In 1992, sterling fell to a short position the Bank of England could not outspend. India's first tokenized bonds sit on a ledger where the RBI sees every position. That protects the rupee. It also keeps the world's capital out.

Canton keeps data private through domain isolation between known, legally accountable validators, Goldman Sachs, BNY Mellon, and a handful of others control what gets seen and by whom. That's enforcement by contract and reputation, and it holds inside one jurisdiction. Cross-border, multi-regulator settlement needs a guarantee that doesn't depend on every party staying cooperative.

Inco pairs Intel TEEs for fast execution with an FHE path for the fully trustless case, and ships EVM compatibility to lower the onboarding barrier for existing Solidity developers. The tradeoff shows up at the edges: TEE security rests on Intel's hardware not being compromised, Spectre and SGX side-channel exploits are documented, not hypothetical and EVM itself has no primitives for encrypted state.

Miden's client-side proving model lets a user compute locally and submit only a STARK proof that the computation was correct — the chain verifies the proof, never the inputs. That protects what was computed, not the values inside it. An encrypted join across two counterparties' balances is outside what a proof-of-correctness system was built to do.

CCCM validates in parallel cube clusters and orders blocks against proofs rather than raw transaction contents, so a validator confirms that a computation was done correctly without ever seeing what was computed. No mempool means no place for a bot to sit and watch pending transactions, frontrunning needs a queue to front-run, and this design removes the queue.

A PMB is an isolated execution environment scoped to one enterprise: its own encrypted dataset, its own access policy, its own Merkle-anchored state proof, with cross-PMB calls available where needed. That's a materially different unit of infrastructure than a shared public mempool where every participant's transaction sits in the same queue.

A loan approval touches a customer record, an account, an entity relationship, and a jurisdiction-specific compliance rule, four joined tables, minimum. Blockchains store state as flat key-value pairs or Merkle dumps with no native concept of a join, which is why "put it on-chain" has stalled at the pilot stage for almost every real enterprise workflow.

L2FhEX runs SQL-style joins, filters, and aggregations over encrypted state, with threshold reveal for the narrow cases where a result actually needs to surface. Getting there requires solving encrypted MVCC concurrency, FHE circuit compilation from query plans, and noise management across chained operations, three separate hard problems that have to be solved together, not sequentially.

FHE performance improved on roughly five-year cycles from Craig Gentry's 2009 thesis through 2018. GPU-accelerated libraries — TFHE, OpenFHE, Microsoft SEAL then cut compute costs by close to four orders of magnitude between 2020 and 2024. That performance curve, not a change in appetite, is why confidential computation at blockchain scale went from a research paper to a deployable system in the space of one crypto cycle.

Cubane runs computation directly on ciphertext using Fully Homomorphic Encryption, so the arithmetic that touches a balance, a diagnosis, or a portfolio position never sees plaintext at any point in the pipeline. Encrypting data at rest and decrypting it to run a query, the standard cloud pattern today, still exposes the data during the one moment it matters most.

Financial institutions spend roughly $206B a year on KYC, AML, sanctions screening, and transaction monitoring, compliance work that now eats close to 19% of a typical bank's annual revenue. Most of that spend exists because current infrastructure can't produce a cryptographic proof of compliance; it has to be reconstructed by hand, after the fact, by people.

Under HIPAA, a hospital can't put patient records on a chain where anyone with a node can read them. Under GDPR, "right to erasure" runs directly into a ledger designed never to forget anything it's written. These aren't configuration problems solvable with an access-control layer bolted on top, the base protocol guarantee is the wrong shape for the workload.

Public chains make every state transition globally readable by design, that's what lets a stranger verify a validator without trusting them. Permissioned chains close that visibility by restricting who can read the chain at all, which is the same move that kills the network effects a shared ledger exists to create.

TradeLens, the Maersk/IBM shipping-logistics chain, shut down in 2022 after years of pilots. @Consensys Codefi wound down the year after, despite owning both the Ethereum-based platform and the application layer on top of it. Same company, same stack, still no path to a sustainable product. That rules out "wrong vendor" as the explanation.

R3 launched Corda with over 200 financial institutions signed on, including @GoldmanSachs and @jpmorgan as founding members. Both left within a year of joining. A shared ledger among direct competitors runs into the same governance question every time: who gets to see what, and who decides.

L2FhEX is cubane's execution layer. smart contracts run on encrypted state through fully homomorphic encryption. the contract computes on ciphertext and never reads the underlying values. the result still verifies.

"Privacy or compliance, pick one" was a limit of the cryptography, not a rule of finance. FHE lets you compute on data without decrypting it. cubane makes that the execution model, not an add-on.

Regulated data has always forced a choice: keep it private, or make it verifiable. cubane is a confidential L1 that removes the choice. queries and joins run directly on encrypted data. testnet q1.2026.

Encryption protects sensitive data while stored and moving. The harder problem starts when someone needs to compute over it. Portfolio analysis, patient record comparison, fraud detection, and financial reconciliation all require computation over sensitive information. Conventional systems typically need access to usable data during processing, creating a confidentiality gap at the exact point where the information becomes useful. Cubane's L2FhEX uses FHE to perform computation over encrypted values. Protecting data during computation is becoming an infrastructure requirement for institutional blockchain.