Is x402 really optimizing payments, or just hiding the complexity of Web3?

x402 is indeed multi-chain and multi-currency at the protocol level, which is fine. But in practical operation, it almost entirely relies on the facilitator provided by Coinbase. In other words, it now functions more like "a payment layer hosted by Coinbase" rather than a truly neutral protocol network.

Let's talk about Gas fees.

Gas hasn't disappeared; it's just paid elsewhere. Now, the facilitator front-runs the gas for all requests. A single transaction on Base costs about a few cents, which doesn't sound like much, but at scale—say 1 million API requests—that's several tens of thousands of dollars in real costs.

The question is, who will pay for this long-term?

Will it be merchants? Users? Or will it be monetized through data, traffic, and behavioral profiling?

Regardless of the path, the core issue is that costs are concentrated in a centralized role, rather than being eliminated.

Now, about confirmation times.

Signatures are instant, but actual payment confirmation still depends on the finality of the chain.

Base takes about 2 seconds, Ethereum mainnet over 10 seconds.

This means HTTP connections must stay open during this period.

In an ideal network, this might be acceptable, but on mobile devices, long agent chains, or unstable network environments, this is a very counterintuitive design.

If this 2–15 second window is interrupted, the client enters an awkward state: Did I actually pay successfully or not?

Retrying might lead to double charges. Not retrying could mean the money was paid but the request didn't complete, so an additional "status query" layer is needed.

And once you need to check the status, you again introduce RPC, indexing services, or continue relying on the facilitator.

This is already a significant departure from the original simplicity of HTTP.