Can the most important crypto projects effectively fortify themselves ahead of the arrival of Q-Day or major AI breakthroughs in math?
That’s the grand question, and one that’s being debated more than ever in the wake of Justin Drake’s recent “bunker mode” comments.
Only time will tell if those remarks underestimated or overestimated the timeline regarding math breakthroughs. But, in any case, both the quantum and AI threats are very real. They’re not distant, abstract dangers, and some crypto projects are taking these specters more seriously than others.
Among these projects is Starknet, whose builders made waves last week when they announced they were considering transitioning the chain from an Ethereum L2 to a standalone L1 network.
Why? To harden against quantum and AI more quickly than the chain can while being directly tethered to Ethereum.
As some of you may recall, StarkWare’s post-quantum roadmap from June 30th didn’t float an L1, and StarkWare CEO Eli Ben-Sasson has since stressed that this idea has evolved recently from the growing concern that traditional cryptography might get broken much sooner than the crypto industry expects.
— (@)
The logic is that as an L1, Starknet could set the pace of its security upgrades rather than follow Ethereum’s, putting it on track for what it bills as “the first fully quantum-resistant network,” targeted for 2027. So why can’t Starknet get there as an L2?
The June roadmap shows why. It splits Starknet’s quantum migration into three phases. The first two (e.g. swapping Pedersen hashing for the quantum-resistant BLAKE2, adding post-quantum signatures like Falcon-512, and migrating legacy contracts) are Starknet’s to execute, while the third phase isn’t, as it covers two pieces Starknet effectively borrows from Ethereum. Namely:
-
Bridge messaging, which uses quantum-vulnerable elliptic-curve calls to move messages between the two chains. StarkWare is explicit that removing those calls before Ethereum migrates would break the bridge.
Enjoying this article?
Subscribe to Bankless or sign in
-
Blob data, i.e. the state data Starknet publishes to Ethereum via blobs, anchored by elliptic-curve-based KZG commitments. A broken KZG would hurt data availability per StarkWare, not the soundness of Starknet’s state, which stays guarded by STARK proofs. Still, the June roadmap leaves this fix entirely to Ethereum.
That said, it might be a few years before Ethereum finishes its own post-quantum evolutions, which would put key pieces of Starknet’s defensive roadmap on ice for at least that long.
This is a point that other projects will undoubtedly contend with, too. Since a rollup inherits the L1’s cryptography alongside its security, its quantum timeline is effectively whichever is slower, its own or the L1’s.
Thus, simply put, Starknet is anticipating coming up against this wall because it can do its side of the work much faster. What’s interesting too is that some of Ethereum’s own quantum plans will make it look more like Starknet.
For instance, Ethereum is pursuing EIP-8141, which among other things would let accounts adopt quantum-safe signatures on their own timelines rather than wait for a network-wide migration. Starknet, meanwhile, has had native account abstraction (i.e. every account is a programmable smart contract) for years. That’s why S2morrow could already build a post-quantum Starknet account around Falcon-512 signatures with no protocol changes.
Moreover, Ethereum plans to swap its BLS validator signatures for leanXMSS, a hash-based scheme, and to compress the bulkier results with zk proofs via leanVM. It also lists STARK-based commitments among the leading candidates to replace KZG. STARKs are hash-based by design and put the “Stark” in Starknet.
In kind, over the past couple of years Starknet has assembled much of the machinery it would need to become an L1, like STRK staking, multiple sequencers, etc. These pieces were originally made to make Starknet a better decentralized extension of Ethereum, but they do shrink the chain’s L1 question considerably.
From what I can tell, the missing parts for now are data availability (i.e. where transaction data gets published so anyone can rebuild the chain), final settlement, and reworking the official bridge as needed.
Speaking of bridging, consider how a StarkGate withdrawal works today: funds only release after Ethereum has verified the proof for the withdrawal’s block and the matching state update has landed there. Put otherwise, Ethereum checks Starknet’s math, so Starknet’s operators can’t forge the state that bridged funds depend on.
An L1 Starknet could land anywhere on that spectrum, though. It might keep posting STARK proofs to Ethereum to back a trust-minimized bridge, or it might cut the cord entirely. Those are very different trust models, so this is the open question I’d watch closely.
Remember, too, that Starknet has long pitched itself as a layer that could one day settle on both Bitcoin and Ethereum. That’s not the same thing as leaving Ethereum, but a proof-connected Starknet L1 bridging to both would arguably rhyme with that vision more than break from it.
For now, nothing official has kicked off but it seems clear that an L1 architecture will be the path forward for Starknet. Whatever happens, the chain wants to move sooner rather than later, because there might not be a later for the projects that don’t make adjustments in time. Fair enough, I say.
