Cardano founder Charles Hoskinson used his 1 May 2025 livestream to unveil the first quantitative performance figures for Ouroboros Leios, the protocol upgrade that—if it performs in production as the prototype already does in simulation—promises what he called “an infinitely scalable protocol, a one-minus-delta protocol.”
Speaking from Colorado, Hoskinson described Leios as the most ambitious iteration in the Ouroboros research line: a telescoping design that “extends Prowse” yet can “collapse back down to the current protocol” should any unforeseen fault occur. Unlike stand-alone Byzantine-Fault-Tolerant (BFT) engines, Leios is engineered to stack on Cardano’s existing proof-of-stake ledger without abandoning the original 50% Byzantine resistance or the network’s 24/7 liveness guarantees.
Hoskinson shared a single slide from the prototyping team that captured the community’s attention. The metric in focus was the input-block (IB) rate, a new adjustable parameter that multiplies throughput by inserting many “input blocks” in parallel before each ranking block is elected.
“With just that one input block, you can see that maximum is about 6 TPS,” he said while pointing to the graphic. “You have five input blocks, 10 input blocks, 20, 30—11,000 TPS for your minimum [250 Bytes], maximum 180 [TPS] at 16,000-kilobyte transactions, which are huge.”
Because IB count is a parameter, network governors will be able to raise or lower it each “tick” period, then dedicate a subsequent “tock” period to software and network optimizations before turning the dial again. In Hoskinson’s words, “year-by-year Cardano gets naturally faster and it doesn’t require a brand-new design or protocol for it.”
Leios is the most invasive change to the Haskell node since the 2020 shift from Ouroboros-BFT to Shelley consensus. The upgrade touches virtually every subsystem: ledger logic, block structure, network stack, consensus rules and cryptography. A “follow-the-sun” development schedule—multiple teams handing code off across time zones—will be adopted to compress time-to-market despite higher cost and developer fatigue.
Hoskinson cautioned that the rewrite is non-trivial: “It typically takes about nine months to twelve months of implementation effort from a SIP to bring this into the protocol.” The formal specification, simulations and Cardano Improvement Proposal (SIP) are expected to be finished “the second half of this year.” Once those artefacts exist, Input Output Global (IOG) will issue a request for proposals to outside firms to co-implement the code in Haskell—and potentially in Rust and Go if alternative clients mature in time.
Leios arrives alongside a separate roadmap of Layer-2 technologies—Hydra, Mithril, Midgard optimistic roll-ups and the recursive-SNARK initiative—each designed to compound throughput without compromising decentralisation. Hoskinson emphasised that the UTXO model makes Cardano uniquely capable of embedding zero-knowledge proofs: “Every output can be a proof as opposed to a transaction.”
The strategy is to let Leios supply base-layer headroom while Layer-2s add specialised scalability paths, together positioning the network to absorb “the enormous transaction volume that Bitcoin DeFi, XRP DeFi, and us becoming an AVS system” could generate.
Hoskinson closed by reiterating Cardano’s research-driven ethos: “Nobody in the cryptocurrency industry writes a SIP, has dense simulations and prototyping, and a formal specification as the definition of an RFP. Nobody does that.” Leios, he said, is the “crowning achievement of ten-plus years of careful thought, research and engineering… the capstone of the Ouroboros agenda.”
If the delivery timetable holds, Cardano stakeholders could be voting on the Leios SIP in early 2026, bringing live throughput of at least 11,000 transactions per second—without abandoning the platform’s signature security model—before the end of next year.
At press time, ADA traded at $0.71.