Comparing AI and Crypto Assets: Distinct Development Paths of Layered Strategies

Recently, there have been voices pointing out that Ethereum's Rollup-Centric strategy seems to have failed, expressing strong dissatisfaction with this L1-L2-L3 nested model. However, interestingly, the developments in the AI field over the past year have also undergone a similar rapid evolution of L1-L2-L3. By comparing these two fields, we can explore the root causes of the issues.

In the field of AI, hierarchical logic means that each layer addresses core problems that the previous layer could not solve. Large language models at L1 resolve basic language understanding and generation capabilities, but there are notable deficiencies in logical reasoning and mathematical computation. The reasoning models at L2 specifically tackle this shortcoming, with certain models capable of solving complex mathematical problems and debugging code, thus filling the cognitive blind spots of large language models. Building on this, the AI agents at L3 naturally integrate the capabilities of the previous two layers, transforming AI from passive responses to active execution, enabling it to autonomously plan tasks, invoke tools, and handle complex workflows.

This hierarchy is one of "progressive capabilities": L1 lays the foundation, L2 addresses deficiencies, and L3 integrates. Each layer achieves a qualitative leap based on the previous layer, and users can clearly feel that AI is becoming more intelligent and practical.

In contrast, the layered logic in the cryptocurrency field is that each layer provides solutions to the problems of the previous layer, but at the same time brings about new and larger issues. For example, L1 public chains have insufficient performance, which naturally leads to the consideration of using L2 scaling solutions. However, after a round of internal competition in L2 infrastructure, although Gas fees have decreased and TPS has cumulatively increased, liquidity has become fragmented, and ecological applications are still scarce, resulting in an excess of L2 infrastructure becoming a major problem. Thus, L3 vertical application chains began to be developed, but these application chains operate independently, unable to enjoy the ecological synergy of general-purpose chains, which instead makes the user experience even more fragmented.

This layered evolution has led to a "problem transfer": L1 has bottlenecks, L2 provides patches, and L3 appears chaotic and fragmented. Each layer seems to merely transfer the problems from one place to another, giving the impression that all solutions are merely aimed at "issuing tokens".

The fundamental reason for this difference lies in the fact that AI layering is driven by technological competition, with major companies striving to enhance model capabilities; while cryptocurrency layering seems to be constrained by token economics, with each L2's core KPI focused on Total Value Locked (TVL) and token prices.

Essentially, one field is focused on solving technical problems, while the other is more about packaging financial products. There may not be a clear answer as to which is right or wrong; it depends on individual perspectives and positions.

Of course, this abstract analogy is not absolute; it is merely an observation of the interesting differences and points of reflection found when comparing the developmental contexts of the two fields.