The privacy technologies, compared.
An honest, neutral comparison of the main approaches to on-chain confidentiality, and where each one fits.
Last updated August 11, 2026
An honest, neutral comparison of the main approaches to on-chain confidentiality.
| Technology | How it works | Latency | Best fit | Sources |
|---|---|---|---|---|
| Garbled Circuits (GC-MPC) · Bubble | Compute on encrypted data via garbled circuits + secret sharing | Low | Confidential finance, portfolio & transactions at institutional scale | gcVM paper, measured benchmarks |
| FHE | Compute directly on ciphertext | Higher | Arbitrary computation where latency is less critical | TFHE paper, Zama litepaper |
| ZK proofs | Prove statements without revealing inputs | Medium | Validity proofs, private state transitions | Aztec on private state, client-side proving |
| TEE | Compute in trusted hardware enclaves | Low | Performance-sensitive workloads with a hardware-trust assumption | SGX.fail, Intel attestation |
Methodology
This comparison is compiled from primary sources. Rows describing Bubble link to the gcVM paper and to benchmarks we measured ourselves (methodology and raw results available on request); rows describing other technologies link to those projects' own publications and documentation. We aim for vendor-neutral wording and welcome corrections through the contact page.
See where GC-MPC fits your stack.
Read the research, or talk through the trade-offs with our team.