There’s an old scare story: if Bitcoin becomes too expensive, transactions will suffocate. If Ethereum shoots into the tens of thousands, the network will collapse under fees. It sounds logical until you look at the mechanics. And the mechanics are boring, engineering-heavy—and therefore reassuringly calm. Coin price and transaction price are only loosely connected. People confuse them mostly because they’ve never looked inside a block.
Let’s start with Bitcoin. Blocks appear not because “there are transactions,” but because there is time and consensus. Roughly every ten minutes the network must agree on a new ledger state. Even if there are zero transactions, a block will still be found. It will inсlude the coinbase transaction—the miner’s reward. That’s issuance, neatly embedded into the process of confirming history. Satoshi designed a scheme where security, time, and coin issuance are fused into a single ritual. So empty blocks aren’t a bug or a tragedy—they’re the systеm breathing normally.
When there are many transactions, an auction begins. Not an auction of “who’s richer,” but an auction of “who’s more urgent.” Each transaction sets a fee rate—sats/vB, satoshis per virtual byte. A block is limited by weight—about 4 million weight units—and the miner selects the set of transactions that maximizes total fees. That’s it. No magic. If you bid 5 sats/vB while the market is at 50, you’ll wait. If the network is empty, 1 sat/vB can be enough. Bitcoin’s dollar price is secondary here—it simply multiplies the outcome.
This leads to a natural question: why move bitcoins at all if they’re digital gold? The answer is plain. They’re moved rarely and in large chunks: for cold storage, custodian rebalancing, and opening or closing second-layer channels. Bitcoin hasn’t been needed for coffee for a long time. But for “holding and proving ownership,” it’s ideal.
To ensure that this “holding” doesn’t get in the way of those who do need to move funds, SegWit arrived in 2017. It was proposed by Bitcoin Core developers and adopted through a painful but revealing consensus process. SegWit moved signatures out of the main transaction structure into a separate area—the witness. That solved transaction malleability—a subtle but fundamental engineering flaw in early Bitcoin that made it unsafe to build complex constructions on top of L1.
The practical effect was twofold. First, transactions became non-malleable and therefore suitable for payment channels and L2 systems. Second, signatures stopped consuming space in the main body of the block. Formally the block remained 1 MB; in practice, it could fit more activity. Today, roughly 80%+ of Bitcoin’s on-chain transactions use SegWit addresses. The rest are legacy formats. They aren’t “banned” for one simple reason: a huge portion of UTXO is historically immobilized—some keys are lost, some sit in old storage setups with no incentive to migrate. That’s not stagnation; it’s archaeology.
Now let’s calculate instead of believing. Suppose Bitcoin is $1,000,000. A typical SegWit transaction is about 140 vB. With moderate network load and a 10 sats/vB fee rate, the fee is: 140 × 10 = 1,400 satoshis = 0.000014 BTC ≈ $14. Even at 20 sats/vB, it’s about $28. That’s the cost of a final, irreversible entry into the most secure ledger on the planet. Not an apocalypse.
On top of that foundation, the Lightning Network grew. It’s often criticized for UX and complexity—and fairly so. But for its purpose, it works exactly as intended. Lightning is not a replacement for L1; it’s an overlay for micropayments. You pay an on-chain fee once to open a channel, then run thousands of operations inside it almost for free, and finally pay to close it. The blockchain records only the net result. Lightning fees are measured in fractions of a cent, and the network already holds several thousand BTC in capacity—tiny relative to total supply, but with enormous turnover. Bitcoin doesn’t have to scale on L1. It scales at the edges while keeping its conservative core.
Now Ethereum. It became a payment network, but it didn’t become the payment unit. Value moves across it constantly—just usually not in ETH. Stablecoins did their job. Why send ETH peer-to-peer when you can send tokenized dollars and ignore volatility? As a result, ether became not “money,” but the fuel of a computer that executes code, stores state, and guarantees that rules cannot be rewritten retroactively.
And that isn’t theory. In 2025, Ethereum was tested by reality. After a major exchange infrastructure hack, assets worth roughly $1.5 billion moved out on L1, and Ethereum did not “roll back” history—because it already knew the price of such decisions. The funds were gone, and the network held. For contrast: in newer ecosystems like Sui, after hacks worth hundreds of millions, the protocol simply pressed a pause-and-rollback button. Fast. Effective. And absolutely centralized. Ethereum chose a different road—and paid for it with a reputation for being iron, not plasticine.
Gas in Ethereum is not a “transfer fee,” but the price of computation. Each smart contract is a small program, and the network honestly charges for the resources it consumes. When there are many such programs, the base fee rises. Again, it’s an auction—but now for compute and data. The dollar price of ETH multiplies the result, but it doesn’t define it.
To understand Ethereum’s scaling, you first have to understand what it decided to become. The base layer is the backend of a decentralized internet. Its job is not to serve millions of tiny operations, but to guarantee immutable state and final settlement. Everything else—speed, UX, cheapness—is deliberately pushed outward.
The next step was separating data from execution. For a long time, rollups paid for publishing data like ordinary smart contracts, via calldata. It worked, but it was expensive. With blobs, Ethereum introduced a separate, cheaper space for temporary L2 data. A blob is needed only at verification time—then it disappears. Ethereum stopped being a “hard drive” and became a center for verification and consensus. Security is permanent; data is ephemeral.
Then come rollups. Arbitrum and Optimism are optimistic rollups. Base made this model mainstream. Today, most user transactions in the Ethereum ecosystem already live on L2, where fees are measured in cents.
The key point is the base layer’s role. Ethereum is deliberately turning into a “court and notary.” L1 is where things are expensive but final. That’s where corporate settlement, issuance and accounting of real-world assets, and finalization of large states happen. If ETH costs even $62,000—as Tom Lee likes to say—then a $6–$12 fee for these operations is not a problem.
Another major line is zero-knowledge. zkSync and other ZK rollups prove correctness mathematically. This isn’t only about speed. It’s the foundation of the future Web3: proving you’re human, that you have a right, that you’re not a bot—without revealing data. In the age of generative AI, the idea of Proof of Human stops sounding like science fiction.
The overall picture is surprisingly calm. Bitcoin can be a million because it doesn’t need to move often, and when it does, SegWit and Lightning help. Ethereum can be tens of thousands because gas is the price of computation—and computation scales via rollups and blobs. Fees don’t put a ceiling on price. They simply reflect demand for trust.
Blockchains don’t break from growth. They mature. And the more valuable the asset becomes, the stronger the incentive to make using it even cheaper.