Quantum computing has long been framed as a distant, almost theoretical threat to Bitcoin. This week, Coinbase made clear it is treating the problem as an engineering deadline. Chief Cryptographer Yehuda Lindell told MARA Foundation TV that the exchange is designing a post-quantum custody system intended to protect roughly 250 billion USD in institutional assets — and, critically, to support any new signature scheme Bitcoin or another blockchain might eventually adopt.
The design goal, Lindell said, is to keep Coinbase’s custody platform usable regardless of which post-quantum signature schemes blockchains select. Bitcoin has not chosen a signature scheme for practical post-quantum use, and different blockchain communities may reach different decisions rather than converging on one common standard. Coinbase is therefore preparing for several possible outcomes instead of building around a single candidate.
The company wants to avoid a scenario in which a blockchain approves a signature scheme its custody infrastructure simply cannot handle.
## The MPC problem at the heart of institutional custody
Most institutional custody platforms rely on Multi-Party Computation, or MPC, which divides control of a private key among several parties. No participant ever holds or assembles the full private key while approving a transaction. The technique underpins much of the institutional custody industry, including the infrastructure safeguarding assets for clients like BlackRock, which uses Coinbase Custody for digital assets connected to its investment products.
Lindell’s warning is that many post-quantum signature schemes may be “not friendly to MPC.” Hash-based signatures present a particular problem: they lack the arithmetic structure on which traditional cryptographic key splitting depends. Without that structure, the standard approach to distributed key control breaks down.
Researchers — including Stanford cryptographer Dan Boneh, who sits on Coinbase’s advisory board — are studying ways to apply MPC-style controls to hash-based signatures. But Lindell characterized the research as highly experimental, and it is not yet clear whether a practical MPC-equivalent can be created for those schemes at all.
## Hardware security modules as the fallback
To reduce dependence on MPC compatibility, Coinbase is exploring a backup architecture built around programmable Hardware Security Modules. HSMs are physically protected devices that store cryptographic material and perform sensitive operations. Under the design being considered, private keys would remain encrypted with post-quantum cryptography and be assembled only inside secure HSMs, allowing the custodian to work with signature schemes that cannot be divided through conventional MPC.
Physical security becomes more important under this model, since the complete key would briefly exist inside the device. Coinbase would need the modules to perform signing without exposing keys to outside software or operators. Notably, the approach would not abandon MPC for schemes that support it — the HSM architecture would serve as an alternative custody method when a network’s chosen cryptography cannot work with distributed key generation and signing.
Programmable modules would also give Coinbase room to add support as blockchain developers settle on new standards, and Lindell said he expects the finished system to let the custodian declare: “I can support any scheme.” He did not provide a completion date, acknowledging the technical work could take time.
## The numbers at stake
The scale of the exposure is substantial. Lindell cited about 250 billion USD in institutional assets during the program. An August U.S. custody review placed Coinbase’s institutional assets nearer 376 billion USD and said the company safeguards more than 80% of assets held by U.S. spot Bitcoin and Ethereum exchange-traded funds. The differences between the two figures likely reflect reporting dates and the specific services or assets each estimate includes — but either way, Coinbase is the single largest institutional key-holder in the U.S. crypto market, and its post-quantum readiness effectively sets the floor for the industry’s.
## A roadmap already in motion
Coinbase’s post-quantum work predates the latest design. In January the company established an independent quantum computing and blockchain advisory board that includes Boneh, Lindell, Ethereum Foundation researcher Justin Drake, University of Texas professor Scott Aaronson, EigenLayer founder Sreeram Kannan and distributed-systems specialist Dahlia Malkhi. Its roadmap includes changes to Bitcoin address handling, updates to internal key-management systems, and research into supporting schemes such as ML-DSA within MPC infrastructure.
Other custodians are testing parallel routes. In June, BitGo tested quantum-safe MPC with Silence Laboratories using an ML-DSA-based protocol, retaining distributed key control, policy checks and separation of duties. ML-DSA is part of FIPS 204, the post-quantum digital signature standard published by the U.S. National Institute of Standards and Technology.
The context is familiar to Bitcoin watchers: the network still uses elliptic-curve cryptography, no publicly demonstrated quantum computer can derive private keys from exposed public keys, and researchers and industry groups — including the G7 and BIP-360 proponents — have nonetheless called for early preparation because changing Bitcoin’s security model is a slow, contentious process. Coinbase’s message this week is that the custodial side of that transition may be just as hard as the protocol side, and it intends to be ready either way.
Market snapshot at press time (Sep. 23, 22:31 UTC, CoinGecko via fetch-prices): BTC 84,493 USD (-2.3% 24h), ETH 2,687.68 USD (-2.8% 24h), SOL 114.93 USD (-3.2% 24h).
250 billion in institutional assets is finally a real budget line for PQ work. Usually this stuff only gets funded after something breaks.
250 billion in institutional assets hedging against a quantum break before Bitcoin has even picked a signature scheme. Coinbase is basically forced to bet on every horse at once.
250 billion in custody and bitcoin still hasnt picked a signature scheme. lindell basically admitting theyre building for a target that keeps moving
That is the whole point of the design though. Keep custody usable whichever scheme wins instead of betting the vault on one standard.
lindell planning for multiple PQ schemes because BTC might never converge on one is the quietly smart take here. standards fights take years
agreed, and MPC custody makes scheme swaps way harder than a normal wallet. glad someone is stress testing this before its an emergency
hash based sigs being hostile to MPC is the real headline. key splitting just breaks without arithmetic structure, the HSM fallback is basically plan B taped on
calling the HSM fallback plan B is generous, its plan D wearing a trench coat lol
^ and boneh on their advisory board. at least the right people are in the room
Meanwhile the actual deadline question nobody answers: what happens to coins sitting in old p2pk addresses if a CRQC shows up before any migration? Custodians will be fine, plebs wont be.
p2pk coins are like what, 20k btc total, satoshi era stuff. real question is whether custodians set the deadline or keep waiting on core to pick a scheme