Buried below the headline story in TIOBE's September 2026 index — Rust's first-ever top-10 finish — is a quieter, equally interesting trend: Julia has climbed to #21 with a 0.74% rating, sitting just one spot behind COBOL at #20. It's not in the top 20 yet, but the direction of travel has been consistent, and it's coming almost entirely at the expense of one very specific, very established competitor: MATLAB. Here's what's actually driving it, and — more usefully — who should bother learning it.
Where things stand this month
For context, the top of the index hasn't moved much: Python still leads comfortably at 17.76% (even as its share keeps eroding year over year), C sits at #2 with 10.28%, C++ is #3 at 8.67%, and Java rounds out the top four at 7.54%. Julia, at #21, is well outside that group — TIOBE's rating is logarithmic in practice, so the gap between #4 and #21 is enormous. What makes Julia worth a separate look isn't its absolute position; it's that it's a genuinely new language (first stable release in 2012, 1.0 in 2018) gaining ground in a specific, well-defined niche, rather than a general-purpose language slowly bleeding or gaining share across the board.
The MATLAB connection
The clearest signal in this month's data is what's happening to MATLAB, which continued sliding and dropped two more spots to #27. That's not a coincidence sitting next to Julia's climb. Julia was built from the ground up to solve the exact problem MATLAB has always been used for — numerical computing, scientific simulation, and technical computing — but as a free, open-source language with performance close to C for the kind of tight numerical loops that make interpreted languages like Python or R slow without workarounds (NumPy vectorization, Cython, etc.). For research groups, universities, and labs that have historically paid for MATLAB licenses, Julia's pitch — write natural, readable, loop-heavy numerical code and get compiled-language speed without leaving a high-level language — has become genuinely competitive over the past few years, and this month's numbers are one more data point in that direction.
TIOBE's own caveat is worth taking seriously
It's worth quoting TIOBE's own analysis directly here, because it's more measured than a "Julia is taking over" headline would suggest: the index's commentary notes Julia's growth "remains concentrated in relatively niche domains and has not yet managed to break through as a general-purpose programming language." That's an accurate read. Julia isn't showing up as a serious contender for web backends, mobile apps, or general application development — its adoption is specific to numerical computing, data science, and scientific/technical fields, and there's no strong signal that's changing. This is a language having a real moment in its lane, not a language crossing over into new ones.
So, should you learn it?
The honest answer depends entirely on what you actually do:
- Learn it if: you're in scientific computing, numerical simulation, data science research, or any field where you'd currently reach for MATLAB or NumPy-heavy Python for performance-sensitive numerical code. Julia's combination of readable syntax and near-C performance is a genuine advantage in exactly that lane, and the ecosystem (DifferentialEquations.jl, Flux.jl, and a mature package manager) has matured enough to be a realistic production choice, not just an academic curiosity.
- Skip it if: you build web apps, mobile apps, or general backend services. Nothing about Julia's climb changes what you should be learning for those domains — Python, JavaScript/TypeScript, Kotlin, and Swift remain the languages with the ecosystems, job markets, and tooling maturity that actually matter there.
- Consider it as a second or third language if you're already comfortable in Python for data work and occasionally hit performance walls that push you into writing C extensions or fighting with vectorization to get NumPy fast enough — Julia is a genuinely pleasant answer to that specific, recurring problem.
How this fits the bigger language landscape
Zoom out and this month's index tells a coherent story about specialization: Rust cracking the top 10 reflects systems-level and performance-critical general-purpose adoption; Julia nearing the top 20 reflects a narrower, domain-specific win in numerical computing. Neither is really competing with Python for the "what should I learn first" spot — Python's continued dominance at #1, even in decline, reflects its role as the default glue language across data work, scripting, and the AI tooling ecosystem that sits underneath most of the coding assistants developers use today. Julia and Rust are best understood as complements to that foundation for specific, demanding use cases, not replacements for it.