Cardano has been running for nearly nine years, and it's still one of the most argued-about assets in crypto. Depending on who you ask, it's either the most rigorously engineered layer-1 in existence or a textbook case of research getting in the way of shipping. Both descriptions are defensible — that's why this series exists.
This is Part 1 of a four-part refresher on where Cardano stands in mid-2026. Before governance, the Midnight privacy chain, or the Hydra/Mithril scaling stack, you need the fundamentals: what Ouroboros is, how staking works under the hood, how the ledger differs from Ethereum's, and where the money comes from. No hype, no dismissal, just mechanics and honest trade-offs.
What Cardano Actually Is in Mid-2026
Cardano was founded in 2015 by Charles Hoskinson and Jeremy Wood, both formerly of the Ethereum project. Development is split across three organizations: Input Output Global (IOG) handles research and engineering, the Cardano Foundation handles stewardship, and Emurgo handles commercial partnerships. That three-way structure is unusual — most chains have one core dev team — and it partly explains why some things move slowly.
Mainnet launched in September 2017 with the Byron era: a federated network running an early version of Ouroboros, Cardano's proof-of-stake consensus protocol. "Federated" meant a small, permissioned set of nodes run by the founding organizations produced blocks — not the decentralized stake-pool network Cardano runs today.
As of mid-2026, ADA's circulating supply sits around 37.27 billion, roughly 83% of its 45 billion hard cap, putting the asset around #18 by market cap. That's a solidly top-20 coin with nearly nine years of continuous uptime behind it.
Here's the honest number to sit with, though: Cardano's DeFi total value locked (TVL, the amount of capital parked in its lending, trading, and staking-derivative protocols) is around $132 million, putting it roughly 27th among blockchains by that measure. A top-20 market cap next to 27th-place TVL is a real gap, and it's the single most common criticism leveled at the chain. We'll come back to it.
Ouroboros: A Peer-Reviewed Proof-of-Stake Protocol
Ouroboros is Cardano's consensus protocol, and its defining trait is that it was published as peer-reviewed academic research before it was deployed, rather than written up after the fact. The original paper — by Kiayias, Russell, David, and Oliynykov — was presented at CRYPTO 2017, one of cryptography's top academic conferences, and has since been cited more than 1,200 times. Most consensus designs exist as whitepapers or documentation, not peer-reviewed papers; Ouroboros went through that process first.
There isn't one Ouroboros. There are several variants, rolled out in stages as the network matured:
- Classic — the original version that bootstrapped Byron in 2017.
- BFT — a short-lived transitional variant used during the move toward decentralization.
- Praos — the version that has produced every Cardano block since the 2020 Shelley era, and still does today. Praos selects the slot leader (the node allowed to produce the next block) privately and unpredictably, weighted by delegated stake. Nobody outside the protocol knows who the next leader is in advance, which makes it hard to target upcoming block producers.
- Genesis — a variant that lets a new node join and verify the chain's history safely from nothing but the genesis block, closing off long-range attacks where an adversary forges an alternate history from far in the past.
Be precise here: as of mid-2026, mainnet still runs Praos. Genesis has not fully shipped — what's live is an interim feature called "bootstrap peers," which helps new nodes find trustworthy peers on first sync but isn't the full Genesis security model. A claim that Cardano "runs on Ouroboros Genesis" is ahead of where the software actually is.
There's also Ouroboros Leios, a redesign of block production and propagation aimed at throughput, which entered public testnet in June 2026. Leios is forward-looking, not live on mainnet, and belongs to the scaling conversation Part 4 picks up properly.
Epochs, Slots, and Slot Leaders
Cardano's clock runs on epochs and slots. A slot is a one-second window in which a block might be produced. An epoch is 432,000 slots, or five days — the unit stake snapshots, reward payouts, and pool statistics all update against.
Not every slot produces a block. The "active slot coefficient" is set to 5%, meaning on average about one in twenty slots has an eligible leader that actually produces a block. Over a five-day epoch that works out to roughly 21,600 blocks. Leaders are chosen through a private, stake-weighted lottery running inside Praos — there's no mining hardware race and no proof-of-work puzzle, just a probabilistic draw where your odds scale with how much ADA is delegated to your pool.
Delegation: Staking Without Surrendering Custody
This is the part of Cardano that holds up best against scrutiny.
When you delegate ADA to a stake pool, your coins never leave your wallet. Delegation publishes a certificate on-chain pointing your stake at a chosen pool; the ADA itself stays exactly where it was, under your keys, spendable at any time. There's no lock-up and no unbonding queue — delegated ADA can be moved or spent immediately, unlike protocols that hold your stake hostage for days or weeks after you ask to withdraw it.
More importantly, there's no slashing. On Ethereum, a validator that goes offline at the wrong moment or gets caught equivocating can lose a portion of its staked principal. On Cardano, the worst case for a delegator is a missed reward for an epoch — never a loss of the ADA itself. That's a structural difference, not a marketing line: the protocol has no mechanism to confiscate delegated stake in the first place.
The trade-off is reward speed, not risk. After you delegate, the first reward doesn't arrive until roughly 15 to 20 days later — a minimum of three epochs — because the reward calculation lags behind the stake snapshot it's based on. After that first payment, rewards arrive automatically every epoch, every five days, with no further action needed.
As of mid-2026, about 56.8% of circulating ADA — roughly 20.7 billion ADA — is actively delegated across approximately 2,698 stake pools, earning a reward rate around 2.18% APY. Pools also have a saturation point: push too much stake into one pool and the per-ADA reward it produces starts to decline, a deliberate design choice meant to discourage stake from concentrating into a handful of mega-pools.
eUTXO vs the Account Model
Cardano uses the "extended UTXO" model, or eUTXO — arguably its least understood feature. It starts from the same foundation as Bitcoin: a ledger made of discrete, spendable coins — UTXOs, or unspent transaction outputs — rather than account balances. When you spend, you consume specific UTXOs and create new ones; there's no single running balance updated in place, the way there is on account-based chains.
Where "extended" comes in: Cardano's UTXOs can carry a script, a datum (attached data), and a redeemer (data supplied at spend time), which is what makes smart contracts possible on top of a UTXO ledger. That's the Plutus smart contract layer.
The eUTXO model has one property that's genuinely valuable and rarely explained clearly: whether a transaction succeeds, and what it costs, depends only on that transaction and the specific UTXOs it references — never on some other piece of mutable global state that could change between signing and submitting. Practically, you can compute your fee and know a transaction will succeed before you send it. No gas-price auctions, no transaction that executes halfway, fails, and still charges you for the attempt.
The trade-off is concurrency. If a shared UTXO represents something like a decentralized exchange's liquidity pool, only one transaction can consume it at a time — a second transaction touching that same pool in the same block has to wait, because a UTXO is either spent or it isn't. Account-based chains don't have this bottleneck, since balances update independently. In production, Cardano DEXs work around it with off-chain batchers — services like Minswap's that collect a batch of orders and settle them in one transaction — and with transaction chaining, which lets a wallet spend a UTXO it just created without waiting for confirmation first. These are real mitigations, not a full fix. Concurrency is a genuine architectural cost, not a rumor.
ADA Tokenomics: Counting Down to 45 Billion
ADA has a hard cap of 45 billion coins, with about 37.27 billion — roughly 83% — already circulating.
New ADA doesn't appear from nothing. It's drawn down from a fixed reserve pool through monetary expansion: each epoch, the protocol withdraws an amount equal to rho (0.3%) of whatever remains in the reserve. Because that 0.3% is calculated against a shrinking pool, issuance decays geometrically — larger withdrawals early on, smaller ones as the reserve depletes. There's no tail inflation planned. Once the reserve is exhausted, no new ADA gets created, ever.
Of every epoch's monetary expansion, plus 20% of all transaction fees collected, a share (tau, also 20% of the expansion amount) is routed to Cardano's on-chain treasury instead of stake pool rewards. That treasury is the pot that funds ecosystem development through on-chain voting — the exact mechanism Part 2 of this series covers when it gets into the Chang and Plomin governance upgrades.
The Academic Method — and Its Discontents
IOG's research output is substantial: its public research library lists more than 250 peer-reviewed papers, and the original Ouroboros paper alone has been cited over 1,200 times. That's the foundation of Cardano's pitch — work out the theory, prove the security properties, then ship.
The explicit trade-off is speed. Peer review, formal specification, and academic rigor take longer than "ship it and patch bugs live," and Cardano's own history shows the cost plainly. Mainnet launched in September 2017 with no smart contract capability at all. It wasn't until the Alonzo hard fork on September 12, 2021 — four years later — that Plutus smart contracts went live and developers could build anything beyond simple token transfers.
Four years with a meaningful market cap and no usable applications is a long time in an industry that moves as fast as crypto does, and it's the direct origin of the "ghost chain" label critics still use today. That criticism didn't evaporate once Alonzo shipped, either — the broader complaint is that Cardano's dApp ecosystem and developer activity still haven't kept pace with Ethereum's or Solana's, and the $132 million TVL figure above is the evidence people point to now.
The counterargument isn't nothing. Cardano hasn't had a consensus-level security failure or an exploit that drained the base protocol. Fees are predictable in a way few chains can match. And the no-slashing delegation model means stake pool operators and delegators alike have never lost principal to a protocol-level penalty. Whether four-plus years of caution was worth that track record is a genuinely open argument, and this refresher isn't going to settle it for you. Both the "ghost chain" critique and the "slow but solid" defense are fair readings of the same history.
Where This Series Goes
This piece covered the operating system: Ouroboros consensus, the eUTXO ledger, and the tokenomics that fund everything else on top of it. The next three parts build directly on this foundation.
Part 2 covers Cardano's on-chain governance — the Chang and Protocol 11 upgrades that hand treasury and protocol-parameter decisions to ADA holders. Part 3 covers Midnight, Cardano's confidential-computing sidechain and its NIGHT token. Part 4 covers the Hydra and Mithril scaling stack, where Leios and Cardano's throughput ambitions actually get tested.
None of it makes much sense without the basics covered here: Praos secures every transaction, eUTXO is what every smart contract and DEX has to work with, and the reserve-and-treasury split funds the governance system Part 2 hands to token holders.
