What programming languages does robotics use in 2026

Rust is now more common than C++ in robotics job postings. That is a genuinely strange sentence to write. C++ is still the actual language most production robots run, and every senior robotics engineer knows it. But when robotics companies write their job posts, they mention Rust 341 times and C++ 23 times.

I pulled the tech stack tags from every active robotics posting to see what companies are actually asking for. What the postings say and what the code contains are not the same thing, but the gap between them is itself the story.

The programming languages picture

Actual language mentions across 5,540 active robotics postings:

Four languages get real mentions. Everything else (Go, Java, Kotlin, Swift, C, Zig) shows up in single digits or not at all. Robotics is a Python-plus-C++ industry with a real Rust wave and a MATLAB fortress in the defense-controls subset.

The frameworks and tools picture

The other stuff robotics companies ask for, roughly in three buckets:

ML frameworks

Simulation and robotics-specific tooling

Cloud and infra

Hardware-adjacent (firmware, controls, embedded)

Some of these overlap heavily with hardware engineering roles, which is why counts like PCB design (230) and Embedded (405) look high for a "programming languages" post. They're not languages, but they're the technical skills companies pair with the languages in the same postings.

What the numbers mean

Python is the default reporting language, not the default implementation language. Every ML-adjacent posting mentions Python. Every simulation posting mentions Python. Almost every job posting that's aimed at "someone who codes" mentions Python by default. It doesn't mean the robot runs Python. It means Python is what companies expect candidates to have used at some point.

C++ is invisible in postings because it's assumed. Every serious robotics stack has C++. Companies don't bother listing it because if you can't do C++ they wouldn't consider you for the role. The 23 mentions are outliers where a company felt the need to specifically call it out (usually because the role is unusually C++-heavy: firmware, real-time controls, hardware bring-up).

Rust at 341 mentions is real. Not necessarily because Rust dominates production stacks yet, but because companies are hiring for it. The signal is intent: robotics companies want Rust engineers because they're planning to migrate parts of their stack to Rust over the next 2-3 years. If you're a mid-career engineer thinking about the next language to learn, Rust is the safer bet in robotics right now than in most other verticals.

MATLAB at 359 mentions is a defense-and-controls story. Every big controls engineering role at Northrop, Lockheed, Boeing, or their supplier ecosystem still uses MATLAB and Simulink. It's not dying. It's just concentrated in the defense-and-legacy-aerospace side of the industry.

The ML framework fight

PyTorch: 232 mentions TensorFlow: 86 mentions JAX: 62 mentions

PyTorch has decisively won the mainstream framework battle in robotics. TensorFlow is fading fast, and most of its mentions are at large companies (Google-adjacent, some Nvidia legacy code, some robotics groups at big corporates that haven't migrated yet). JAX is the interesting rising number. Google/DeepMind-adjacent teams and some research-forward companies (Physical Intelligence, Wayve) are using it.

If you're picking a framework to specialize in for the robotics job market, PyTorch is the clear answer. JAX is the "hedge" for the top research-forward companies.

The simulator landscape

Isaac Sim: 51 (Nvidia) MuJoCo: 42 (DeepMind, now open-source) Gazebo: 22 (ROS legacy) Unreal: 23 (Wayve, some AV) Drake: 3 (TRI, small footprint)

Isaac Sim is winning among the well-funded startups (Nvidia is subsidizing adoption aggressively). MuJoCo is the winner in the RL research community and has a real toehold in production teams that came out of DeepMind lineages. Gazebo is dying. Drake is a research niche.

Skills you can actually see in job posts

Beyond languages, the specific skill tags that show up most often:

• Computer vision (2,324) and sensor fusion (2,221) are almost universal
• Motion planning (168), SLAM (63), MPC (15) are role-specific signals
• Reinforcement learning (113) and imitation learning (37) are the ML-for-robotics tags, both growing
• Isaac Sim (51), MuJoCo (42), Drake (3) are the simulator tags

The gap between "has a real ML/perception background" and "just knows Python" shows up in these. If you can talk credibly about sensor fusion algorithms, ML for perception, and one of the simulators, you're inside the top 20% of candidates for most robotics roles.

Practical takeaways

Two things if you're planning skills to learn.

Rust is worth prioritizing over Go, over Zig, over any other systems language for robotics. The demand signal is unambiguous even though production stacks haven't migrated yet.

If you're a mid-career engineer without robotics-specific experience, the fastest bridge into the field is not learning ROS. It's picking up either Isaac Sim or MuJoCo and building something demonstrable. Simulation-first is the modern robotics onboarding path, not "get a robot and learn ROS."

Methodology

Tech stack tags come from the techStack JSON array on each active job posting. Most companies populate this field via their ATS integration, so mentions reflect what employers tag on their own postings, not scraped text from descriptions. I aggregated across 5,540 active robotics-niche postings pulled April 28, 2026.

If a tag doesn't appear it usually means one of two things: companies didn't tag it, or the field is too tiny to show up in a 40-row cutoff. C++ is the classic case where the tag is absent but the underlying use is universal.