Why Gateway pivoted from ZK to MPC

When we started Gateway, we assumed zero-knowledge proofs would be the right long-term primitive for credential privacy. We were right on cryptographic elegance, but wrong on product timing.

What changed

The first hard constraint was latency. Issuers and verifiers did not care how elegant the circuit design was if issuance blocked user onboarding. The second was developer ergonomics. Most partners needed a secure default they could integrate in weeks, not months.

Why MPC won

MPC gave us a path to production reliability under pressure:

• Faster iteration loops for protocol and backend changes.
• More predictable performance across real workloads.
• A deployment model partners could understand and audit.

The point was not that ZK is bad. The point was sequencing. We optimized for the constraint that mattered most at that stage: shipping a system people could trust in production.

What I learned

Architecture choices are rarely permanent. They are usually timed bets. A good decision is often the one that keeps optionality while solving the current bottleneck.