Does it matter if a new AI project is an L1 or an L2? Binance’s CZ answers

On Thursday, Binance co-founder Changpeng Zhao (CZ) started a debate on social media platform X, discussing whether developers should build their AI-focused blockchain projects on their own Layer 1 (L1) networks or rely on existing Layer 2 (L2) solutions.

Through his official X account, Zhao said the query crossed his mind after having a conversation with a project founder. He surmised that the main goal of AI-driven blockchain projects is not to create a new blockchain but to “use blockchain technology for AI-related economic models.”

“Having your own L1 feels like you have more sovereignty, more decentralization, etc. But it’s also a lot more work. Nodes, validators, etc. You are also kinda on your own island,” the ex-Binance CEO wrote. 

“L2 saves you all that work. And doesn’t seem to have much value leakage to the L1 you are on (like ETH). You get to use the DEX, perps, existing tools, and the community.”

Concluding his thoughts, Zhao asked the crypto community which of the two is considered “superior. ” He acknowledged that it was an aged debate but wondered if there is a clearer winner now.

Decentralization vs. practicality: Community discusses

L1 networks, like Bitcoin and Ethereum, are independent blockchains that handle all transactions and security on their own, processing transactions directly and requiring validators or miners for consensus. 

L2 solutions such as Ethereum’s ZK-Rollups operate on top of existing L1 chains and process transactions off-chain or in batches before settling on the main blockchain to reduce congestion and fees. L2s allow projects to use pre-established ecosystems without the need to maintain a network.

Building L1 is more like building an apartment building. You need to maintain the building, employ security and find real estate agents who then find you tenants to make the building viable.

Similarly, L2 is like buying one of those apartments to rent it out further. Smaller…

— Annkurr (@ghaiankur) March 6, 2025

In response to CZ’s question, an account using the pseudonym Silent Capital insisted that the decision should depend on what the project’s goals are. According to the user, L2 networks are innately decentralized because they leverage an existing blockchain network.

According to Silent Capital a users own L1 won’t give the same level of decentralization.

They also talked about how, in order to build a new L1, a developer would have to make almost everything from scratch. They have to cover consensus mechanisms, validator networks, and security models.

Marketing researcher Clytheronix also gave a similar perspective, insisting that L1 networks tend to be more decentralized but come with greater complexity. 

L1 networks haven’t lost their relevance

Still, not everyone agrees with the idea that L2 is the best option for AI projects. MatrixAI community head Sercan Muslu listed some benefits of L1’s sovereignty, asserting that AI applications require a level of optimization that existing L2 solutions fail to provide.

“L1 sovereignty isn’t just about control [it] helps optimize consensus mechanisms for AI workloads. [In addition, it] integrates AI models natively into the chain and improves scalability without external dependencies,” Muslu remarked. 

The community head concluded that L2 solutions feel more plausible, but they depend on the L1 parent chain’s architecture and are not designed to handle AI’s high computational demands.

Some crypto enthusiasts like Music artist Marco Tonetti went beyond technical considerations and saw the choice between L1 and L2 through the scope of making a business strategy. 

 “L2 is way better if you want to build a product. L1 is better if you want to gamble your luck and become richer faster,” he claimed.

Cryptopolitan Academy: Coming Soon - A New Way to Earn Passive Income with DeFi in 2025. Learn More