Proof in Silicon: How Hardware Is Reinventing Trust in Crypto

Why TEEs Are Reshaping MEV Infrastructure, Privacy, and Decentralization

May 20, 2025

In the evolving arms race of MEV (Maximal Extractable Value), where milliseconds and confidentiality define the winners, infrastructure must do more than scale—it must earn trust. Trusted Execution Environments (TEEs), once dismissed for their security concerns, are now being reimagined as foundational primitives for high-performance, decentralized computation.

In the 3rd MEV Space Twitter session, Shea Ketsdever (Flashbots) and Chua Zheng Leong (Automata Network) offered a deeply technical and candid conversation on how TEEs are transforming block building, privacy models, and protocol design.

MEV Space brings together the sharpest minds in crypto to dissect the evolving MEV landscape.

From cross-domain sequencing to bottom-of-block arbitrage, the two speakers highlighted how TEEs can rebuild both the trust and performance layers of the crypto stack. Their insights point not to an abstract future, but to systems already in production—BuilderNet, multi-party computation (MPC) hybrids, and open attestation libraries that make TEE coordination a reality.

1. TEEs vs. MPC and FHE: Practicality Over Purity

We’ve actually tried MPC and FHE prototypes—the runtimes are just insane.
—Shea Ketsdever

Shea emphasized that TEEs have become an essential solution in MEV architecture because they are fast, private, and allow for architectural decentralization—without requiring central intermediaries. In contrast, she found that techniques like MPC and FHE, while theoretically powerful, lacked the performance needed for MEV use cases. Flashbots tested several prototypes and found runtimes “insane”—simply not viable for millisecond-sensitive auctions. The emergence of high-performance TEEs like Intel’s TDX unlocked real deployment paths, enabling strong privacy and integrity guarantees while staying fast enough for live markets.

Chua positioned TEEs as a “powerful generic tool” in the cryptographic toolbox. He stressed that TEEs offer practical integrity and confidentiality at low cost and overhead, making them ideal for applications where speed and privacy both matter. Rather than being perfect, TEEs are valuable as part of “defense-in-depth,” where multiple security layers reduce systemic risk. He gave the example of combining TEEs with MPC in Worldcoin’s IRIS-matching AMPC protocol: TEEs enforce protocol correctness and deter collusion, even under honest-but-curious adversary models.

2. TEEs in MEV Auctions and Block Building

Stop thinking about block builders as nouns. Start thinking about block building as a verb.
—Shea Ketsdever

Shea described BuilderNet as the product of a long-standing goal at Flashbots: building a decentralized block builder capable of resisting the centralizing pressures of exclusive order flow and cross-domain MEV. She pointed out that early ideas from the Suave whitepaper directly informed BuilderNet’s design. TEEs enabled this shift by allowing performant, privacy-preserving coordination between independent actors. What makes this novel, she explained, is that now you can run atomic arbitrage bots inside enclaves, share order flow securely, and construct more profitable blocks through collaborative logic without centralized trust.

We built an SGX builder in 2023—but without private order flow, it just wasn’t competitive.
—Chua Zheng Leong

Chua recounted Automata Network’s early experiments with SGX-based block building on Ethereum mainnet—before TDX became viable. Their blocks performed well with public transactions, but were not competitive against builders who received private order flow. This real-world limitation reinforced the need for infrastructure like BuilderNet, which can pool encrypted flows in a distributed, verified manner. He also highlighted how TEEs remove the need for social trust in builders and relays—shifting enforcement from off-chain agreements to verified machine behavior.

3. Security: Designing for the Inevitable Breach

Shea framed TEE security as an “arms race,” where robustness doesn’t mean invincibility, but resilience. She acknowledged that TEEs can be vulnerable to side-channel attacks, especially with physical access. To mitigate this, Flashbots runs enclaves in cloud environments, requiring collusion between cloud and hardware providers for a breach. She emphasized designing systems that can respond quickly to new exploits: version gating, rapid patching, and runtime monitoring. Importantly, she noted that current MEV infrastructure carries worse trust assumptions—TEEs, while imperfect, are “an order of magnitude better.”

Chua agreed, stating that systems should be built with failure in mind. He drew parallels to traditional sysadmin practices like patching Linux kernels after exploits, arguing that TEEs should be treated similarly. He emphasized remote attestation as the cornerstone for verifying machine integrity and suggested that enclave revocation could be handled on-chain. If an enclave is compromised, it can be blacklisted and rendered useless after one attempt, minimizing the blast radius. He also advocated using TEEs only for short-lived secrets to limit potential data exposure.

4. TEEs Beyond MEV: New Frontiers in App Design

Shea offered compelling non-MEV use cases, especially in what she called the “account encumbrance” space. She described Teleport, an internal Flashbots app that delegates control of Twitter accounts to TEEs governed by LLM-based behavioral constraints. TEEs here serve not just as secure sandboxes but as programmable identity agents—users can set complex permissions on what their accounts can post. She also mentioned a playful experiment where a TEE helped players secretly collude in an online game so effectively that the creators were banned—yet their tool remained in use by others.

ZK proofs are strong, but expensive. TEEs give us cheap redundancy.
—Chua Zheng Leong

Chua, on the other hand, highlighted TEEs as lightweight proving systems that complement ZK rollups. His team is working with projects like Scroll and Linea to implement multi-proof systems—akin to client diversity in Ethereum, but for rollup proofs. ZK provides strong correctness; TEEs add speed and redundancy. He also touched on physical isolation approaches like “space TEEs”—enclaves on satellites for key recovery or backup signing. These emerging ideas showcase how TEEs can reinforce systems where physical trust separation or computational cheapness are advantageous.

5. Decentralizing the TEE Stack

BuilderNet isn’t here to replace builders—it’s here to give them a better platform.
—Shea Ketsdever

Shea made it clear that BuilderNet is not designed to displace existing builders like Titan or Beaver Build but to offer them a neutral, open platform. The idea is to abstract away the boilerplate of builder infrastructure—negotiating order flow, securing subsidies—and let teams focus on their core competencies. Flashbots is already working with Beaver Build to migrate their stack to BuilderNet. Shea emphasized that TEEs enable a reframing of block building from a single actor’s role to a collaborative process—“a verb, not a noun.”

Chua extended this vision with a decentralized encrypted mempool model. Instead of central TEEs, he described a mesh of mutually attested enclaves, each running RPC endpoints and sharing encrypted order flow. This would allow small searchers to access value flows typically locked behind private builder deals. Chua stressed that mutual attestation and TE-based networking can create a level playing field in a world where MEV profits increasingly depend on exclusive access. Decentralizing the TEE layer, in his view, is not just a technical problem—it’s a redistribution of economic power.

Conclusion

TEEs are no longer an academic curiosity—they’re becoming critical components in the decentralization of both infrastructure and trust. For Shea and Chua, TEEs represent not just secure execution, but new architectural principles: programmable collaboration, enforceable privacy, and fault-tolerant coordination. They both recognize the imperfections, but also agree that today’s trust models in crypto are worse. What’s emerging is not a silver bullet, but a layered evolution—more robust, more open, and surprisingly fast.

“Don’t let perfect be the enemy of good. TEEs aren’t flawless—but they’re already better than what we have.”
—Shea Ketsdever

As rollups, auctions, and identity systems grow in complexity, TEEs could quietly become their hidden engines—less visible than consensus, but no less foundational. And in that transformation, the very concept of trust in crypto may shift—from ideology to instruction set.

Click this link or open https://bit.ly/hfdefi to download the free ebook Head First DeFi, Decoding the DNA of Crypto Transactions & Strategies.

Wisdom of DeFi by EigenPhi

Discussion about this post