S11.5 · Media, Entertainment & Creative
A ~$189-197B games market, mobile-led, where AI is compressing QA, localization and art-production cost for outsourced dev-services shops.
The global games market runs at roughly $188.8B in 2025 by one widely used estimate (Newzoo), with other trackers citing figures near $197B — a spread that reflects differing scope definitions, not genuine disagreement about direction. Mobile is roughly 55% of revenue and remains the primary growth driver. Esports, often treated as the category's bellwether, is neither: a small niche of only $1-2B whose growth has decelerated since its earlier hype cycle, and a poor proxy for where the money in this market actually sits.
APAC — led by China, Japan and Korea — is the largest revenue region, while development talent is more globally distributed, with Canada, Poland and India growing as production hubs. Tencent, Sony, Microsoft/Activision Blizzard, Electronic Arts, Take-Two and HoYoverse capture a large share of the market, and a long tail of independent developers reaches players through digital storefronts. Engines such as Unity and Unreal form the infrastructure layer, with developers and publishers above them and platforms and esports organizations downstream.
Three economic models share the segment and should not be blended. Live-service franchises behave like rights businesses: high-margin recurring revenue built on an installed player base. Work-for-hire development studios run on project and services economics. And AAA game development is capital-intensive regardless of eventual commercial outcome — development spend on a major title is committed years before launch, with no guarantee the installed-base economics of a live-service hit will ever materialize. That capital intensity is a risk factor in its own right, distinct from whatever the underlying market growth rate does.
Publishers pushing franchises into film and television adaptation, and streaming platforms pushing into cloud gaming, are running the same play from opposite ends: diversification away from dependence on any single title's commercial performance.
AI's cost compression is specific — quality-assurance testing, localization and art-asset production. The moat under pressure is the labor-cost advantage that large art and QA teams have historically provided, and the damage is unevenly distributed. It hits outsourced development-services shops — the studios doing work-for-hire art, localization and testing for publishers — harder than the publishers themselves, because publishers retain the IP and the player relationship regardless of how the underlying assets are produced. The services layer absorbs the displacement; the rights layer captures the savings. A new tier of AI-native procedural-content studios, built around radically smaller headcount than traditional development teams, is emerging as a distinct competitive class that can plausibly compete on production values that once required a large outsourced studio. The compression in QA, localization and art production is underway now; the AI-native studio model as a durable competitive category is expected within 2-5 years.