Highlights from the Stack Overflow Developer Survey 2025 (vs 2024)
Python Is Back on Top, Docker Is in the Air, and AI… Is Getting Tiring.
TL;DR: Python sees its biggest leap in a decade, Redis is on a tear, Docker becomes “the default,” FastAPI gains momentum, and developers use AI more — but like it less. A paradox worth paying attention to.
Python’s Big Comeback (and the End of “JS Forever”?)
Python jumped 7 percentage points year over year to reach 57.9% among all respondents — the fastest growth it’s had in years. It’s now firmly the “AI + data + backend” language. JavaScript still holds #1 at 66%.
Among those learning to code, Python is the clear leader (71.8%), up from 66.4% in 2024. The next wave of developers is clearly thinking “Python first.”
Why it matters: If you’re building products, training material, or hiring pipelines, Python + AI ecosystem is the safest bet. From a company perspective: onboarding docs, code examples, and training should start with Python.
Docker Is Everywhere, FastAPI Picks Up Speed, Phoenix Is the “Secret Favorite”
Docker saw a historic jump (+17 percentage points) to 71.1% usage — the biggest single-year gain of any tool in the survey. In 2024 it was already popular among pros (59%). Now it’s near-universal.
FastAPI grew by 5 percentage points, riding the wave of Python backend development.
Phoenix (Elixir) remains the most admired web framework (79%) — small community, big satisfaction. It’s a perfect “edge-tech” for employer branding: rare skills, high loyalty.
Why it matters:
Operationally — standardize CI/CD with Docker; it’s no longer a competitive edge, it’s table stakes.
Product-wise — if you need quick APIs, FastAPI aligns perfectly with Python’s surge.
Brand-wise — “we build in Phoenix/Elixir” can be a recruitment magnet.
Redis Is Having a Moment
Redis is up 8 percentage points year over year (from ~20% to 28%). As applications get more complex, caching and in-memory data structures are now a “must have.” PostgreSQL remains the most popular database.
Why it matters: If you want quick performance wins, add Redis first. It’s also a great click-friendly topic: “How we cut TTFB by 60% with a simple cache.”
The AI Paradox: More Use, Less Love
84% of respondents use or plan to use AI for development (2024: 76%). Among pros, 51% use it daily.
Yet positive sentiment fell to ~60% (from 72% last year).
More developers distrust AI (46%) than trust it (33%). The top frustration? “AI gives almost correct answers” (66%), which then take extra time to debug. Many devs are back on Stack Overflow… fixing AI’s mistakes.
LLMs: OpenAI GPT is still the most used (81.4%), but Claude Sonnet is the most admired and #2 in “want to use” (33%). Gemini Reasoning is also worth watching.
Why it matters:
AI is eating the workflow, but governance and validation are the new productivity currency.
Work: Agile Compromise Over Dogma
Remote work is ~32% globally, 45% in the US. In Germany, “your choice” setups (full flexibility) hit 21%. Less ideology, more pragmatism.
Stack Overflow Didn’t Die — It’s Thriving
81% have an account (2024: 76%), and 82% visit at least several times per month. One-third of visits are triggered… by AI mistakes. Younger devs want more lists, articles, and chat formats.
Strategic Takeaways
1. Go Python-first — in both code and narrative.
Examples, starter projects, docs, and training in Python will resonate with the largest (and fastest-growing) audience.
2. Add a “Redis Pass” to your roadmap.
Look for caching opportunities before buying more hardware. It’s cheap, measurable ROI — and it makes for great engineering case studies.
3. Treat Docker as your org’s operating system.
Standardize images, policies, scans, and signing. The competitive edge is now in process quality, not adoption.
4. AI: less magic, more control.
Introduce an “AI QA Gate” with tests and linting for hallucinations.
Use defensive prompts and counter-prompts (e.g., “give me three reasons this won’t work”).
Track “time to correct build” for AI-generated code as a key KPI.
5. Experiment with FastAPI / Phoenix.
FastAPI rides the Python wave; Phoenix delivers developer joy. Both make for high-signal conference talks and blog posts.
6. Content for younger devs: tool lists, interactive tutorials, and “AI gone wrong” repair guides are rising in popularity.
Final Word
The 2025 survey shows a market without illusions: we use AI every day, but we’re building guardrails around it. Python is in turbo mode thanks to the AI wave, Redis and Docker are baseline infrastructure, and FastAPI and Phoenix are rising stars in prestige.
If you have limited resources but want results: Python + Docker + Redis as your base stack, controlled AI workflows, and a couple of bold experiments with FastAPI/Phoenix. The rest is execution.
More on our IT blog.