What Comes After Stablecoins? (ft. Circle CEO Jeremy Allaire and Chris Dixon)

Circle’s path to Arc began with an effort to expand Bitcoin, which Jeremy Allaire said was limited by ideological disputes within its community. Ethereum later offered enough capability for Circle to build the dollar protocol that became USDC, but by 2020–2021 its fees, cost, and speed no longer met the company’s needs. Allaire said Circle built Arc as a new operating system and network for a coming mainstream era, drawing on blockchain governance, platform ecosystems such as Android, and open-source development to support real money and investment contracts.

Arc had launched on mainnet a few weeks before the discussion with about 10 or 11 infrastructure partners; Allaire said it could eventually have 40 slots. Validators and operators will vote on membership growth. He argued that credible neutrality and internet-scale reach require geographic and geopolitical diversity, while operators must meet high operational, information-security, and compliance standards. Arc is public, permissionless, and open source for users and builders, but its validation operators are identifiable and selected under identity, operational, and legal covenants. Allaire called this distributed but not fully decentralized model a “middle way” suited to institutions, companies, and households that want known operators and transparent governance.

Circle’s network centers on USDC, alongside a euro stablecoin and other digital assets. Allaire said more applications increase utility and attract developers, while global liquidity—from primary issuance and redemption to secondary on-chain markets—helps drive network value. Circle mints and redeems about $150 billion of USDC per quarter. Allaire said its Cross-Chain Transfer Protocol, or CCTP, is the most widely used interoperability protocol across blockchain networks, and USDC the most transferred digital currency across them. CCTP handles about half of value moved between chains, reaching as much as 70% in recent quarters; Arc-issued assets, including tokenized equities and credit products, can also use it. Deep liquidity from Binance, Coinbase, and Hyperliquid benefits products that depend on efficient capital, while Circle’s Payments Network had almost a couple hundred financial institutions using stablecoin rails and blockchains to move value globally.

Allaire characterized stablecoin markets as winner-take-most, not winner-take-all, leaving room for multiple networks as the market grows. He said mainstream adoption remains ahead: most blockchain apps are not viable for mainstream users, and no network has a substantial developer-mindshare advantage or moat. He expects the next three to five years to reshape the sector as AI creates software and performs or coordinates economic work on blockchains. AI could learn and implement deterministic software infrastructure and APIs quickly, but Chris Dixon said enterprise AI adoption was about 2% and consumer adoption of agents remained very low.

Allaire sees open platforms combining money, financial contracts, and software primitives enabling new credit instruments and marketplaces involving people and machines. Tokenized real-world assets are an early sign of broader participation in financial markets, he said, with potential for more long-tail credit and equity issuers. Arc Studio, which Allaire described as a vibe-coding app for safely deploying smart contracts, addresses risks in publishing contracts and has drawn strong developer engagement. Builders are also exploring financial apps, games with economic systems, and projects combining AI, computer networks, and incentives, though Allaire cautioned that one unusual AI-network experiment may not become a major direction.

Better software experiences, Allaire argued, could make governance and complex economic arrangements accessible without workflows requiring multiple crypto-wallet signatures. One project coming to Arc aims to arbitrate work involving AI, using AI and potentially human arbitrators, with security, staking, and payment-release mechanisms. He predicted automated arbitration could scale for economic, creative, and intellectual-property work and create new possibilities for labor. Allaire also expects machines to take on much future creative and cognitive work, potentially multiplying economic activity and value exchange while distributing more services and creative output. Dixon said machines could plausibly make up most of the internet, with economic activity naturally included; Allaire noted that bots already account for most online activity and said an experimental Arc app received about 15,000 bot visits a day, including probes for exposed credentials and crypto vulnerabilities.

Dixon backed Arc because of Allaire and his team, and sees blockchains as a global finance computer with platform potential. He emphasized that they are both social communities and computers, making operating-system and programming-language design important. Their composability lets applications interoperate like Lego bricks, he said, so early projects can look like toys before developing into larger systems. Dixon also argued that open blockchains could counter AI’s centralizing tendency through open services, global access, and low developer fees; policymakers, in his view, should allow that decentralizing force to flourish alongside AI. Both speakers see AI and crypto as forces that could reconfigure the digital economy, though Dixon framed the horizon as 10–20 years.