Verifiable Calculation - RiscZero CEO Podcast Interview The rapid development of AI enhances productivity, but also reduces the cost of counterfeiting/fraud. More RWAs will be put on the chain, but if the assets on the chain are systematically verified and audited, without verification guarantees, these RWA assets will face great risks. I remember there was a wave of CEX asset zk audits in 23-24, but in the future, more RWAs will be put on the chain, and asset verification and tracking audits will be more needed. In the future, the verifiability and program transparency of the Internet will become key features. The potential for the combination of blockchain and AI is enormous, and verifiable computing is the core of infrastructure. In verifiable computing, the zk projects @ RiscZero and @ boundless_xyz implement verifiable computing through RISC-V and ZK VM. Attention since 2003 RiscZero has been developing from zkvm to specific zk L1. Below is a podcast interview with their founder @ zeroknowledgefm to gain a deeper understanding of the team and project's technology choices, business ideas, and more. This interview was conducted before the release of @ boundless_xyz, but previous interviews have already revealed L1's plans. If the English listening is good? Listen directly to the original version: https://zeroknowledge.fm/podcast/251/ Interview Summary: The core technology of Risc0 is ZK VM and RISC-V: Risc0 builds a ZK VM that runs RISC-V code, supports programs compiled in languages such as C, C++, Rust, etc., similar to virtual microcontrollers, with zero knowledge properties (hidden inputs and programs). Choose RISC-V: Due to its open source, no intellectual property restrictions, simplicity and efficiency, as well as compliance testing and formal model support, it is suitable for implementation in ZK circuits. Privacy and Scalability: Supports zero knowledge proof of program execution and the program itself, initially focusing on scalability to support complex application ecosystems, with privacy (such as ZK identity) as the future direction. Memory and Uncertainty: Memory is represented by 32-bit numbers and encoded with Merkle structure to represent the initial state; Support non deterministic data requests, guest (ZK VM) requests data from host and verifies it to maintain privacy. Technical Implementation and Acceleration of STARKs Applications: Using STARKs to prove repetitive structures (such as processor time steps), suitable for VM's temporal execution model. Recursive support: In development, Rust validators can be run within ZK VM to validate proofs and implement incremental proofs (such as from genesis blocks to N+1 blocks). Hardware acceleration: Currently, it reaches 30000 cycles per second on the M1 Pro, with an estimated 1 million cycles per second for GPU acceleration (Metal/CUDA) and approximately 10 million cycles per second for the 3090 GPU. Support accelerator circuits such as SHA-256, and plan to add finite field and large integer modular multiplication accelerators in the future. Parallelization: Proving that the process can be parallelized, suitable for GPU acceleration, reducing latency, and supporting large-scale computing (such as compiler execution). Vision and Applications Cloud 2.0: The goal is to build a decentralized public cloud that supports DeFi, games, NFTs, and complex applications such as SQL databases and social media, achieving "transaction costs as low as no need for measurement". L1 blockchain: Develop L1 and use ZK optimized consensus mechanism ("transaction proof") to determine the longest chain based on computational complexity rather than transaction count, supporting billions of transactions per second. Use cases: including supply chain security (proving that binary code comes from audited source code), physical asset tracing, etc. Ecological and Future Ecological Connection: We plan to integrate with other blockchains (such as Ethereum) through bridges or validators, supporting the verification of Risc0 proofs or running EVMs on Ethereum. Roadmap: The current version is open source, and the next version (including recursion and GPU acceleration) will be open source soon. We plan to launch DevNet by the end of 2022 or early 2023, and register for testing through platforms such as Twitter. Does Risc0 exist as L1 and connect to other networks or network clusters? Things like @ VitalikButerin's eth, @ Gavinwood's @ Polkadot, Cosmos, etc? Brian said he will almost certainly connect. Regard Risc0 technology as a widely applicable tool. Although we have our own L1 vision, we also hope that this technology can serve other ecosystems. Challenge and Reflection on Programming Usability: Lowering the threshold for developers through existing languages and toolchains, without the need to learn new languages or arithmetic circuits. Prospects for hardware acceleration: In the short term, GPUs are dominant due to immature mathematics and long ASIC development cycles; FPGA and ASIC may be applied in specific SNARK circuits. Competition and positioning: Unlike other ZK VM/EVM projects, Risc0 emphasizes generality and cloud level scalability, balancing the Ethereum ecosystem with other networks. Risc0 uses RISC-V and STARKs to build ZK VM, solving the scalability problem of distributed systems. The goal is to decentralize public clouds and support large-scale, privacy protected computing. The current open-source version is available, and GPU acceleration and recursion support are about to be released. We plan to build L1 and integrate it with other ecosystems, suitable for DeFi, supply chain security, and other scenarios Complete interview content: Long article pictures