Rust entered the TIOBE top 10 for the first time in July 2026 and is still there in September, at #10 with 1.34%, two places above Go. In the same twelve months the number of UK permanent job ads citing Rust fell from 398 to 278, and the median advertised salary dropped 10% to £90,000.
Both of those are true, and most articles about Rust in fintech only tell you the first one. This guide was first written in March 2026 and refreshed on 6 September 2026 with the current numbers, because "most wanted" is a claim that should survive a bad hiring year. It does, but not for the reasons the language advocates give.
Is Rust Still in Demand in Fintech in 2026?
Yes, relative to everything else, and that qualifier matters.
ITJobsWatch's six-month window to 6 September 2026 shows the whole UK programming-language market contracting. Rust ads are down 30% year on year. Go ads are down 52%, from 976 to 471. Java is down, Python is down, C++ is flat. Against that backdrop, Rust holds the highest median salary of the five:
| Language | UK permanent ads (6 months) | Median salary | Change |
|---|---|---|---|
| Rust | 278 | £90,000 | -10.0% |
| Go | 471 | £80,000 | -11.1% |
| Java | 2,872 | £70,000 | -6.7% |
| Python | 6,904 | £70,000 | -3.5% |
| C++ | 1,502 | £65,000 | flat |
In London specifically: 90 Rust ads, median £100,000, 90th percentile £144,300. A year earlier the London median was £110,000. So the premium narrowed but did not close, and the top of the range is where the payments and trading roles sit.
The demand signal on the language side is the opposite of the hiring signal. Rust was #14 on TIOBE in March 2026 when I first wrote this, #18 in the summer of 2025, and hit #10 in July 2026. Stack Overflow's 2025 survey had it as the most admired language for the ninth consecutive year at 72%. The State of Rust 2025 survey, published 2 March 2026 from 7,156 responses, found 84.8% of people using Rust at work said it helped them achieve their goals, and the top worry was "not enough industry adoption" at 42.1%. That last figure is the honest one. Rust developers are not short of enthusiasm for the language; they are short of employers.
Which Fintech and Payments Companies Use Rust in Production?
Fewer than the conference talks suggest, and the ones that do are serious about it.
Kraken is the clearest case. Its December 2025 engineering post describes millions of lines of Rust across hundreds of services, with Rust as the primary backend language "by a wide margin", REST APIs on Tokio and an in-house Kafka stream framework. That migration was from PHP, not Go. TrueLayer is the strongest UK payments example. Its PayDirect payment lifecycle runs as a Rust state machine, it has published Rust crates since 2021 (reqwest-middleware, ginepro), it hosted Rust Global: London at its office in February 2025, and it was advertising a "Rust Backend Engineer, Payments" role this year. Stripe uses Rust for tooling rather than the payment path: rubyfmt, the formatter it runs across a 42-million-line Ruby codebase, is written in Rust. Tempo, the payments chain incubated by Stripe and Paradigm, is a separate entity whose node is Rust on the Reth SDK, live on mainnet since 18 March 2026 with Visa running a validator. Circle's Arc chain runs Malachite, a BFT consensus engine written in Rust, with mainnet scheduled for 16 September 2026. Base, Coinbase's L2, shipsbase-reth-node as its execution client. Bitpanda advertised a Rust and Python role for its trading and pricing infrastructure in June 2026. Fireblocks and Anchorage list Rust in security and platform engineering roles, which is job-ad evidence rather than a blog post, so treat it as such.
Now the ones people assume and get wrong. Monzo is a Go shop and has never published a Rust post. Wise's 2025 stack write-up is Java and Kotlin across a thousand-plus services. Revolut's engineering ads are Java. Paxos prefers Go and merely "values" Rust. Jane Street is OCaml. If you are learning Rust to get into a UK neobank, check the actual stack first, because the firms hiring for Rust in UK fintech are the infrastructure and crypto-settlement companies, not the consumer apps.
Why Do Payment Systems Keep Landing on Rust?
Three reasons, and none of them is "it's fast".
Memory safety is a money-safety property. A use-after-free in a settlement engine is not a crash; it is a ledger that disagrees with the bank. Rust's ownership model removes that class of bug at compile time, and the compiler is cheaper than the reconciliation team. No garbage collector in the critical path. An Open Banking checkout has to complete bank authentication, initiate the payment and return inside a few seconds. FX conversion has to execute at the quoted rate. A GC pause at the wrong moment is a timeout the customer sees and a stale rate the treasury desk pays for. Concurrency the compiler checks. Webhooks from three providers arrive in parallel, settlement confirmations need matching against pending transactions, and every one of those must be processed exactly once.Send and Sync are not ergonomics; they are the reason a data race in the matching loop fails to build rather than fails in production at 3am on a Bacs cut-off.
What Do Rust Developers Actually Build in Payments?
The domain problems, not the language, are what make the work hard. In my own payments work Rust ends up in:
Payment orchestration: routing across Open Banking APIs and per-bank quirks. The VRP consent model alone has a 40-character idempotency key limit, a mandatory one-second resend delay and per-bank rate limits with no standard, and the service that enforces all of that is exactly where you want the type system doing the remembering. Event pipelines: webhook ingestion, idempotent processing, retry with backoff that differs per failure class. Every payment notification must land once, even when the provider sends it twice. Settlement and ledger: matching incoming funds to pending transactions, executing FX, updating double-entry positions atomically. PostgreSQL with compile-time checked SQL is the usual pairing. Message construction for rails: FedNow has no REST payment endpoint, so the work is buildingpacs.008 and handling the ISO 20022 structured-address rules that Swift has now deferred. Strongly typed message builders are where Rust's enums earn their keep.
Compliance services: KYC/KYB screening, Travel Rule checks and transaction monitoring, all of which must not add latency to the payment flow.
Rust vs Go for Fintech Backends in 2026
Both are good choices and most teams I know run both. The difference is where the money moves.
| Rust | Go | |
|---|---|---|
| TIOBE, Sep 2026 | #10, 1.34% (+0.33pp) | #12, 1.10% (-1.22pp) |
| Stack Overflow 2025 usage | 14.8% | 16.4% |
| UK ads, 6 months to Sep 2026 | 278 (-30%) | 471 (-52%) |
| UK median salary | £90,000 | £80,000 |
| Hiring pool | smaller, more specialised | larger |
| Compile-time data-race prevention | yes | no |
| GC pauses | none | yes, short |
| Time to first production service | slower | faster |
JetBrains' 2025 developer ecosystem report has Go at 11% and Rust at 10% on "want to learn", with both among the languages developers see as having the highest growth potential, and JetBrains has reported one in six Go developers considering a switch. I could not find a single named fintech team that publicly moved from Go to Rust in 2026, and I looked. The migrations that get written up are from PHP (Kraken) and Ruby. That tells you Rust is winning greenfield infrastructure, not displacing working Go.
What Is the Rust Fintech Ecosystem Like in 2026?
Mature at the bottom, thin at the top. crates.io passed 329,000 crates and 429 billion downloads in September 2026, more than double the 160,000 I quoted in March. The pieces a payments service actually depends on:
- Tokio 1.53.1 (July 2026): the async runtime, and effectively the only one that matters.
- Axum 0.8.9 (April 2026): the web framework. 0.9 is on main with breaking changes and no release timeline; there is no 1.0, and a payments team should pin accordingly.
- SQLx 0.9.0 (May 2026):
query*()now takesSqlSafeStr, static strings by default, which is a compile-time guard against string-built SQL. The project also moved to a new organisation, so check your dependency source. - Serde 1.0.229 (July 2026): serialisation, now on syn 3.
- Rust 1.98.0 (20 August 2026): the current stable.
The other ecosystem fact from 2026 is the one nobody wanted. On 20 August, malicious versions of arrayref, internment and append-only-vec pulled in a typosquatted proc-macro1 whose build script downloaded a payload; they were live for between 86 and 107 minutes, and arrayref has 245 million all-time downloads and 403 direct dependents. I wrote up what that attack means if your build touches money. The short version: cargo vet and a private registry mirror are not optional for a payments codebase any more.
What Do Rust Developers Earn in the UK in 2026?
From ITJobsWatch, six months to 6 September 2026:
| Ads | Median | 25th pct | 75th pct | 90th pct | |
|---|---|---|---|---|---|
| UK | 278 | £90,000 | £63,750 | £100,000 | £120,000 |
| London | 90 | £100,000 | £144,300 |
The 25th percentile fell hardest, from £85,000 to £63,750, which says the market for junior and mid Rust roles cooled far more than the senior end. The 90th percentile in London at £144,300 is the trading-infrastructure and crypto-settlement bracket, and those roles want payment domain knowledge as much as they want the language.
Ignore the self-reported salary aggregators for Rust in the UK. The sample sizes are tiny and the figures contradict the advertised ranges.
Will Rust Replace Go in Fintech?
No, and I would not want it to.
My prediction: by 2028 the pattern is settled at roughly what the serious shops run today. Rust owns the money-moving path, the signing infrastructure and anything with a hard latency budget. Go owns the edges, the internal tools and the services where hiring speed is the constraint. The teams that try to make everything Rust burn their velocity on services that never needed it, and the teams that try to make everything Go end up with a ledger they cannot fully trust.
The hiring numbers support that reading. Rust ads fell less than Go's in a shrinking market and kept the salary premium, which is what you would expect if Rust roles are concentrated in the parts of the stack that do not get cut when budgets do.
What This Means for Developers Learning Rust for Fintech
If you want the roles at the top of that salary table:
1. Target the companies that actually run Rust. Kraken, TrueLayer, the settlement-layer teams at Circle and Coinbase, the trading-infrastructure firms. Not the neobanks.
2. Learn a payment rail properly, not just the language. Open Banking VRP, FedNow's ISO 20022 surface, or Verification of Payee. A candidate who can explain why PeriodAlignment matters is rarer than one who can explain lifetimes.
3. Build the missing domain library as your portfolio. A typed pacs.008 builder, an idempotent webhook processor with per-failure-class retry, or a double-entry ledger on PostgreSQL with the isolation level justified in the README. Those are the things employers cannot download.
4. Know the supply-chain tooling. cargo vet, cargo audit, a registry mirror. After August 2026 every payments team interviews for this.
5. Pin your framework versions and say why. Axum without a 1.0 and SQLx with a fresh 0.9 are fine in production if you can articulate the upgrade policy.
The demand for Rust in fintech is real. It is also narrower than the hype, concentrated in infrastructure rather than product, and priced accordingly. That is a better market to aim at than a broad one, provided you aim at the right part of it. More on the systems side of this at Tom Wang.