There is a wonderfully awkward question at the center of Input Output Group: Can the company that became indispensable to Cardano prove its success by becoming less indispensable? For a decade, the Singapore-headquartered, globally distributed research and engineering operation has been the brain trust most closely identified with the blockchain. It designed Cardano from first principles, developed the Ouroboros proof-of-stake protocol and supplied much of the engineering talent that kept the network moving. Now it is deliberately pushing pieces of that work toward independent teams while widening its own orbit.
That makes Input Output less like a conventional software vendor and more like a three-part contraption: university laboratory, open-source infrastructure company and venture studio. Its inputs are cryptography, formal methods and game theory. Its outputs include networks, wallets, APIs, identity frameworks and experiments with names such as Midnight, Lace, Blockfrost and Quantum Hosky. The name itself has changed several times - IOHK became Input Output Global and then Input Output Group, often shortened to IO - but the method has been remarkably consistent.
The paper before the product
Most startups try to reduce the distance between an idea and a release. Input Output inserts an academic department in the middle. From 2015 through 2020, the company says it produced more than 250 peer-reviewed papers, including the Ouroboros series. By late 2024, IO Research said its work had passed 10,000 citations. This is not decorative scholarship. Formal specifications and mathematical analysis are meant to expose security or incentive failures while those failures are still equations, before they become expensive incidents on a live network.
How an idea travels through IO
The practical payoff is high-assurance software: systems built for transactions, identity and governance where a small bug can become a large financial event. The cost is pace and complexity. Cardano's extended unspent transaction output model is powerful, but even Input Output's 2026 developer proposal acknowledged that its tooling has not always offered newcomers an easy lift. The company's differentiation and its recurring product challenge are therefore the same thing. Rigor attracts serious builders; rigor must also be translated into documentation, APIs and interfaces that ordinary builders can tolerate.
“Technology should serve people, not the other way around.”Input Output's stated design belief
A portfolio, not a monolith
Cardano remains the foundation and the reference implementation of the approach. It gives developers a proof-of-stake base layer for assets, smart contracts, governance and decentralized applications. Stake-pool operators maintain the network; application teams build finance, identity and consumer services on it; ADA holders participate as users, delegators and voters. The company said in April that more than 3,000 stake pools depend on the core security tooling it helps maintain.
The portfolio answers a commercial problem. Protocol research alone does not guarantee recurring revenue, and Input Output is privately held, with no public breakdown of its finances. Its emerging model mixes engineering contracts, milestone-based work funded by the Cardano treasury, commercial partnerships, venture investments and revenue from studio companies or infrastructure products. The supplied company record estimates annual revenue at $34.3 million, but the company has not publicly confirmed that figure.
Its customers are equally mixed. Developers use Blockfrost and Cardano tooling. Consumers hold assets and enter applications through Lace. Institutions explore data privacy on Midnight. Governments and enterprises have tested identity and loyalty systems. In Ethiopia, an education-credential pilot registered 38,000 pupils by June 2023, according to Input Output, though the proposed national rollout did not materialize. The underlying work evolved into Identus. DISH and Boost tested blockchain-backed loyalty infrastructure. Brave integrated Cardano into its wallet. These projects reveal both the appeal and the friction of the market: decentralized infrastructure can improve verification and user control, but deployment still depends on regulation, partners, integration and patient execution.
The moat is a method
Input Output competes with organizations clustered around Ethereum, Polkadot, Solana, Polygon and Avalanche, plus a separate rival set for each portfolio product. Lace faces wallets such as MetaMask. Blockfrost meets API platforms including Alchemy and QuickNode. Midnight enters a field of privacy and zero-knowledge networks. Pogun joins a crowded effort to make Bitcoin useful beyond holding and payments.
What separates IO is not that it owns every layer. It is the repeatable bridge between academia and deployment. Chief scientist Aggelos Kiayias also holds a chair in cybersecurity and privacy at the University of Edinburgh. The company supports research hubs and laboratories at Edinburgh, Stanford and, as of April 2026, the Archimedes Research Unit in Athens. The Athens lab is pointed at the overlap of blockchain, privacy-preserving artificial intelligence and healthcare - work where using sensitive data without exposing it is more than a clever technical demonstration.
A company learning to let go
The clearest demonstration of Input Output's current strategy arrived when it stopped two projects. Acropolis was a Rust-based alternative Cardano node; tiered pricing was intended to manage network demand. Newer research and shifting priorities made both less useful. In April 2026, the company said it would return ₳4.1 million of allocated funds rather than continue. Acropolis code remained open source, and its data node had already reduced a sync process described as taking four days to about one hour.
That decision matters because Cardano's treasury has become a significant customer. For 2026, IO proposed nine workstreams totaling $46.8 million, down from $97.5 million the previous year. The work covers developer experience, scaling and other core infrastructure, with milestones subject to community governance and oversight through Intersect. Input Output says the goal is to ask for less each year as delivery capacity spreads to a wider collection of organizations.
The organizational version is already visible. Functions and product teams associated with Cardano are moving toward specialist entities. In August 2026, four engineering functions - platform engineering, quality assurance, cryptography, and performance and tracing - began transitioning from IO Labs to ICAN Group with their existing leadership and teams. Other components, including wallet and scaling work, have followed or are exploring similar paths. This is decentralization applied to the org chart, not merely the ledger.
There is an obvious business tension. Cardano still supplies reputation, demand and funding, while independence reduces IO's direct control. The bet is that a healthier ecosystem, staffed by multiple capable engineering organizations, is more valuable than permanent dependence. Input Output can then concentrate on venture creation, frontier research and products that travel across networks. Midnight's privacy model, Lace's multichain direction and Pogun's Bitcoin focus all test whether the company can carry its method into markets where the Cardano name is an ingredient rather than the entire meal.
The most interesting output may be an ecosystem that no longer needs one input.The decentralization test facing the company
Where it fits
Input Output occupies a narrow shelf in the technology market. It is too research-heavy to resemble a typical crypto app company, too product-minded to be an academic institute and too committed to open protocols to behave like an enterprise software licensor. That can make the company hard to summarize. It can also make it useful. A government evaluating credentials, a developer needing blockchain data, an institution exploring private computation and a community funding a protocol upgrade all encounter different faces of the same organization.
Its future will be judged less by how many papers it publishes than by how cleanly those papers become tools, and less by its dominance inside Cardano than by whether Cardano develops strong alternatives. The company has already shown it can design a network. The more revealing experiment is now under way: building products, businesses and independent teams around that network without losing the rigor that made it distinctive. That is harder to diagram than a consensus protocol. It may also be the more consequential proof.