The era of free bidding proved by ZK has arrived: from centralized constraints to infinite boundaries of decentralized computing In the widespread implementation of Rollup, DID, and AI generated content, ZK proof is gradually becoming an indispensable trust tool. But generating an effective zero knowledge proof often means high-intensity calculations lasting from seconds to minutes. The current dilemma of ZK track: Who will do this? How do I pay? How to ensure its timely delivery? Currently, most projects choose to bind a centralized prover or build their own complex ZK infrastructure. This choice is costly, lacks flexibility, and introduces obvious trust bottlenecks. ZK is currently becoming another 'miner's privilege', gradually moving away from the decentralized spirit it should represent. Based on this, Raspberry @ boundless_xyz proposed a new solution: Transform the ZK proof task into a service marketplace that allows for free bidding, where any requester can issue a task, set a budget and timeline, and invite interested Provers to bid and execute it, with encrypted proofs ensuring performance. So, will Boundless solve the problems currently faced by traditional ZK-as-a-Service? Is it possible for it to become the "computational coordination layer" of modular blockchain? ZK is no longer 'free': proof that it is devouring on chain computing ZK proof is rapidly expanding its application boundaries: Rollup, DID, privacy payments, AI auditing, on chain game verification... but its costs are becoming increasingly apparent. For example, generating a zkSNARK proof may require 1-10 seconds of CPU/GPU resources, and using zk STARK proof volume provers may require longer time, higher memory, or specialized hardware (FPGA, GPU cluster) to complete. And verifying a proof only requires a few gas units on the chain. But who will bear the responsibility for generating this proof? At present, the mainstream approach is: ❶ Self built prover services (such as zkSync, Starknet); ❷ Use closed APIs (such as the provisioning end of Polygon zkEVM); ❸ Or rely on a set of licensed nodes to execute (such as Scroll). These methods cannot solve two core problems: The task requester does not have the right to choose; Task executors do not have a bidding mechanism. In short, the execution end of ZK is currently "closed and unpriced". This goes against the market spirit of blockchain. Boundless model: ZK computing task becomes public auction Boundless does not directly "provide computing power", but acts as a coordinator to turn ZK proof generation tasks into a competitive service transaction. The mechanism design is closer to DePIN and on chain order book system. I have also introduced the specific process. Here is a brief explanation of the roles and functions involved in the process, as shown in the figure: The core advantage of this mechanism lies in: The service price is determined by the market and not specified by the platform; Enforcers have no licensing threshold, reducing barriers to participation; Budget - Delay - Cost feedback is generated to automatically adjust system load and efficiency. The implementation of ZK has never been a victory of algorithms, but a long-term game between system design and incentive structure. Boundless proposed an attempt to replace central scheduling with market mechanisms and enhance execution reliability through responsibility games. This solution is not omnipotent, as it still poses challenges to network stability, arbitration efficiency, task separability, and so on. But it does show us a direction: ZK computing can not only be self built and hosted, but also be auctioned, scheduled, and traded like an order book. This is not the end, more like a starting point