Problems · Before you buy
What should I check before buying a laptop for programming?
A programming laptop is chosen with the hands and eyes first: you will spend thousands of hours on this keyboard and screen, and no core count compensates for hating either. After those, the technical checks: RAM for your actual stack, sustained performance if your builds are long, and the ports and docking of a desk-centred life.
The return window checks: type on it for real days, run your real project's build, open your real environment. Synthetic anything tells you less than one honest workday.
First, notice when it happens
The causes, ranked
1. Keyboard and screen: the thousand-hour surfaces
most commonHow to confirm it
Type real code for an hour in the window: key travel, layout quirks (where did they put the arrows), flex, and a screen tall enough for code. 16:10 height and comfortable text rendering at your working scale matter daily.
The fix
If either annoys you in week one, no spec redeems it by year two. This is the least glamorous check and the one most owners wish they had weighted.
2. RAM: count your real stack, not a rule of thumb
most commonHow to confirm it
Open your actual daily set: editor, containers, local services, browsers, a call. Read the memory pressure. Web stacks with containers routinely clear 16GB; heavy IDEs and VMs clear more.
The fix
16GB is the floor for light stacks, 32GB is the honest default for container-era work, and soldered RAM makes overshooting the only revisitable decision. Buy for the stack you are growing into.
3. Sustained performance, scaled to your build times
commonHow to confirm it
A ten-second incremental build cares about single-core snap. A ten-minute clean build or test suite is a sustained load, where thin machines settle and build times stretch.
The fix
Match the spend: comfort-first machines for short-cycle work, sustained evidence (a fan and a flat curve) where the compiler is your longest meeting. Run a sustained test on the candidate in the window; your builds live on that curve.
4. The OS and toolchain fit
commonHow to confirm it
Your deploy target and toolchain choose more than preference does: container-native work, Unix tooling, platform-specific builds. WSL has closed much of the gap on Windows; verify YOUR tools, not the discourse.
The fix
Install your real environment in the return window. An afternoon of setup friction now predicts a year of it.
5. Desk life: ports, docking, externals
commonHow to confirm it
Two external monitors is the common programmer desk: verify the machine actually drives your pair at full resolution through your dock or ports, a spec sheets bury in footnotes.
The fix
Check the display-out limits for the exact model before buying, and test the dock in the window. Monitor disappointment is the most common programmer return reason nobody warns about.
6. Battery and noise, weighted by where you work
occasionalHow to confirm it
Cafe-and-train workers need the all-day battery and silence that desk-dwellers can trade away for sustained speed.
The fix
Be honest about your actual week, then pick the trade deliberately. The quiet efficient machine and the sustained workstation are both right answers to different weeks.
Measure it instead of guessing
The candidate's sustained curve is your future build time in one picture: flat means the ten-minute build stays ten minutes on the hottest afternoon; a slide means it grows. Run it in the window, next to your real project's clean build, and the purchase decision becomes two numbers instead of a feeling.
What software cannot tell you here
- We cannot run your build or your containers; one real workday in the return window is the test that matters most.
- Keyboard feel and screen comfort are yours alone to judge, and they outrank everything we can measure.
Asked, in people’s own words
- Is 16GB enough for programming in 2026?
- For scripting, study and light web work, yes. Add containers, a heavyweight IDE, local databases and modern browser tab counts, and 32GB stops being luxury. If the RAM is soldered, buy the stack you are growing into, not the one you have today.
- Do programmers need a GPU?
- For most software work, no: integrated graphics drive the monitors and the money belongs in RAM and the screen. The exceptions know who they are: game development, ML work and GPU compute, which move you to the AI and gaming spec guides.
- MacBook or Windows or Linux laptop for coding?
- Deploy target and toolchain first: match what you ship to. Beyond that it is keyboard, screen and RAM per unit of money. All three platforms carry professional toolchains now; the wrong choice is buying the platform argument instead of the machine in front of you.