What Is a Blockchain, Really?
A blockchain is a specific type of distributed ledger — a database replicated across many nodes with no central administrator. The “chain” part refers to how new data bundles (blocks) are linked to previous ones via cryptographic hashes, creating an append-only history.
But here’s the important context: the term “blockchain” is massively overused. When you hear “we’re putting X on the blockchain,” the first question should be: why does X need to be on a distributed, append-only ledger rather than a regular database?
The blockchain test: Before believing any “blockchain solution” pitch, ask: Does this require multiple mutually distrusting parties to share a ledger? If yes, blockchain might be relevant. If it’s one company managing their own data — it’s just a database with extra steps and worse performance.
The Core Components
1. Data Structure: Blocks and Hashes
Each block contains:
- A block header: timestamp, reference to previous block’s hash, a nonce (for PoW chains), and the Merkle root
- A set of transactions
The Merkle tree is a clever data structure: every transaction is hashed, hashes are paired and hashed again, all the way up to a single root hash. This means you can prove a specific transaction is in a block without downloading the entire block — critical for mobile clients and SPV proofs.
The link to the previous block is a hash of that block’s header. Change any transaction in any historical block, and its hash changes, breaking every subsequent link. This is the tamper-evidence property.
2. Consensus Mechanisms: How Nodes Agree
With thousands of nodes holding copies of the same ledger, how do they agree on which transactions are valid and what the canonical chain is? This is the consensus problem.
Proof of Work (PoW)
Used by Bitcoin. Nodes expend real energy to earn block-production rights. Makes attacks economically costly. Criticism: energy-intensive, slow (Bitcoin: ~7 transactions/second, 10-minute blocks), limited programmability.
Proof of Stake (PoS)
Used by Ethereum (since “The Merge” in 2022) and most newer chains. Validators lock up (stake) cryptocurrency as collateral. The protocol randomly selects validators to propose blocks, weighted by stake. Dishonest validators lose their stake (slashing).
Key difference: PoS replaces energy cost with economic stake. Much more energy-efficient (Ethereum’s energy use dropped ~99.95% after The Merge). Criticism: arguably concentrates power among large token holders, “nothing at stake” theoretical concern (largely addressed by slashing).
Practical Byzantine Fault Tolerance (PBFT) and variants
Used by permissioned enterprise blockchains (Hyperledger Fabric, some consortium chains). Faster, deterministic finality, no mining. Tradeoff: requires a known, limited set of validators — not permissionless.
For FIs: When a bank or consortium runs their own blockchain (like JPMorgan’s Onyx/JPM Coin), they’re usually using a permissioned chain with PBFT-style consensus. The trust model is completely different from Bitcoin — they know and vet every validator. The value proposition is interoperability and programmability, not decentralization.
3. Finality
In traditional finance, when a payment settles, it’s final. In blockchains, finality is more nuanced:
- Probabilistic finality (Bitcoin): Transactions are never 100% final, but the probability of reversal drops exponentially with each new block. After 6 confirmations (~1 hour), a Bitcoin transaction is practically irreversible.
- Deterministic/economic finality (Ethereum PoS): After certain checkpoints, validators would lose their entire stake to revert — making reversal economically impossible.
- Instant finality (many permissioned chains): The consensus protocol guarantees finality in one round.
For financial settlement use cases, finality type matters enormously. T+2 settlement has finality issues too — fails and recalls exist. Some argue blockchain-based settlement could actually improve finality semantics.
Public vs. Permissioned Blockchains
| Public (Bitcoin, Ethereum) | Permissioned (Hyperledger, R3 Corda) | |
|---|---|---|
| Who can participate | Anyone | Known, vetted validators |
| Consensus | PoW / PoS | PBFT variants |
| Transaction speed | 7–30 TPS | Thousands of TPS |
| Privacy | Pseudonymous (public ledger) | Configurable privacy |
| Finality | Probabilistic | Deterministic |
| Decentralization | High | Low-medium |
| Use case | Censorship resistance, open access | Consortium workflows, settlement |
Most serious enterprise blockchain projects use permissioned chains precisely because they don’t need permissionless access — they need shared infrastructure among known counterparties.
Smart Contracts: Code That Runs on the Chain
A smart contract is code deployed to a blockchain that executes automatically when predefined conditions are met. The code is immutable once deployed and runs identically on every node.
Ethereum pioneered this with its Ethereum Virtual Machine (EVM) — a Turing-complete runtime that lets developers write arbitrary logic in a language called Solidity.
Simple example: an escrow contract that holds funds and releases them to the seller only when a delivery oracle confirms the package arrived. No escrow agent needed.
More complex: an entire lending protocol where interest rates adjust algorithmically based on supply and demand, loans are auto-liquidated when collateral falls below a threshold, and no human intervention is required at any step. This is what DeFi protocols do.
The “code is law” problem: Smart contracts are immutable and execute exactly as written. When there’s a bug, there’s no recourse — you can’t call the bank. In 2016, the “DAO hack” drained $60M from a smart contract through a code vulnerability. The Ethereum community controversially hard-forked the chain to reverse it. This revealed a deep tension: true immutability vs. human error correction.
Scalability: The Blockchain Trilemma
Vitalik Buterin popularized the blockchain trilemma: it’s very hard to achieve all three simultaneously:
- Decentralization (many independent validators)
- Security (resistant to attacks)
- Scalability (high throughput, low fees)
Bitcoin chose security + decentralization. Ethereum has been progressively trading some decentralization for scalability through Layer 2 solutions.
Layer 2 Solutions
Layer 2 protocols sit on top of a “Layer 1” chain (like Ethereum) and process transactions off-chain, periodically settling compressed proofs or batches back to L1.
- Rollups (Optimistic and ZK): Bundle thousands of transactions into a single L1 transaction. Ethereum L2s like Arbitrum, Optimism, and Base now process far more transactions than Ethereum mainnet.
- State channels: Two parties transact off-chain and only settle net results on-chain (the Lightning Network for Bitcoin works this way).
- Sidechains: Separate chains with bridges to L1 (Polygon).
This is how blockchain systems are scaling toward the throughputs needed for financial applications.
Where Blockchain Adds Genuine Value in Finance
After years of experimentation, the clearest use cases are:
- Cross-border settlement: Removing correspondent banking friction. JPMorgan’s Onyx processes billions in institutional repo settlement. Stellar and Ripple are used for international remittances.
- Asset tokenization: Representing real-world assets (bonds, real estate, private equity) as tokens on a chain. BlackRock tokenized a money market fund on Ethereum in 2024.
- Trade finance: Letters of credit, supply chain financing — currently paper-heavy, multi-party, slow. Blockchain provides a shared audit trail.
- Digital identity: Self-sovereign identity and credential verification.
Key Takeaways
- A blockchain is a specific tool — a distributed, append-only ledger. Not every problem needs one.
- The core innovation is achieving consensus (agreement) without a central authority, using cryptography + economic incentives
- Consensus mechanisms (PoW vs PoS vs PBFT) represent different tradeoffs between energy, decentralization, speed, and finality
- Smart contracts enable programmable, self-executing financial logic — the foundation of DeFi
- Layer 2 solutions are solving the scalability problem; throughput is no longer a hard barrier
- For financial institutions, permissioned chains and tokenization are the highest-probability near-term use cases