Technology 2026 | Stack Overflow Developer Survey

What’s new, what’s old, and something for everyone in technologies surveyed this year

The technology leaderboards have been a constant feature in our Developer Surveys from the beginning. That’s the core of the software developer’s toolkit. What’s changed are the categories of tech on this list: cloud providers, coding agents, LLMs, and security ops. Many of the top technologies on some of these lists have been there for years. Some lists show clear leaders in the space. But there's still plenty of room for alternative technologies to gain ground- each of these lists includes a long tail of tools in the single digits for usage (and we had to cut the lists off somewhere).

Most popular technologies

For the most used programming languages, JavaScript is at the top of the list for the 12th year in a row

View and share

JavaScript
SQL
HTML/CSS
Python
Bash/Shell (all shells)
TypeScript
Java
C#
PowerShell
C++

	Respondents	Percent
JavaScript	8,742	62.0%
SQL	8,237	58.4%
HTML/CSS	8,212	58.2%
Python	8,184	58.0%
Bash/Shell (all shells)	7,231	51.3%
TypeScript	6,171	43.8%
Java	3,885	27.6%
C#	3,826	27.1%
PowerShell	3,472	24.6%
C++	3,425	24.3%

The core stack: Technologies that stand the test of time

For the most used programming languages, JavaScript is at the top of the list for the 12th year in a row

Most popular technologies

Postgres continues reign as the most used database, used by a majority of respondents.

View and share

PostgreSQL
SQLite
MySQL
Redis
Microsoft SQL Server
MariaDB
MongoDB
Elasticsearch
Oracle
Dynamodb

	Respondents	Percent
PostgreSQL	6,984	57.9%
SQLite	5,083	42.1%
MySQL	4,235	35.1%
Redis	3,414	28.3%
Microsoft SQL Server	3,270	27.1%
MariaDB	2,446	20.3%
MongoDB	2,297	19.0%
Elasticsearch	1,784	14.8%
Oracle	1,135	9.4%
Dynamodb	1,116	9.3%

Postgres continues to rule in the database space. In an interview on Stack Overflow’s podcast Shireesh Thota (Corporate VP of Azure Databases at Microsoft) said "I tell everyone that Postgres is the Linux of databases... It's one of the best open source projects out there where the community really rallies. They care deeply." Is it also the AI gateway drug? That’s how Brian O'Grady (Head of Field Research at Qdrant) put it when he summarized in a separate interview with us the ease of starting with its vector capabilities: "I've called it almost like a gateway drug to vector search. As soon as you use pgvector, you're in."

“I tell everyone that Postgres is the Linux of databases... It's one of the best open source projects out there where the community really rallies. They care deeply.”

Shireesh Thota, Corporate VP of Azure Databases at Microsoft source

Builders are still building in the age of AI

Popular predictions of an AI-driven "SaaSpocalypse" turn out to be greatly exaggerated. Developers are building at high volumes, but their focus remains on internal utilities and glue code rather than replacing commercial software. 77% of developers build small scripts and automations, while 63% create team-level internal tools. 54% build customer-facing applications—a reminder that a vast portion of engineers work inside traditional enterprise sectors like retail, finance, and manufacturing. Vendor replacement sits at the bottom: Only 19% of respondents build in-house software to replace paid vendors.

But when you shift to look at what people build with AI, SaaS vendors might get nervous: 54% of those building with AI are replacing vendors. The good news is that this is limited to the AI builder subgroup, builders are more likely to be using AI in general for all projects. Vendor replacement got a lot of press this year, but it’s far from an AI breakout story—it's actually the workload where AI assistance sees the lowest ceiling for adoption.

Build vs. AI

Small scripts are most common for builders to work on, 19% say they’re building tools to replace paid vendors.

View and share

	Respondents	Percent
Small scripts/automations	10,304	77.3%
Team-level internal tools	8,398	63.0%
Customer-facing products	7,254	54.4%
Cross-team/platform tools	6,130	46.0%
Replacing paid vendors	2,482	18.6%

When cross-referencing these builds with AI tool usage, a subtle trend emerges: AI adoption steadily drops as system complexity and operational risk increase. When you buy tools, you offload security compliance, SLAs, and continuous maintenance. Replacing them with internally-built tools transfers technical debt back onto internal teams, right where AI errors create the most pain. Developers are happy to let AI generate low-risk automation scripts, but when it comes to business-critical SaaS, most teams still prefer paying a vendor over managing unmaintainable, AI-generated technical debt.

Build vs. AI

While only 19% say they’re building tools to replace paid vendors, of that cohort more than half are doing so with the help of AI.

View and share

	Respondents	Percent
Small scripts/automations	7,640	74.2%
Team-level internal tools	5,588	66.5%
Customer-facing products	4,754	65.5%
Cross-team/platform tools	3,831	62.5%
Replacing paid vendors	1,349	54.4%

Frontier models’ next evolution

While OpenAI and Anthropic sit neck-and-neck at the top of general LLM usage (74% and 75% adoption, respectively), practitioner sentiment reveals a rift between usage and the desire to keep using it. Anthropic holds a clear lead in forward-looking interest, with 49% of developers wanting to work with Claude over the next year, compared to 32% for OpenAI. Spend tracking data from Ramp mirrors this shift, showing Anthropic capturing market share rapidly as engineering teams transition from simple chat interfaces to structured, programmatic agent architectures.

Most popular technologies

The two companies focused on frontier models are the top two in LLM usage: Anthropic at 75% and OpenAI at 74%.

View and share

	Respondents	Percent
Anthropic	8,871	74.6%
OpenAI	8,834	74.3%
Google AI	6,483	54.5%
Microsoft	3,142	26.4%
DeepSeek	2,253	18.9%
Alibaba Qwen	1,203	10.1%
Mistral AI	1,162	9.8%
xAI	906	7.6%
Meta AI	754	6.3%
Moonshot AI	479	4.0%
Other	427	3.6%
Zhipu AI	263	2.2%
Snowflake	234	2.0%
Databricks (DBRX)	194	1.6%
Black Forest Labs	120	1.0%
Cohere	112	0.9%
Stability AI	102	0.9%
Voyage AI (MongoDB)	92	0.8%
Cascade	82	0.7%
01.AI	56	0.5%
Luel	41	0.3%
Guide Labs	39	0.3%
Reka	39	0.3%
Minimax	13	0.1%
GLM	4	0.0%

Disruption and security remain areas for improvement. Few respondents reported working with security technology, suggesting AI security ops is still a relatively isolated field—and possibly a good target for anyone looking to add future-forward tools to their toolkit. Even setting aside dedicated security practitioners, 23% of users said their team lost valuable time to security or privacy issues stemming from AI.

Most popular technologies

For all the talk of security work shifting left, only 23% of respondents have used security tools in the past year.

View and share

	Respondents	Percent
I used a programming, scripting, or markup language for any reason	13,697	84.3%
I used a development environment (IDE) to write code or interact with agent-assisted developer tools	12,423	76.5%
I used a database for storage, querying, deployment	11,910	73.3%
I used an LLM (large language model)	11,655	71.8%
I built, maintained or learned about web technologies	10,176	62.7%
I used a knowledge management platform (such as a wiki)	10,074	62.0%
I used a cloud platform or any tools related to cloud technologies	9,659	59.5%
I used a community platform or online public forum to search, comment, post to, or read from for technology-related content	9,062	55.8%
I used an enterprise search platform (a secure way to search for disparate internal content)	3,987	24.6%
I used security software for permissions and compliance operations	3,755	23.1%
None of these	364	2.2%
Other	197	1.2%