Buying RAM
How much RAM do you need for programming and software development?
For most programming in 2026, 32GB of RAM is the right amount. You can get by with 16GB for web development, Python or scripting in a light editor like VS Code, but Docker, Android emulators, big IDEs and large C++ or game projects quickly outgrow it. Go to 64GB if you run several virtual machines, a local Kubernetes cluster, Unreal Engine projects or local AI models alongside your tools.
What actually uses RAM when you code?
The code itself is tiny. What fills your memory is everything running around it:
- Your editor or IDE. VS Code is light. Full IDEs like IntelliJ IDEA, Android Studio or Visual Studio index your whole project so they can autocomplete and refactor, and that index lives in RAM.
- Language servers and linters. The background processes that check your code as you type. Big TypeScript, Rust or C++ projects make these hungry.
- Containers and virtual machines. Docker on Windows and macOS runs inside a small virtual machine that reserves its own slice of memory. Each database, cache or service you start adds more.
- Emulators. An Android emulator is a whole phone running in RAM.
- Builds. Build tools often run one compile job per CPU thread, and each job needs its own memory. On a big C++ codebase, a fast many-core CPU can run out of RAM before it runs out of cores.
- Your browser. Documentation, Stack Overflow, the app you're testing and its dev tools. This is often the biggest single user.
That's why the minimum requirements on tool websites can be misleading. They cover one tool on its own, and you never run just one.
What do the tool makers recommend?
| Tool | Minimum | Recommended |
|---|---|---|
| Visual Studio Code | not stated | 1GB |
| IntelliJ IDEA | 8GB total RAM | not stated |
| Docker Desktop (Windows, WSL 2) | 8GB system RAM | not stated |
| Visual Studio 2026 | 4GB | 16GB for typical professional solutions, "works best" with 64GB |
| Android Studio | 8GB, or 16GB with the emulator | 32GB |
| Unreal Engine 5 | not stated | 32GB |
These come from each vendor's requirements page (the IntelliJ and VS Code figures via SizeMyLaptop's roundup of vendor pages). Two stand out. Google recommends 32GB for Android Studio, and Microsoft says Visual Studio runs best on Windows 11 with 64GB of RAM and a CPU with 16 or more cores. When the people who make the tools suggest that much, take it seriously.
How much RAM for each kind of development?
| What you work on | Workable | Comfortable |
|---|---|---|
| Learning to code, scripts, small Python projects | 8GB | 16GB |
| Web front end in VS Code, a few services | 16GB | 32GB |
| Back end with Docker Compose (database, cache, API) | 16GB | 32GB |
| Java or Kotlin in IntelliJ, .NET in Visual Studio | 16GB | 32GB |
| Android apps with the emulator | 16GB | 32GB |
| Large C++ codebases, game engines, Unreal | 32GB | 64GB |
| Several VMs, local Kubernetes, data science on big datasets | 32GB | 64GB or more |
"Workable" means it runs, but you'll close things to make room. "Comfortable" means you can leave your browser, IDE, containers and chat apps open all day without the system slowing down.
Why do Docker and WSL eat so much memory?
On Windows, Docker Desktop and the Windows Subsystem for Linux (WSL 2) run Linux in a lightweight virtual machine. That VM takes a chunk of your RAM and holds on to some of it even when your containers are idle, so Task Manager can show a process called "Vmmem" or "VmmemWSL" using several gigabytes.
You can cap how much it's allowed to use with a .wslconfig file in your user folder, and Docker Desktop has its own resource settings. That helps on a 16GB machine, but it's a trade-off: give the VM too little and your containers slow down or crash instead.
Docker's own Windows requirements ask for 8GB of system RAM just to run it. Add a database, an IDE and a browser, and 16GB is already tight.
Does RAM speed matter for programming?
Much less than capacity. Faster memory can shave a little off long compile times, but you won't notice the difference between DDR5-5600 and DDR5-6000 while editing code. Running out of memory, on the other hand, makes everything stall while Windows swaps to the SSD.
So if you're choosing between 32GB of ordinary DDR5 and 16GB of fast DDR5, take the 32GB. On a desktop, 2x16GB of DDR5-6000 is a sensible, well-supported default. Our guide on why DDR5-6000 is the sweet spot explains why going faster rarely pays off.
Do make sure you're running two sticks in dual channel. One stick halves your memory bandwidth, which shows up in builds and test runs. Our guide on dual channel covers the details.
Laptop or desktop: does it change the answer?
On a desktop, you can start with 32GB and add more later if your projects grow. Check the board has two free slots, or plan to replace the kit with 2x32GB or 2x48GB, since four DDR5 sticks often have to run slower. See 2 sticks vs 4 sticks of DDR5.
On a laptop, check whether the memory is soldered. Many thin developer laptops use LPDDR5X that can never be upgraded, so buy enough up front. For professional work, that means 32GB at least. Our guide on how much RAM a laptop needs goes through the decision.
What about AI coding assistants?
Cloud-based assistants run on someone else's servers, so they barely touch your RAM. Running a coding model locally is different: the model has to fit in memory, on top of everything else you have open. Small models fit alongside a normal setup on 32GB, but larger ones need 64GB or a graphics card with plenty of its own memory. Our guide on RAM for local AI has the numbers.
Is it worth paying for 32GB with prices this high?
If you write code for a living, yes. RAM is expensive in 2026, but the time you lose to a sluggish machine every day costs more than the difference between 16GB and 32GB. If you're learning to code or work on small projects, 16GB is fine for now, and you can upgrade a desktop later when prices settle. Our guide on whether to buy RAM now or wait can help with the timing.
The short version
- 32GB is the right default for professional development in 2026.
- 16GB is fine for learning, scripts and web work in a light editor.
- Docker, Android emulators, Visual Studio and big C++ projects are what push you past 16GB.
- Go to 64GB for several VMs, Unreal Engine or local AI models.
- Capacity matters far more than speed. Take 32GB of ordinary DDR5 over 16GB of fast memory.
Sources
- Android Developers, Install Android Studio (system requirements)
- Docker Docs, Install Docker Desktop on Windows (updated Nov 19, 2026)
- Microsoft Learn, Visual Studio 2026 system requirements (updated Apr 1, 2026)
- Epic Games, Hardware and software specifications for Unreal Engine
- SizeMyLaptop, RAM requirements of pro apps, from the vendors' own pages (read Nov 2026)