AI makes the cost of writing code close to zero, but the moat of financial business is not in the code. Multicoin partner Spencer Applebaum published the latest long article, dismantling the three moats that DeFi can really make money in the AI ​​era: distribution channels, capital and proprietary data. (Previous summary: Multicoin Capital is optimistic about SOL, HYPE and ZEC! Partners reveal 2026 must-buy list and valuation model) (Background supplement: "No longer believes in the prospects of web3" Multicoin Capital founder quickly resigned after deleting the PO post: Capital is still behind decentralization) Spencer Applebaum of Multicoin Capital published an article on the X platform on November 10, pointing out that AI As software costs approach zero, finance is one of the few industries that still has a moat, and DeFi is in the best position. His latest long article published simultaneously on the Multicoin blog reorganizes the evolutionary framework of DeFi value capture and proposes three moats: distribution channels, capital and proprietary data. Original post by Spencer Applebaum on X . Source: build, with core lending or trading software potentially assembled in a weekend. An essay by Spencer Applebaum argues that code alone will no longer provide a lasting edge. It identifies three durable DeFi moats: user… — Coincamps (@coincamps) November 11, 2026 As software costs approach zero, financial barriers become more obvious. Spencer Applebaum first points out a cruel reality in the article: To build a DEX or lending protocol five years ago, the team had to first raise 300 Ten thousand dollars, five engineers, and six months to create a minimum viable product. Today, the founder uses AI tools such as Claude Code to build the matching engine, risk control engine, trading interface and API of the perpetual contract exchange in a weekend. He believes that AI is lowering the competitive threshold in almost all software segments, but financial services will be one of the most difficult businesses to be commoditized. The reason is simple: the most valuable part of a financial company has little to do with the software itself. Accumulating loyal customers, growing a balance sheet, and honing proprietary data and underwriting capabilities all require years of relationships, track records, and trust. AI can make software cheaper, but it cannot make these capabilities replicable. He used a comparison to illustrate this point: the big language model can write the code of the lending agreement in an afternoon, but it cannot change the $10 billion in stable currency deposits flowing to the contract, nor can it attract 28 million customers for the new online brokerage firm. In other words, programming code is no longer a scarce resource; what is truly scarce is what software cannot create. From Fat Protocol to Moat: Multicoin’s Thinking Iteration Applebaum reviewed the evolution of Multicoin’s thinking on DeFi value capture over the years. In the early days, they studied how application tokens can capture value through work tokens and destruction/minting balance models; later they focused on the defensible and unforkable status of DeFi, arguing that anyone can fork open source code, but cannot replicate liquidity or network effects; more recently, they demonstrated that the maximum value can be captured when DeFi protocols must manage some kind of real risk. This article is the next iteration. He first stated his premise: DeFi will continue to engulf CeFi and eventually the entire traditional financial system. The current spot market share of DEX relative to CEX is at a record high and is still rising; at the same time, tokenized assets such as U.S. Treasury bonds, private credit, and stocks are ushering in real adoption. It is not just native crypto users that are driving growth. Traditional financial giants, financial technology companies, institutional asset management, and traditional securities firms are all building products on the chain and bringing existing customers into the on-chain world. But the key issue is that today’s blue-chip DeFi protocols may not necessarily be able to reap this wave of growth dividends. As the cost of software and data analysis falls, value will flow to those who control the "scarce elements that software cannot create." He proposed the three core moats of DeFi: distribution channels, capital and proprietary information. There is also a fourth element "risk tolerance", which works in different ways and is more like a wedge that cuts into the market. Moat 1: Distribution pipeline, waterfall value chain from stablecoin to wallet Applebaum starts with the stablecoin value chain. As long as it is scaled up and the interest rate is not close to zero, stable currency issuance has always been a good business. Tether and Circle hold US dollar reserves to earn interest, and most of the time they hardly distribute profits to holders. Since 2021, Tether alone has made more than $20 billion on this model, with profit margins approaching 100%. But Hyperliquid’s case is a game changer. More than $6 billion in USDC is deposited on its platform as derivatives margin, and these retained balances generate significant reserve interest income for Circle and its distribution partners. At a certain stage, Hyperliquid realized that it was the fundamental reason for the existence of this balance, so it held a bargaining chip: if Circle was unwilling to share the revenue, Hyperliquid could completely replace USDC with a stable currency that it was willing to share, or theoretically issue its own stable currency. Earlier this year, Hyperliquid reached a cooperation agreement with Coinbase based on the Aligned Valuation Asset Framework. According to the agreement, Hyperl... "AI makes code worthless?" Multicoin dismantles the three moats that make real money in DeFi. This article was first published in BlockTempo, the most influential blockchain news media.