AI Reduces DeFi Development Costs: Multicoin Identifies Three Key Resources

The development of artificial intelligence (AI) reduces the cost of creating applications for decentralized finance (DeFi), but does not eliminate the main barriers to building a sustainable business. This conclusion was reached by Spencer Applebaum, a partner at the investment company Multicoin Capital, in an analytical article published on November 10, 2026.

According to the author, five years ago, developing a decentralized exchange or lending protocol could require about $3 million, a team of five engineers, and six months of work. Now, tools like Claude Code allow significant portions of a software product to be created over a weekend. However, this refers to development speed, not the system's readiness for safe commercial operation.

Applebaum believes that as software becomes cheaper, the main economic value will be concentrated in projects possessing three resources: access to clients, capital, and proprietary data.

Access to users allows for more revenue

The first competitive advantage the analyst mentions is distribution—the ability to attract clients and direct their trading operations to specific platforms.

An example is Hyperliquid, which hosts over $6 billion in USDC stablecoins, primarily used for ensuring derivatives trading.

Thanks to control over significant user balances, the platform was able to participate in the distribution of income from USDC reserves. In cooperation with Coinbase, about 90% of the corresponding yield, after adjustments, is directed to the Assistance Fund, which uses the funds to purchase HYPE.

According to Multicoin's calculations, with a certain volume of reserves, such a mechanism can provide about $185 million per year. This is an estimated figure, not confirmed protocol profit.

A similar principle operates in the crypto wallet market. For example, MetaMask charges a 0.875% fee for swaps through its interface. In comparison, the average trading fee on Uniswap over the past 12 months was about 0.13%, with a significant portion going to liquidity providers.

According to Applebaum, such differences show that control over client access can be more profitable than the direct technical execution of operations. However, this advantage is limited by the ability of users to quickly switch between wallets and trading platforms.

Capital and proprietary data remain important DeFi resources

The second advantage is related to access to capital. It is necessary for providing liquidity, lending, and covering losses after financial incidents.

Multicoin highlights two promising areas: DeFi prime brokerage, which allows collateral to be used for operations on multiple platforms, and risk insurance mechanisms for protocols.

Examples include the HLP liquidity pool in Hyperliquid and the Umbrella system from Aave, whose participants take on certain financial risks.

However, user capital can quickly leave the protocol in search of higher yields. Therefore, the analyst considers the platform's own funds, accumulated through retained earnings, to be a more sustainable resource.

The third advantage is information unavailable to competitors. Most cryptocurrency operations are public, so risk assessment models for lending against bitcoin or ether collateral are relatively easy to replicate.

The situation is different in insurance and lending of real assets, where closed information about borrowers, loss histories, and contract terms is necessary. According to Applebaum, combining such data with risk assessment experience can become a sustainable advantage for a financial project.

How AI will change competition between blockchains

The study also considers the possible weakening of the advantages of first and second-level blockchains (L1 and L2).

Previously, launching an application on Ethereum or Solana provided access to users and assets already in the respective network. However, the development of cross-chain transfers and the emergence of financial applications operating with multiple blockchains are gradually reducing projects' dependence on specific infrastructure.

Moreover, AI simplifies the transfer of software between different blockchain environments.

However, the author does not believe that L1 and L2 networks will necessarily lose economic value. Their revenues may increasingly depend on transaction volumes and demand for settlements.

As examples of new business models, Multicoin cites Ethena, which has developed a synthetic dollar product to $5.4 billion in assets under management, as well as projects working with lending secured by digital assets and graphics processors.

According to Applebaum, small teams are capable of competing with large financial organizations by entering markets that traditional companies have not yet served. However, the further success of such projects will depend on their ability to attract clients, form a sustainable capital base, and accumulate proprietary data.