How Long Does It Take to Make a Video Game? (2026 Data)
Game development takes 48 hours (game jam) to 7+ years (AAA). Real 2026 timelines by genre and team — plus how AI compresses the schedule."
Jabali Team
Aug 7, 2026
8 min read
How Long Does It Take to Make a Video Game? (2026 Benchmarks)
Quick answer: A video game takes anywhere from 48 hours to 7+ years depending on scope. A game jam entry: one weekend. A small solo 2D game: 3–12 months part-time. A commercial indie title: 1–3 years. AA: 2–4 years. AAA: 3–7+ years. The schedule is dominated less by "building" than by iteration — finding the fun, cutting what doesn't work, and polishing — which is why AI production tools that collapse build time change the calendar more than any engine improvement ever has.
The timeline ladder: real benchmarks by scope
| Game scope | Typical timeline | Calendar driver |
|---|---|---|
| Game jam entry | 48–72 hours | Fixed window; scope discipline is the whole game |
| Prototype (one core mechanic) | 1 day – 4 weeks | Enough to answer "is this fun?" |
| Small solo 2D game (part-time) | 3–12 months | Evenings/weekends; one mechanic, one loop |
| Solo commercial indie (full-time) | 1–3 years | Balatro: ~2 years, one developer |
| Small team indie (2–10 people) | 1.5–3 years | Most Steam indie releases |
| AA production | 2–4 years | 10–75 people, publisher milestones |
| AAA production | 3–7+ years | 100–1,000+ people across multiple studios |
| Live-service game | 2–5 years to launch, then forever | Launch is the midpoint, not the end |
Solo data points for calibration: Balatro took LocalThunk just over two years (hobbyist pace, with pauses) before selling 5 million copies in under a year. Stardew Valley took Eric Barone roughly 4.5 years of solo work — 41 million copies later, it's the definitive proof that timeline ≠ team size. What those timelines bought wasn't code; it was iteration.
The phases every game goes through (and where time really goes)
Whatever the scope, the shape of production is the same:
| Phase | Share of schedule | What happens | Where teams lose time |
|---|---|---|---|
| Ideation & pre-production | 5–15% | Concept, core loop definition, riskiest-question prototypes | Skipping it entirely ("we'll find it as we go") |
| Prototype | 10–20% | Ugly, fast proof that the core mechanic is fun | Polishing the prototype before it's proven |
| Production | 40–60% | All content and systems built | Scope creep — the schedule's #1 killer |
| Alpha/beta & iteration | 15–30% | Playtesting, balancing, cutting | Not cutting — keeping features that don't earn their place |
| Polish & launch prep | 10–20% | Bug fixing, onboarding, performance, store setup | Underestimating certification, store pages, trailers |
| Post-launch | Ongoing | Patches, community, content | Planning to "figure it out later" |
Design insight: Ask any shipped designer where the time went and you'll get the same answer: not building — deciding. Is this jump fun? Is this economy fair? Should this feature exist at all? Games are found through iteration loops, and every loop has the same cost formula: time to build the change + time to judge the change. Traditional tools make building the bottleneck. When AI collapses build time, the limiting factor becomes how fast you can judge — which is a much better problem to have, because judgment is the skill that makes designers great.
What actually determines your timeline
Five variables decide the calendar more than any other:
- Scope (the multiplier). Content volume — levels, enemies, systems — scales time almost linearly. A 10-level puzzle game and a 100-level puzzle game are the same project with a zero moved. Cut content before you cut corners.
- Genre complexity. A visual novel and an online multiplayer shooter are different species. Networking alone can double a schedule; branching narrative multiplies writing and QA load.
- Team size and coordination. Nine women can't make a baby in a month: adding people past ~10 adds communication overhead. This is why 40% of the industry now works solo or in micro-teams of 2–10 (GDC 2025).
- Tooling and pipeline. Custom engine work can consume 30–50% of a schedule. Mature engines reduce it; AI production tools (Jabali Studio's Bali, self-healing builds, in-pipeline asset generation) attack it from the other direction by removing implementation work altogether.
- Iteration speed. Hours per playable change is the metric that predicts your finish date. Fast-iterating teams find fun early and cut confidently; slow teams find it late or never.
Tips from producers who ship on time:
- Timebox the prototype. Two weeks maximum to prove the core loop is fun. If it's not fun by then, it won't be fun in two months — pivot or cut.
- Playtest from week one. Every week without a real player is a week of decisions made on guesses.
- Plan to cut 30%. Every experienced team cuts a third of planned features. Plan which third in advance so cutting is a strategy, not a crisis.
- Finish something small first. Your estimate for game #2 is always better after you've shipped game #1 — even a tiny one.
- Watch the tail. The last 10% (onboarding, performance, store assets, age ratings) routinely takes 25% of the schedule. Book it early.
How long does it take to make a game with AI?
Short answer: Playable prototype in minutes; a polished small game in days to weeks of iteration rather than months. The schedule doesn't shrink to zero — it shifts almost entirely to the "judge and refine" phases.
With an AI producer building the implementation, the phase table changes shape:
| Phase | Traditional small indie | With Jabali Studio |
|---|---|---|
| Prototype | 2–6 weeks | Under an hour — describe the game, play it |
| Core iteration (finding the fun) | 3–9 months | Days–weeks — each change is a conversation with Bali |
| Content production | 3–12 months | Days–weeks — assets and levels generated in-pipeline |
| Bug fixing & breakage | 20–40% of total | Largely absorbed by self-healing builds |
| Polish & launch prep | 1–3 months | Weeks — onboarding and pacing remain human craft |
| Total, small 2D game | 1–3 years | Days to a few weeks of directed iteration |
What doesn't compress: playtesting with real humans, taste decisions, difficulty tuning, and marketing lead time (Steam wishlists still need months to accumulate — start your store page the day your game is fun, not the day it's done).
The strategic implication: when production time approaches zero, the number of attempts becomes your advantage. A developer who can test ten game ideas in a month will out-design one who can test one per year — every time.
Frequently Asked Questions
Q: How long does it take to make a simple 2D game? A: Traditionally, 3–12 months part-time for a solo developer, depending on scope and experience. With an AI game maker, the same scope compresses to days or weeks of directed iteration, because implementation time approaches zero while playtesting and tuning remain.
Q: Can you make a game in 48 hours? A: Yes — game jams like Ludum Dare and GMTK Jam are built around exactly that constraint, and many successful commercial games started as jam entries. The trick is radical scope discipline: one mechanic, one level, one clear loop. AI tools are increasingly the jam advantage — prototyping in minutes leaves more of the weekend for polish.
Q: How long did famous indie games take to make? A: Balatro: ~2 years, one developer. Stardew Valley: ~4.5 years, one developer. Most commercial indie titles: 1–3 years with small teams. Timelines track scope and iteration count far more than team size.
Q: Why do AAA games take 5+ years? A: Content volume and coordination. Hundreds of people producing thousands of assets, cinematics, and systems — plus the organizational overhead of keeping them coherent. AAA timelines are a management problem as much as a technical one, which is why they don't translate to indie scope.
Q: What's the fastest way to make a game as a beginner? A: Pick the smallest game you genuinely want to play — one mechanic, one screen — and use tools that don't make you wait to play your own ideas. An AI game maker like Jabali Studio produces a playable version of your concept in minutes, so your learning loop starts on day one instead of month three.
Sources
- GDC, State of the Game Industry 2025: 40% of developers work solo or in teams of 2–10; industry timeline norms.
- PlayStack / developer AMAs and press: Balatro — "just over 2 years" of solo development; 5M copies in under a year (Jan 2025).
- ConcernedApe interviews and press: Stardew Valley — ~4.5 years of solo development; 41M copies sold as of Dec 2024.
- Newzoo, Global Games Market Report (June 2026): market context for launch-window planning.