banks to issue their own payment stablecoins through subsidiaries, under the supervision of their existing federal regulators. JPMorgan and Bank of New York are already running tokenized deposit pilots. Once regulated bank-issued dollar tokens reach scale, market demand for third-party stablecoin issuers like Circle could narrow.

Arc doesn't directly solve this problem, but owning proprietary on-chain infrastructure can create network effects and switching costs. It's a defensive line to hedge against profit-taking or vertical integration risks from everyone from Canton to Ripple to JPMorgan's Kinexys.

Circle Agent Stack is an Offensive Move

The Agent Stack is a developer toolkit for building AI agents that can transact using USDC. It consists of wallets, a marketplace, and a nanopayments layer enabling transfers as low as $0.000001. The core logic is: as AI agents autonomously handle more operational and financial tasks, the transaction sizes and granularity they require will be unsupported by existing payment rails (like card networks, ACH, SWIFT) due to their high fixed costs, making sub-cent transactions economically unviable. A USDC-native chain supporting programmable micropayments has no such cost floor. For an AI agent needing to pay per API call, per second of compute, or per data query, no perfect solution currently exists in the market.

Ramp launched Agent Cards in March 2026. In short, it allows businesses to issue virtual cards for autonomous agent spending. Stripe, after acquiring Bridge in late 2024, has its own answer: issuing agent-specific cards via Bridge, providing wallet infrastructure through Privy, and supporting stablecoin payment acceptance in 32 markets.

• Ramp's Agent Cards: Built for enterprise expense control.
• Circle's Agent Stack: Targeting USDC-native micropayments on the Arc chain.
• Stripe: Positioning itself as a full-stack layer (offering fiat, stablecoin, and wallet infrastructure under one API).

The Circle vs. Stripe Showdown

Where Circle has a structural advantage is the asset itself.

USDC is the dominant compliant stablecoin, serving as the unit of account for a significant portion of on-chain activity. Stripe's Bridge, on the other hand, issues its own stablecoins via "Open Issuance." USDH, one of Bridge's flagship issuances, announced its shutdown this week after failing to compete with the $5 billion USDC presence on Hyperliquid, with Coinbase stepping in as the official USDC treasury deployer. Building agent infrastructure on top of USDC means agents inherit existing liquidity and network depth from day one. This asset advantage has proven much harder to replicate than it might seem.

As mentioned, Stripe also incubated Tempo, a Layer-1 blockchain specifically tailored for payments. However, Tempo is positioned as a general-purpose payment settlement layer supporting any stablecoin, while Arc is built entirely around USDC. Both companies are betting that the future of payments will settle on specialized, proprietary chains rather than general-purpose ones like Ethereum.

Differences in capital structure are also noteworthy. Circle raised $222 million via a pre-sale for Arc ($30 billion FDV). Stripe, a privately held, consistently profitable company with a recent valuation of $70 billion, can fully fund Tempo and Bridge's expansion from its own balance sheet cash without diluting equity through tokens.

The types and scale of ammunition these two companies can deploy to absorb and subsidize the cost of bootstrapping a new chain ecosystem are fundamentally different.

Ultimately, the capabilities and inclinations of a "payment processor (like Stripe)" and a "cash-equivalent financial instrument issuer (like Circle)" are vastly different. The former excels at distribution, sitting on countless merchants and customers in its ecosystem; the latter has a presence in nearly every exchange and crypto wallet. We believe blindly pursuing vertical integration and getting caught in an expensive arms race would be a mistake.

The Math on the Revenue Ledger

Circle's business model today is simple: $77 billion in USDC outstanding, earning roughly a 4.1% return on reserve assets, with a significant portion flowing to Coinbase per distribution agreements. Its full-year 2025 revenue was $2.75 billion.

Analysts project around $3.2 billion in revenue for 2026, implying roughly 15% growth. This is quite modest compared to last year's 64% growth rate, reflecting two real headwinds:

• Falling interest rates compressing reserve asset yields.
• The GENIUS Act imposing constraints on how reserve income can be shared with distribution partners, subjecting the Coinbase agreement to regulatory scrutiny.

The new products must be understood in this context. Circle estimates non-reserve revenue of $150 million to $170 million for 2026, higher than the $110 million in 2025 but still less than 6% of total revenue. Arc transaction fees, Agent Stack developer revenue, and CPN (Circle Payments Network) fees are all in very early stages. To achieve the valuation re-rating from "interest rate agency tool" to "infrastructure platform," these business lines need not just absolute growth but a material increase in their revenue share. Based on the current trajectory, Circle's narrative is running ahead of its financial numbers.

Stock performance reflects this tug-of-war. CRCL went public at $31 in June 2025, briefly surged to nearly $300, and has since settled around $114. Following the Q1 earnings release, JPMorgan raised its price target to $155, Needham to $150, while Deutsche Bank set one at $101. Consensus expectations settle between $125 and $130, implying very cautious upside from current levels.

Bull Case vs. Bear Case

The bull case requires three conditions to be met simultaneously:

• USDC circulation grows fast enough to offset the impact of falling reserve yields.
• Arc generates significant fee income and partially replaces or breaks away from the Coinbase agreement.
• The Agent Stack establishes its dominant underlying infrastructure position in the agent payment space before Stripe crushes it with scale.

If all three are achieved, Circle will have successfully transformed into a payment infrastructure company, with its valuation multiple driven by transaction volume and network effects rather than being hostage to the Fed's interest rate cycle.

The bear case is much simpler:

Interest rates fall faster than circulation grows; renegotiating the Coinbase agreement reduces distribution channels without effectively compensating for lost volume; Arc fails to migrate a meaningful amount of USDC onto its own chain; Stripe or Ramp launch superior agent infrastructure at lower cost, successfully encircling Circle.

Circle's announcements are undoubtedly the right strategic moves. But for now, they are chips and bets, not yet transformed into real businesses. Circle is asking investors to pay for the option value of these three conditions being met simultaneously, while its core business model faces tangible structural headwinds. This request isn't unreasonable—it just feels a bit expensive at current valuation levels.