Trusted Execution Environments (TEEs)

A trusted execution environment (TEE) is a secure section of a computer's main processor that keeps certain code and data isolated from everything else running on the machine, including the operating system, other applications and, in many designs, the system administrator. Data inside a TEE is encrypted in memory and can only be processed by the authorized code, and the hardware can produce a cryptographic report, called an attestation, that proves to an outside party exactly what code is running inside. TEEs are built into most modern phones and many server processors.

TEEs are a leading technology in confidential computing, the protection of data while it is being used, as opposed to while it is stored or sent. Well-known implementations include Intel Software Guard Extensions (SGX) and Trust Domain Extensions, AMD's Secure Encrypted Virtualization, Arm TrustZone in mobile chips and Apple's Secure Enclave. In crypto, TEEs have become a tool for privacy, key management and trading infrastructure, offering an alternative or complement to purely cryptographic techniques.

Imagine a bank branch with an ordinary office where staff work, and inside it a sealed glass booth with its own lock. Documents passed into the booth can be read and worked on only by the specialist sitting inside. Even the branch manager cannot open the booth or read the papers, and the booth has a certificate on the door, issued by its manufacturer, that anyone can check to confirm who is inside and what they are allowed to do.

A TEE is that booth inside a computer chip. A program can hand sensitive information, such as a password, a private key or medical records, to code running in the booth. The rest of the computer, even if it has been hacked, sees only scrambled data. The trade-off is that users must trust the chip maker that built the booth and the certificate on its door.

• Isolation. The processor reserves protected memory regions, sometimes called enclaves, which other software cannot read or modify.
• Memory encryption. Data leaving the processor for main memory is encrypted with keys that never leave the chip.
• Remote attestation. The hardware signs a measurement of the code loaded into the enclave, so a remote user can verify that the expected program, and not a tampered version, is running before sending it secrets.
• Sealing. Enclaves can encrypt data so that only the same code on the same hardware can later decrypt it.

TEEs differ from a hardware security module, which is a separate, dedicated device for storing keys, though both aim to protect secrets in hardware.

On phones, TEEs protect fingerprint and face data, encryption keys and payment credentials, supporting mobile wallets and banking apps. In cloud computing, they let firms process sensitive data, such as financial or health records, on servers owned by someone else. Banks and payment firms use them to protect cryptographic keys and run fraud models on pooled data without revealing it.

In crypto, TEEs are used in several ways:

• Wallets and custody. Some wallet providers use enclaves to generate and store private keys or to run parts of multi-party computation schemes.
• Private smart contracts. Some blockchains run smart contracts inside enclaves so that inputs and state stay hidden from node operators.
• Block building. Projects linked to Flashbots have used TEEs to build blocks without letting operators see or front-run pending transactions, as a way to manage maximal extractable value.
• Oracles and AI agents. Enclaves can prove that off-chain data was fetched or that an automated agent ran unmodified code.

Other privacy technologies rely on mathematics rather than hardware. Zero-knowledge proofs let someone prove a statement without revealing the data behind it, and fully homomorphic encryption allows computation directly on encrypted data. These approaches avoid trusting a chip maker but are usually far slower and more complex. TEEs offer near-normal speed today, which is why many projects combine them with cryptographic methods as part of a wider toolkit of privacy-enhancing technologies.

The security of a TEE depends on the hardware vendor's design and on keeping its attestation keys safe. Researchers have repeatedly found weaknesses, especially side-channel attacks that infer secrets from timing, power use or processor behaviour; several published attacks extracted data from Intel SGX enclaves, and in 2022 researchers showed that such flaws could have exposed secrets held by a privacy-focused blockchain that relied on SGX. Vendors issue fixes through firmware and microcode updates, but older hardware may stay vulnerable. TEEs also concentrate trust in a few chip makers, conflicting with the trust-minimizing goals of public blockchains, and attestation services may depend on the vendor being online. Supporters argue that, used carefully alongside other defences, TEEs significantly raise the cost of attacks.

• Confidential Computing Consortium