How Many Game Studios Use AI? Vivrato's 2026 Survey of 101 Studio Leaders, and What It Can't Tell Indies

How many game studios use AI? The 2026 survey most outlets quote comes from Vivrato. It reports that 97% of 101 senior studio leaders in Europe and North America have used AI at least once, and 94% actively use it in at least one production stage. The same report also gives a median score of 3/10 for how deeply AI is embedded in pipelines. Most headlines leave that number out.

This note covers every figure the secondary coverage reports, the low ones included. It then explains why a survey of 101 studio decision-makers says very little about solo and indie developers.

What did Vivrato survey, and what does 97% mean?

Vivrato is a games consultancy. According to This Week in Videogames (TWIV), its "AI Adoption in Videogames" report draws on one-on-one interviews with 101 senior studio leaders and decision makers in Europe and North America. WN Hub says the interviews ran from April to September 2026.

The headline result comes in two parts:

• 97% of respondents have used AI at least once.
• 94% actively use AI in at least one of nine pipeline stages.

TWIV and the Bastion weekly digest, which cites GamesIndustry.biz, both give these figures. The gap between them matters. "Used at least once" covers a single experiment that went nowhere. "Actively use" sets a higher bar, but even that can mean one tool in one department.

The word "AI" is also broad here. TWIV reports that the survey covered six types of AI, including generative AI, agentic AI and classical machine learning. WN Hub says Vivrato itself noted that studios may think about generative AI differently from AI in the wider sense. So "97% use AI" does not mean "97% use generative AI." A studio that runs a machine-learning model on player analytics counts the same as one generating concept art.

A note on sourcing: the original GamesIndustry.biz article could not be opened for this piece. The figures here come from TWIV, WN Hub and the Bastion digest, which cite the report and GamesIndustry.biz.

Which parts of game development use AI the most?

TWIV lists the share of respondents using AI in each of the nine stages:

WN Hub repeats the top three figures: 86% for code, 83% for general tasks and 79% for concepting. It also says analytics and monetisation have the lowest adoption.

The ranking shows a pattern. Internal work leads: code, admin and early concepting. Work that players see, such as final art and narrative, comes lower. One reasonable reading is that studios use AI first where the output stays inside the team and gets reviewed before anyone outside sees it. The reports don't test that explanation, though.

For indie readers, the 86% code figure is the closest match to the workflows this site covers. It is the same territory as the guides in coding agents and the walkthrough on setting up Claude Code for game development. Still, "uses AI for code" in a leadership survey could mean anything from autocomplete in an IDE to agents editing whole projects. The secondary coverage doesn't say which.

How deeply is AI integrated into studio pipelines?

This is the figure that qualifies the headline. TWIV reports that respondents rated how deeply AI is embedded in their pipelines on a scale of 10:

• The median score was 3/10.
• 52% rated their studio at 3 or lower.
• Only 7% rated it 9 or 10.

The Bastion digest puts it more simply: more than half of the 101 leaders said AI isn't deeply embedded in production.

TWIV also gives regional averages. Europe, including the UK, averaged 4.2/10 and North America 3.5/10. With 101 respondents split across two regions, that gap should be read with caution. The coverage doesn't give the number of respondents per region.

Taken together, the data supports "nearly every studio leader has tried AI" far better than "AI runs studio production." A median of 3 suggests AI is common at the edges of the workflow and uncommon at its core.

These are also self-ratings by leaders, not audits of pipelines. A studio head's sense of "deeply embedded" could differ from what an engineer or artist on the same team would report. Nothing in the coverage says Vivrato checked the scores against actual tooling or output.

Does the Vivrato survey include indie developers?

Not in any way the coverage reports. Several features of the sample limit what it says about solo and small-team developers.

Who was asked

The 101 respondents are senior studio leaders and decision makers. That describes people who set policy at studios, not people working alone. None of the three secondary sources gives studio sizes, and none gives a breakdown for solo developers or very small teams. A two-person studio might be in the sample, or it might not. The coverage doesn't say.

Where they are

The sample covers only Europe and North America. Asia and other regions are not included. TWIV's own article mentions large-company examples, including Capcom reworking its RE Engine for AI tools and an Epic launcher described as vibe-coded. Those companies sit outside the survey's sample, which shows how much of the industry the 101 interviews leave out.

Likely sources of bias

The following points are analysis, not findings from the sources:

• Who ran it. It is a consultancy's own study. A firm that advises studios on AI has a stake in how the topic is framed. That doesn't make the numbers wrong, but it is worth checking the method.
• Sample size. 101 respondents is a small base, especially once it is split by region or stage. One respondent is roughly one percentage point.
• Self-reporting. Use and depth were reported by the respondents, not observed.
• Leaders, not makers. Senior leaders may describe AI use differently from the people who write the code, draw the art or run QA.

Why studio adoption and indie adoption differ

A studio leader needs a policy before staff can use AI: legal review, platform rules, disclosure, tool budgets. A solo developer makes that choice alone and can make it in an afternoon. So studio adoption rates are a poor guide to how often indies use AI, in either direction.

The survey also, as reported, says little about the questions indies tend to care about most: output quality, real costs, time saved, or how players and storefronts react to AI-assisted games.

Verdict: who should use these numbers?

Useful for:

• Pitch decks and publisher conversations. A published figure from 101 studio leaders gives you something to cite when a partner asks about AI in your pipeline.
• A rough sense of where peers start. Code, general productivity and concepting lead the list, and live ops and monetisation trail it.
• Setting expectations. The 3/10 median is a fair counterweight to claims that the industry already runs on AI.

Less useful for:

• Estimating how many indie or solo developers use AI. The sample wasn't built for that.
• Showing that AI-assisted games sell, review or ship better. The survey doesn't measure outcomes.
• Separating generative AI from other machine learning. The headline figures combine all six types.

Before citing any of these figures, read the method and sample details in the original report. TWIV links to it at galaxy.vivrato.com/vaia. If you quote the 97%, add the 3/10 median beside it so the two numbers give an honest picture together.

For what the code-stage tooling looks like in practice, see the coverage of Claude Code Game Studios' 49 agents and 74 skills and Unity's official Claude Code plugin. More industry updates are in Lab Notes.