LayerZero has released Akita, a post-quantum cryptographic scheme now running inside a16z crypto's Jolt prover, cutting proof sizes to as little as 65 kilobytes.
Key Points:
- Akita is the first production-ready lattice-based commitment scheme built for zero-knowledge proving systems.
- a16z crypto rebuilt its Jolt virtual machine around Akita and shipped the result as Lattice Jolt.
- The swap makes proving two to three times faster and roughly halves the memory a prover needs.
Akita Cuts Post-Quantum Proof Sizes
The interoperability firm announced Akita on Sept. 9 in a blog post that called it the first production-ready primitive of its kind, built with researchers at Carnegie Mellon University and the University of Southern California.
The scheme handles a single job inside a proving system, committing to a record of a computation and then proving to an outside party that the record is correct. Swapping that one piece for a quantum-secure version protects the whole stack, because the remaining components hold up on their own.
Everything else in Jolt stayed put.
a16z crypto folded Akita into its open-source zero-knowledge virtual machine and shipped the rebuild as Lattice Jolt, dropping Dory, the elliptic-curve scheme that Jolt had relied on since its launch. Proofs now measure 65 to 80 kilobytes, against 200 kilobytes or more for the hash-based systems that carry nearly all post-quantum proving in production, a reduction of two to eight times. The code is open source, and Jolt underpins Zero, the blockchain LayerZero is building toward mainnet.
Also Read: Ethereum Faces $2,800 Supply Wall After ETF Inflows Jump To $216M
Why A16z Crypto Backs Lattice Math
Justin Thaler and three colleagues at the venture firm argued that lattice assumptions beat hash-based designs on both speed and size, and that the hash route quietly carries assumptions of its own. Akita rests on Module-SIS, a problem NIST already standardized for post-quantum signatures and key exchange.
Elliptic curves forced Jolt to work over 256-bit fields, while lattices reach comparable security at half that width, and the prover spends most of its time multiplying those numbers together. Lattice Jolt clears 2 million processor cycles per second on a laptop, up from roughly 1 million, and passes 10 million once Apple graphics acceleration is switched on. Memory use per cycle dropped from about 300 bytes to 200.
One gap remains, because the build is not yet fully zero knowledge, a property that privacy applications require.
Crypto Races A 2029 Quantum Deadline
Ethereum (ETH) developers set December 2029 as their deadline for a quantum-resistant base layer this month, matching migration targets already adopted at Google, Cloudflare and Microsoft. Coinbase gathered developers, custodians and hardware wallet specialists at Stanford the same week to weigh signature options for Bitcoin (BTC), and the closed session ended without agreement on any single path. That chain still carries no formal timeline of its own, which leaves proving systems and infrastructure teams to move first.
Read Next: Nvidia's Jensen Huang Says Security Vendors Benefit From AI Threat Panic
