Stellar can now process 7 million payments per second. What does that even mean?
Picture a supermarket checkout. One cashier scans items one at a time, and the line grows. A traditional blockchain works the same way: transactions line up strictly one after another.
The Stellar team asked: do we even need that line? If you’re paying for coffee while someone else is paying for a taxi, they don’t need to know who came first. So in Project Thruput, they removed the ordering within a block. Payments are processed simultaneously, as if lots of checkout lanes had opened at once.
The main rule remains the same: you can't spend money you don't have. Suppose there's 100 coins in the account, and payments of 30, 40, and 50 come in. The first two go through, leaving 30 in the account, and the system rejects the third. The runtime enforces application limits and won't let you spend more than you have.
How it scales. Previously, to make a blockchain work faster, it was divided into parts (sharding). This works differently: one validator distributes the work across many machines, which act as a single unit with shared state, so there's no need to split the blockchain into shards. One machine handled 1.18 million payments per second; eight handled 7.02 million. For comparison, Stellar previously processed around 150 transactions per second under real-world load.
Another feature: there's no fee for each transaction. When the network is congested, space goes to those with more XLM in their balance: the more XLM an account holds, the more capacity it gets.
What's next. For now, this is an experiment on a test network. Developers can already try it out for themselves, since the code and benchmarks are open. Stellar is aiming for a future where payments are made not only by people, but also by billions of AI agents.
Bottom line: this is a bold promise for the future, but not a finished product yet. The 7-million figure is impressive, but whether it will make it to the live network, and in what form, remains to be seen.
