Bryan Cantrill offered a thought-provoking judgment at Rust NYC: as large language models take on more code generation and modification, competition among programming languages is expanding from “who runs faster” to “who makes it easier for AI to maintain reliably.” Rust’s performance advantage remains, but development efficiency is becoming a more direct cost factor.
What this is
This talk focused on Rust’s position in the era of large language models. Rust is a systems programming language designed around memory safety and concurrency reliability, intended to prevent a common class of program crashes and vulnerabilities. Cantrill has long advocated its adoption in servers and infrastructure.
The change brought by large language models is not simply faster code writing. It makes code generation, testing, review, and iteration more frequent. Whether a language works well with AI collaboration, whether its dependencies are clear, whether compiler errors are easy to locate, and whether existing engineering systems can absorb automatically generated code have all entered the language-selection equation.
Industry view
Optimists argue that Rust’s explicit rules and strong type constraints can reduce subtle errors in AI-generated code. As infrastructure complexity rises, the cost of such errors may become more valuable than raw generation speed.
But the risks are equally clear: Rust’s learning curve, compile times, and relatively high engineering barrier may offset the efficiencies AI brings. For most teams, “helping AI write code faster” remains easier to quantify than “building more controllable systems software.” Rust is therefore more likely to retain its advantage first in high-cost settings such as databases, operating systems, and infrastructure, rather than become the default choice for every AI project.
Impact on regular people
- For enterprise IT: When purchasing development tools, companies should look beyond marketing claims and ask about the defect rate of model-generated code, test coverage, and post-deployment maintenance costs.
- For individual careers: The value of programming is shifting from writing code line by line toward defining requirements, reviewing results, and understanding system boundaries. Business knowledge and verification skills now matter more than simply writing code faster.
- For the consumer market: Software delivery may continue to accelerate, but prices are unlikely to fall in step. Companies will still have to pay for reliability, computing power, and ongoing maintenance.