Researching next-gen blockchain architecture (as of 2026) to achieve ultimate scalability in permissionless setting and fully resolve Blockchain Trilemma. May or may not succeed but must be fun.
The current status is heavy WIP with somewhat regular updates on the website. Read the book for architecture details and check the code for details otherwise. Most of the things are missing right now, but that'll change over time.
While this repository is focused on blockchain R&D, there are some useful Rust crates that others might find useful too:
ab-aligned-buffer- efficient abstraction for memory buffers aligned to 16 bytes (u128) with both owned and shared variantsab-blake3-const fnand GPU-friendly (rust-gpu) BLAKE3 primitivesab-chacha8- small GPU-friendly (rust-gpu) software implementation of ChaCha8ab-direct-io-file- cross-platform APIs for working with files using direct I/Oab-io-type- infrastructure for zero-cost zero-copy serialization/deserializationab-merkle-tree- high-performance Merkle Tree and related data structures (Merkle Mountain Range, Sparse Merkle Tree) using BLAKE3 (can be generalized to other hash functions if necessary)ab-riscv-primitives- composable RISC-V primitives (instructions, registers) and abstractions around themab-riscv-interpreter- composable and generic RISC-V interpreter
Many of these crates are no_std, do not require an allocator and make efficient use of stack through const generics.
Where possible, no-panic is used to guarantee absence of panics at compile time for high reliability and more
efficient code generation. None of the listed crates have any dependencies on anything in this repository that is
specific to this project, so can be used externally with minimal dependencies.
Note that these crates may use experimental Nightly Rust features to achieve their goals, and as a result, most of them do not work on stable Rust (yet).