Suddenly, I have a feeling that blockchain public chains and AI large models are somewhat similar; both are undergoing a lot of redundant construction and serve as the underlying layers of these two industries. Discussing this with GPT was quite profound: Similarities 1. Both are general-purpose infrastructures, and competition is inevitable. Whether it's large models or public chains, they essentially belong to foundational general technologies. The core characteristics of such technologies are: high entry barriers + high potential returns + ecosystem-based competition. Therefore, every country, company, and fund hopes to have its own version; even if the underlying layers are similar, they must be rebuilt to establish independent control. 2. Both involve "redundant construction," but for reasonable motives. What seems redundant is actually a necessary differentiation of technology, computing power, compliance, and interests. Different companies retrain under the same architecture (Transformer) because of differences in data privatization, computing power distribution, and fine-tuning directions; public chain projects are rebuilt under the premise of EVM compatibility to meet different performance, regulatory, and economic models. The "redundancy" reflects not waste, but the natural result of system diversification. 3. Both are shifting from "creating technology" to "creating ecosystems." The technological barriers have been breached, and the real competitive focus has shifted to ecosystem integration and user retention. Large models compete in application ecosystems (Copilot, AI Agent, enterprise embedding scenarios); public chains compete in asset ecosystems (DeFi, RWA, L2 scaling, gaming applications). After technological homogenization, ecosystem capabilities determine long-term differences. Differences 1. The concentration of technical architecture differs. The trend for large models is the centralization of computing power and capabilities— the number of leaders is decreasing, and the model scale and inference costs are extremely high, ultimately forming a pattern of a few giants; public chains, on the other hand, are modular and layered—Ethereum becomes the underlying consensus layer, with numerous L2, L3, and application chains emerging above it, forming a networked competitive structure. 2. The role of open source differs between the two. Open-source models (like LLaMA, Mistral) mainly lower the entry costs but do not change the pattern of computing power and data centralization; whereas the open source of public chains (OP Stack, Polygon CDK) truly expands the space for sovereignty and economic innovation, allowing more teams to build new ecosystems on the Ethereum framework. 3. The direction of commercial logic implementation differs. The main line for large models is service commercialization—API subscriptions, enterprise solutions, vertical integration; the main line for public chains is capital financialization—token incentives, ecosystem investments, liquidity economy. The former is driven by cash flow, while the latter is driven by asset pricing. Conclusion > Large models and public chains are indeed similar: both are redundantly constructing the underlying layers and competing for ecosystem dominance. However, the paths have diverged: large models tend toward centralization and commercial reuse of open source, while public chains tend toward modularization and decentralized reuse of Ethereum. One pursues a unified intelligence, while the other pursues diverse value.