Author: Eddy Lazzarin (CTO of a16z Crypto)

Translation: Deep Tide TechFlow

Deep Tide Overview: The rise of open-source software has completely disrupted the development paradigm with "composable code," allowing developers to quickly integrate on top of mature modules without reinventing the wheel. However, as large language models push the marginal cost of writing software toward zero, "vibe coding" significantly lowers the barrier to replicating application layer interfaces, rapidly diminishing the scarcity of code itself, and traditional software moats are collapsing. Eddy Lazzarin, CTO of a16z Crypto, points out that the core of current productivity restructuring is shifting from reusable code to "reusable networks."

When software supply is rampant, what is truly scarce is underlying embedded assets: deep liquidity, immutable reputation, interoperable identity graphs, and settlement credit validated by consensus across the entire network. As a shared state layer of internet norms, blockchain constitutes an irreplaceable structural barrier for decentralized ledgers. Interfaces and code can be cheaply replicated, but it is the reusable networks that bear the historical coordination and consensus value that represent the ultimate moat in the age of intelligent agents.

Open source has fundamentally reshaped the software industry by endowing code with composability. Developers no longer need to build everything from scratch; they can assemble complete applications directly from existing code libraries, frameworks, and protocols.

Today, as artificial intelligence drives the cost of software production toward zero, the essence of composability is shifting: it is no longer just about reusable code but about reusable networks.

Software itself is no longer a moat; the network is.

Whether in the field of cryptocurrency or from a broader perspective encompassing numerous software developers and startup teams across industries, composability remains one of the most compelling core ideas.

Early decentralized finance (DeFi) protocols (e.g., Uniswap) allowed new applications to be built directly on existing liquidity without having to rebuild liquidity pools from scratch, vividly demonstrating the powerful potential and vast prospects of composability.

However, this trait has never truly become the most decisive mainstream narrative in the cryptocurrency industry. The fundamental reason lies in the excessively high costs of trial, error, and experimentation in innovation.

Building on-chain applications requires highly vertical expertise in smart contracts, security protocols, code audits, and blockchain infrastructure. Exploring this vast design space demands tremendous amounts of time and capital, so the group of builders able to participate has always been quite limited.

The advent of AI has completely disrupted this situation. Now, anyone with an idea, including non-programmer builders without technical backgrounds, can generate a runnable prototype in a matter of hours, complete integrations with existing protocols, and achieve product iterations that once took months within just a few hours.

Revisiting the Most Ambitious Ideas in Cryptocurrency

With a cliff-like plunge in the cost of software construction, the number of innovative experiments is witnessing explosive growth. This is the perfect time for us to revisit some of the most ambitious ideas in the cryptocurrency space:

• NFT (Non-Fungible Tokens): Created digital native assets, including collectibles and art, event tickets, real-world assets (RWA), community identity credentials, and intellectual property (IP), which all applications can seamlessly recognize and expand functionalities upon.

• DAO (Decentralized Autonomous Organization): Provides an unprecedented experimental ground for internet-native organizational collaboration and new forms of on-chain governance.

• Decentralized social networks: Unleash fully portable digital identities and social graphs controlled by users.

• On-chain games: Truly transform the virtual world into a fully open underlying infrastructure and a thriving internet economy.

Only when a vast number of real users and numerous AI agents choose to continuously build on these networks can the above ambitious ideas truly unleash their immense power.

AI can almost replicate and generate any application. It can clone front-end interfaces in minutes, rewrite API interfaces, and produce entirely new software products. AI not only makes software development extremely inexpensive, but it also fundamentally disrupts the definition of scarcity in economics.

As software is becoming exceedingly abundant, the networks themselves are anything but.

Imagine a scenario where building crypto wallet software becomes as easy as clicking a button, allowing traders to generate a personalized wallet that belongs solely to them; even as AI agents execute on-chain transactions directly on behalf of users, the traditional wallet software interface may completely disappear.

In such a future world, the importance of front-end interfaces will fall far behind the underlying networks they are reliant upon. The truly scarce assets are no longer "the best wallet client," but rather access to the deepest liquidity pools, the most powerful historical reputational endorsements, the most robust identity relationship graphs, and the most trusted settlement networks recognized by the entire network.

AI can easily generate an NFT, but it can never create the historical provenance behind that NFT, the community consensus cohesion, or the extensibility to other external networks.

AI cannot create your on-chain digital identity, your credit reputation, your social relationship chain out of thin air, nor can it generate the vast liquidity embedded within the protocol. It also cannot replicate the historical sediment, multi-party coordination, and consensus trust that a network accumulates over many years.

You can never rely on "vibe coding" to conjure up a USDC.

But the Internet Needs a Shared State

Perhaps this is the ultimate mission that blockchain has been destined to undertake since its inception: to become the most authoritative and normative shared state layer across the entire internet (Canonical Layer of Shared State).

As upper-layer applications gradually devolve into commoditized goods, the entire technology ecosystem's value capture points must inevitably shift to the networks that these applications depend on. In a future world filled with AI agents, this unified global shared state will be more indispensable than ever before.

Front-end interfaces can be instantly copied and cloned, and software code can be rapidly reconstructed repeatedly by AI. However, the underlying core states, including asset ownership, identity verification, liquidity depth, reputation systems, and network collaboration relationships, still possess an irreducible absolute scarcity.

For today's developers and entrepreneurs, these "reusable networks" with strong network effects represent the grandest historical opportunity of the AI era.