This is a segment from the 0xResearch newsletter. To read full editions, subscribe.
Bitcoin is a dumb base layer. In its native scripting environment, computation is stateless, meaning each transaction is validated independently — with no built-in memory of prior events or intermediate results. This limits Bitcoin to simple, one-off logic like multisig, timelocks or a basic inheritance contract.
True Bitcoin layer-2 (L2) networks will need stateful computation on Bitcoin, which is where StarkWare’s newly proposed ColliderVM comes in.
The idea is to let Bitcoin validate complex computation across transactions, something which not long ago was thought to require a new soft fork upgrade. While still in the early phases of development, ColliderVM joins a growing class of trust-minimized L2 bridge architectures that seek to bypass the current stalemate around new Bitcoin opcodes like CTV or CAT. No forks required.
ColliderVM builds on ideas from BitVM2 and StarkWare’s earlier ColliderScript, using hash-collision-based puzzles to pass data across Bitcoin transactions. This makes it “at least x10,000 more efficient” than ColliderScript, according to StarkWare co-founder Eli Ben-Sasson.
But as BitVM creator Robin Linus noted to Blockworks, that may not be as impressive as it sounds, since “ColliderScript is impractically slow.”
ColliderVM avoids the fraud proofs central to BitVM2. That means operators aren’t forced to prepay withdrawals while waiting for fraud windows to expire. Instead, computation is validity-based and verified directly on Bitcoin.
“Think of this as step two on the journey to deliver [zk validity proofs],” Ben-Sasson told Blockworks.
The tradeoff is cost. As Bitcoin PIPES creator Misha Komarov puts it: “It requires about 30 hours of the whole Bitcoin network hash rate compute to spend one covenant. [That] results in a cost of millions of dollars [he estimates $40m] per tx.”
Ben-Sasson acknowledges that ColliderVM is still expensive, and not ready for prime time. “The newly released research is important because it shows that it works — not that it’s financially viable,” he said.
However, ColliderVM’s capital efficiency and simplicity, with no need for onlookers or interactive challenge protocols, may appeal to developers frustrated by BitVM2’s complexity.
Linus himself is looking forward to further Bitcoin upgrades. “CTV is great for BitVM,” he said, especially when paired with CSFS to “eliminate the existential honesty assumption” and simplify bridge logic. ColliderVM, on the other hand, circumvents the need for CTV entirely — but at the cost of real-world feasibility, for now. That could change with the advent of specialized hardware.
Meanwhile, PIPES represents an alternative approach — one preferred by Botanix Labs’ co-founder Willem Schroé. In his view, BitVM is “too complex,” and ColliderVM is “brilliant, but energy-intensive.” PIPES is “very simple, very clean, but very theoretical and might not work.”
Komarov agrees. “PIPES, in comparison to this hash collisions trick, are much more experimental from the theoretical point of view,” he told Blockworks. “But once they work, they’re much cheaper.”
Schroé has strong views on the bridge verification space. Regardless of what approach becomes production-ready first, he thinks they do something crucial: They show users want covenant-like behavior.
“Bitcoin Core is not even thinking about covenants,” Schroé said. “But PIPES and Collider allow you to show that demand.”
