By Ben Muse, VP of AI at Rankplan, operator of PSP Dex.
What the problem actually is
Bitcoin’s base layer processes a limited number of transactions. Blocks arrive roughly every ten minutes, and each block has a maximum weight of 4 million weight units, a limit introduced by the SegWit upgrade in 2017. Depending on the size mix of the transactions included, that works out to roughly 3 to 7 transactions per second.
For comparison, card networks such as Visa process thousands of transactions per second on average, with capacity well above their average load. If every payment in the world had to settle directly on Bitcoin’s blockchain, the chain could not keep up. That gap between base-layer capacity and potential global demand is the bitcoin scalability problem.
Why Bitcoin does not simply raise the limit
The constraint is deliberate, and it comes from how Bitcoin stays decentralised. Every full node independently verifies every transaction and stores the chain’s history. If blocks were made much larger, running a node would demand more bandwidth, storage, and processing power, and fewer people could do it on ordinary hardware. A network only a few companies can verify is a network those companies can control. 
Bitcoin therefore treats base-layer block space as a scarce settlement resource rather than a retail payment rail. Demand for that space is rationed by fees: when many users compete for the same block, the transactions offering the highest fees are confirmed first, and low-value payments are priced out. Congestion episodes in 2017, 2021, and 2023 showed this clearly, with average fees spiking each time demand surged.
The layered answer
The approach Bitcoin has converged on is to scale in layers, keeping the base layer conservative and pushing volume to systems built on top of it.
SegWit (2017). Segregated Witness restructured how transaction data is counted, raising effective capacity and fixing transaction malleability. The malleability fix was a prerequisite for secure payment channels.
The Lightning Network. Lightning lets two parties lock bitcoin into a shared channel and exchange payments off-chain, settling the net result on the base layer only when the channel closes. Payments can be routed across multiple channels, so a user does not need a direct channel with every recipient. The result is near-instant payments with very low fees, at the cost of new constraints: funds must be locked in channels, routes must have enough liquidity, and receiving payments requires either existing channel capacity or a service that manages it.
Other layers and techniques. Exchanges and payment processors batch many withdrawals into single transactions. Sidechains such as Liquid settle transactions on a separate chain pegged to bitcoin. Taproot (2021) made complex spends cheaper and more private, which improves the efficiency of channel operations. Research continues into channel factories, Ark, and other designs that share one on-chain transaction among more users.
Where this leaves the debate
Nobody serious disputes the arithmetic of the base layer. The live disagreement is about emphasis: how much scaling should happen on Lightning and similar layers, how usable those layers need to become for non-technical users, and whether fee pressure on the base layer will fund security in the long term as the block subsidy halves every four years. What the history shows is that Bitcoin has consistently chosen to protect cheap verification over raw throughput, and to scale everywhere else.
Key facts: Block target time about 10 minutes. Block limit 4 million weight units (SegWit, 2017). Base-layer throughput roughly 3 to 7 transactions per second. Lightning settlement off-chain, with the base layer used to open, close, and enforce channels.
