The Stellar Development Foundation (SDF) has unveiled Project Thruput, a technology preview made public on Wednesday 28 October as its Meridian 2026 conference opened in Lisbon. It is a new execution engine with no per-transaction fees, and the foundation says it processed 7.02 million payments per second in early benchmarks. That number is the execution capacity of a single validator made of eight AWS machines, running one test workload, and the SDF post itself says the benchmarks do not measure end-to-end network throughput. Thruput runs on FutureNet, Stellar’s network for developers, not on the Stellar XLM mainnet, where the token is used and payments settle.
For scale, the live Stellar mainnet handled about 14.16 million transactions on 23 October, an average of roughly 163.9 per second, according to figures published by Crypto Briefing. The two numbers do not measure the same thing. The foundation has given no date and no plan for the mainnet. According to Crypto Briefing, it has not yet decided how the new network will relate to Stellar at all.
On the same morning, the tokenization firm Securitize said it had added Stellar to the blockchains it supports, CoinDesk reported. That is a separate announcement, made at the same conference, and nothing in either source says Securitize will use Thruput. Our Stellar (XLM) page explains how the existing network works, and the dates and speakers of the conference are on our Meridian 2026 event page.
What Stellar Project Thruput changes in how transactions run
Most blockchains put transactions in one strict global order. Some run unrelated transactions in parallel, but once two of them touch the same balance or the same data, they go back to waiting in line. The SDF argues that this is why extra hardware often sits idle: the queue, not the machine, sets the pace.
Thruput takes a different route. All transactions in a block start from the same snapshot of the ledger and run side by side, even when they change the same state. Their compatible changes are combined when the block is committed. The foundation’s example is a balance: a transaction records “subtract 5” rather than “the balance is now 95”, so many payments can hit one account in the same block and still add up correctly. Two payments still cannot both spend the same last dollar: the combined changes must respect the balance’s limits.
Contracts can also set limits that apply across transactions, such as a cap on the total withdrawn from a blockchain that runs exactly as written when it is called, holding and moving funds according to its code, without an intermediary able to change the outcome.">smart contract holding reserves of two or more tokens, supplied by users, that a decentralized exchange uses to execute trades automatically at a price set by a formula.">pool in one block. Developers pick storage types that tell the engine which changes can be merged and which limits must hold. The SDF describes the result as “a technology preview of a new execution model that removes the bottleneck”.
The other half of the idea is hardware. Each validator can spread its work over many machines that act as one, against the same shared state. There is no sharding, so an application does not have to split its data to get more capacity. Compatible updates to one account can run on different machines and still land in the same block.
The SDF page does not call Thruput a layer 2, a sidechain or a payment channel, and it does not discuss consensus, meaning how many validators would agree on blocks and how. The Defiant and Crypto Briefing describe it as a new network. Crypto Briefing reports that its final structure is undecided.
What the 7.02 million payments per second figure measures, and what it does not
The headline number comes from a scaling test. On one machine, execution capacity inside a single validator was 1.18 million payments per second. On eight machines, it reached 7.02 million, nearly six times more. The SDF ran the test “under heavy contention”: every payment moved between just 100 accounts, so most payments touched the same few balances, the situation the SDF says forces other designs to queue.
The workload was ERC-20 style token transfers, the standard used on Ethereum, run as RV32 code (the 32-bit RISC-V instruction set). The SDF says the measurement covers execution, signature checks and the work of opening and committing a block in Thruput’s core engine. Each figure is the median of three runs on AWS r8a.16xl instances. And the SDF adds a limit of its own: “They don’t measure end-to-end network throughput.”
The test shows how fast one well-equipped validator can process a block of payments. It leaves out what a public network adds: transactions arriving from users around the world, many independent validators that must agree, messages moving between them and blocks being shared. Crypto Briefing notes that the benchmark is internal and has not appeared in public analytics, so outsiders cannot check it yet.
Several details are missing from the published material we read. The page does not give the number of cores per machine, block size or timing, the size of the ledger state, or what such a validator would cost to run. The SDF says its benchmarks are reproducible and has published the code, but we found no outside team that has repeated them as of 28 October.
Stellar XLM mainnet today: about 164 transactions per second on average
The figures for the existing network come from Crypto Briefing’s report, not from a Stellar dashboard we checked ourselves. They describe actual traffic, which depends on demand, so they are not a ceiling.
| Measure | Figure | Source |
|---|---|---|
| Thruput, one machine in one validator (test) | 1.18 million payments/s | SDF benchmark, FutureNet code |
| Thruput, eight machines in one validator (test) | 7.02 million payments/s | SDF benchmark, FutureNet code |
| Stellar mainnet, 23 Oct 2026, daily average | about 163.9 tx/s (14.16 million tx) | Crypto Briefing |
| Stellar mainnet, 24 Oct 2026, peak | above 224 tx/s | Crypto Briefing |
| Mainnet record after Protocol 28 (100-block window) | 217.4 tx/s | Crypto Briefing |
| Theoretical ceiling of future mainnet releases | above 3,000 tx/s | Crypto Briefing |
| Mainnet fee at a recent peak | about $0.00012 per transaction | Crypto Briefing |
According to the same report, Protocol 28, nicknamed “Adapter”, went live in September 2026 and let validators handle consensus more efficiently, which is when the 217.4 record was set. Dividing 7.02 million by 164 gives a multiple above 40,000, but it compares a lab test of one machine cluster with the measured traffic of a live public network. The useful comparison will be the day Thruput runs across many independent validators under real load.
How “no transaction fees” would work on Thruput
The fee-free label applies to Thruput, the preview on FutureNet. Stellar mainnet still charges fees, tiny as they are. In Thruput, there is no per-transaction fee: a micropayment and a heavy calculation both run without a gas charge, the SDF says.
That does not mean unlimited computing. Gas, the unit that measures how much work a transaction asks of the network, still measures and caps each transaction. When demand exceeds capacity, an account’s share of the scheduling capacity depends on its unlocked balance of the native token. Scheduling does not spend those tokens, and there are no priority fees to jump the queue. Storing data works in a similar way: it locks part of the balance, which is released when the data is deleted.
On Stellar, the native token is XLM. The SDF post does not spell out what this model would mean for XLM if Thruput stays a separate network, and that is the gap Crypto Briefing points to: on most chains fees help keep spam out and pay for the network, and the foundation has not said what would do those jobs here. The SDF admits the design is unfinished. “There’s plenty left to figure out,” it writes, while setting the goal of free smart contract execution at scale.
No date and no mainnet plan; Protocol 30 “Bell” is the next real step
The post ends with an invitation, not a schedule. “We’re just getting started,” the SDF writes, asking developers to test the engine with contracts that were too expensive to run elsewhere and with AI agents that need no per-transaction fees. What testers find will help decide where the project goes next, the foundation says. There is no timeline for a mainnet, a separate launch or a token change.
The next dated change to the live network is something else. Protocol 30, called “Bell”, reaches testnet on 3 November and goes to a mainnet validator vote on 23 November 2026. None of the sources we read ties Thruput to Bell.
For XLM holders, nothing changes today. XLM traded near $0.1715 at 11:45 UTC on 28 October, up 2.17% over 24 hours according to our market snapshot. None of the reports we read links that move to the announcement.
Securitize adds Stellar to its supported blockchains
The second piece of Stellar news on 28 October came from Securitize, the company that issues and manages BlackRock’s BUIDL tokenized money market fund. Securitize said it had added Stellar to the blockchains it supports, next to Ethereum, Solana and Avalanche, according to CoinDesk, which published the story at 09:00 UTC and said the firms announced the agreement at Meridian 2026.
The firms said Stellar is fast and that assets on it can be set up for customer identification checks (KYC), clawbacks and sanctions compliance. “The Stellar network has built the fourth-largest ecosystem for tokenized real-world assets,” Securitize CEO Carlos Domingo said in a statement quoted by CoinDesk.
What is missing is the substance of the deal. CoinDesk does not say which funds or assets Securitize will issue on Stellar, how much, or when, and we found no announcement on Securitize’s own website or in other outlets at 11:55 UTC. Treat it as an added option on Securitize’s platform until a first product on Stellar is named. CoinDesk also disclosed that Stellar paid for its reporter’s trip to Portugal to cover the conference.
Securitize listed its shares on the New York Stock Exchange in July 2026 under the ticker SECZ and put its own stock on Solana and Avalanche at the same time. The shares rose as high as $13.70 on the first day and closed at $8.16 on Tuesday 27 October, CoinDesk reported. Our Securitize profile covers the transfer-agent business behind BUIDL and why the company has no token.
Tokenized funds keep arriving on Stellar this week
The Securitize news lands one day after another fund manager picked the same chain. On 27 October Janus Henderson and Centrifuge announced JQQQ, a tokenized Nasdaq-100 fund on Stellar, open only to non-US professional investors with a $100,000 minimum, as detailed in our report on Janus Henderson’s JQQQ. A retail investor in Europe cannot buy it.
Franklin Templeton already runs a tokenized money market fund on Stellar, and MoneyGram moves money across the network daily, according to CoinDesk. Crypto Briefing puts tokenized real-world assets on Stellar at $3.4 billion to $4.1 billion in late September and October 2026, and SDF blog headlines speak of a “$4B+” ecosystem. Quarterly stablecoin transfer volume reached $25.1 billion, Crypto Briefing adds.
Those businesses run on the mainnet as it is, with its small fees and its few hundred transactions per second at peak. Nothing announced on 28 October says any of them will move to Thruput.
What to watch after the Thruput benchmark
Four things will show whether Thruput becomes more than a lab result:
- An outside team repeating the 7.02 million figure from the published code, then tests across several independent validators instead of one.
- A decision by the SDF on what Thruput is: an upgrade path for Stellar, a separate network, or a research track.
- A worked-out fee and spam model, including what validators earn and what role XLM plays beyond scheduling.
- A named first asset from Securitize on Stellar, with an amount and a date.
The near dates are fixed. Meridian 2026 continues on 29 October in Lisbon, and Bell reaches the Stellar testnet on 3 November, ahead of the mainnet vote on 23 November.