Gaming, virtual worlds & immersive media

Gaming, virtual worlds & immersive media

Industry ID: 15 | Slug: gaming-immersive | Researched: 2026-09-15 | Analyst: agent

Methodological note on this sector's vocabulary. This industry mixes bookings, revenue, players, MAU and DAU more casually than any other in the programme, and the substitution is frequently deliberate. Throughout this dossier: bookings = cash received in the period; revenue = bookings recognised over an estimated user life, and therefore a lagging indicator; DAU/MAU = engagement, not customers; players = a survey-derived population estimate, not accounts. Roblox's Q2 2026 — revenue +36%, bookings +8%, next-quarter bookings guided to −14% to −18% — is the canonical demonstration and is treated as such in §8 and trend T-15-17.

1. Definition and boundaries

In scope. Console, PC and mobile game development and publishing; platform holders and their storefronts (PlayStation, Xbox, Nintendo, Steam, App Store, Google Play); game engines and middleware (Unreal, Unity, Godot, Havok, Wwise); user-generated-content platforms with internal economies (Roblox, Fortnite Creative/UEFN, Minecraft); live-service operations and their monetisation; esports and competitive infrastructure; XR hardware and the content made for it; virtual worlds; game AI, both the classical discipline and the generative wave; and platform/app-store economics as they apply to games, which in practice means most of the app-store economy, since games are the majority of mobile consumer spend.

Explicitly out of scope, and who owns it.

The boundary disputes are real and worth naming.

1. Is Roblox a game company or a social platform? It reports like a games company (bookings, DAU, hours engaged) and is regulated increasingly like a social network (age assurance, state attorney-general attention, California AB 1709). The answer determines which comparables and which rulebook apply, and it is genuinely unsettled.
2. Is app-store regulation a gaming story? It is adjudicated in cases brought by a games company (Epic) over a game (Fortnite), and games are the largest single category of affected commerce — but the remedies apply to all apps. This dossier covers the games incidence and leaves the general case to sector 02.
3. Is XR one market or three? Consumer VR gaming, enterprise XR and smart glasses have almost nothing in common technically or commercially, yet are consistently reported as one number. Meta's Reality Labs segment bundles all three, which is why segment revenue growth of +16.5% coexists with VR studios closing (see §10 and T-15-16).
4. Generative AI: sector or input? Treated here as an input to games production. The model layer belongs to sector 01; what belongs here is the evidence of adoption in shipped games, which Steam's disclosure field uniquely provides.

2. Subcategories

1. AAA premium console/PC publishing — $100m+ budgets, multi-year cycles, increasingly dependent on a small number of tentpoles. Take-Two, Rockstar, Sony first-party, Nintendo.
2. Live-service operations — persistent titles monetised by season passes, battle passes and cosmetics. The economic centre of gravity; Take-Two's recurrent consumer spending was 84% of net bookings in the quarter ended 2026-06-30.
3. Mobile free-to-play — the largest segment by revenue ($121.1bn of Newzoo's $213.9bn 2026 model) and the one most directly re-priced by the app-store changes of 2025–26.
4. UGC platforms — platforms whose content is made by users and whose economics are a closed virtual currency plus a cash-out rate. Roblox, Fortnite Creative/UEFN, Minecraft.
5. Independent PC development — the long tail on Steam, where the median successful title now prices at $16.99 and roguelites and life-sims have displaced the 2021 genre mix.
6. Engines and middleware — Unreal and Unity duopoly, plus Godot's growing open-source share and an expanding audio/physics/analytics tool layer.
7. Platform storefronts and distribution — the 20–30% toll layer, currently the most volatile part of the value chain.
8. Esports and competitive infrastructure — tournament operators, leagues, teams, and an increasingly betting-financed revenue base.
9. XR hardware and immersive content — Quest, Vision Pro, Steam Frame, Galaxy XR and the contracting studio base that supplies them.
10. Game AI and production tooling — from classical pathfinding and behaviour trees to generative asset, audio and localisation pipelines and, newly, agentic engine assistants.
11. Games-as-advertising-inventory — in-game ads, brand activations and playable ads; growing but still small relative to IAP.
12. Regional publishing and licensing — China's banhao regime, Korea's publisher ecosystem, Japan's mobile duopoly. Structurally distinct businesses, not just geographies.

3. Market structure

Concentration: winner-take-most at the platform layer, oligopoly at the publisher layer, extreme fragmentation at the developer layer. This three-tier asymmetry is the single most important structural fact about the sector and it explains almost every conflict in it.

• Platform layer (2–4 firms per surface). Apple and Google control mobile distribution; Valve dominates PC; Sony, Microsoft and Nintendo control console. These firms do not compete for the same developer, so each has near-monopoly pricing power within its surface. This is why the commission question had to be settled by courts and legislatures rather than by competition.
• Publisher layer (oligopoly). Tencent, Sony, Microsoft, NetEase, EA, Take-Two, Nintendo, Krafton, Ubisoft and a handful of others. Tencent Q2 2026 domestic games revenue alone was RMB47.3bn (+17% YoY).
• Developer layer (fragmented to the point of atomisation). Tens of thousands of studios; 311 games cleared $1m gross Steam revenue in 2025, out of many thousands released.

Where margin actually sits. Not where the headlines suggest.

The commission changes of 2025–26 move roughly 4–10 percentage points of gross from the first row to the rest of the table, with much more available to anyone who can shift a buyer to a web checkout. That is the largest single value transfer in the sector's recent history.

Barriers to entry. Low to make a game, extreme to be discovered. Discovery, not production, is the binding constraint: engine and asset costs have collapsed while the attention available to a new title has not grown. A second barrier has appeared on top: compliance. Age assurance, per-jurisdiction fee agreements, AI disclosure and child-safety architecture are fixed costs that scale poorly for small studios, so regulation intended to constrain the platforms is simultaneously raising the barrier below them.

Who has pricing power. Platform holders retain it, but for the first time it is externally capped. Evergreen live-service operators have it over their own players. Nobody else does: the median price of a top-selling Steam game fell 15% from $19.99 (2021) to $16.99 (2025), while console hardware prices rose (Sony, 2026-04-02: PS5 Digital +$50, PS5 Pro +$150).

Market size — with the modeller named, and treated as an estimate.

• Newzoo models the 2026 global games market at $213.9bn (+6.1% YoY): mobile $121.1bn (+6.8%), console $46.9bn (+5.1%), PC $45.9bn (+5.3%); 3.7bn players; APAC $100.7bn (47%), North America $56.9bn, Europe $38.5bn, LatAm $9.3bn, MEA $8.5bn. Published 2026-08-25 (full report 2026-09-10). Estimate, not a reported aggregate. Note Newzoo's own caveat that console growth is contingent on a single November release.
• ESA/Circana/Sensor Tower report actual US consumer spending of $60.7bn in 2025 (+1.4%), revised to $60.8bn in March 2026: content $52.3bn, hardware $5.4bn (+9%), accessories $2.95bn (−7%), subscriptions +20%, mobile content $26.7bn (+1%). This is a measured figure, not a model, and is the more citable of the two.
• China passed $50bn for the first time in 2026 (estimate, via trade press citing market modellers) — roughly a quarter of the global market in one country.

4. Who matters

Leading companies. Tencent (https://www.tencent.com) · Sony Interactive Entertainment (https://www.sie.com) · Microsoft Gaming (https://www.xbox.com) · Nintendo (https://www.nintendo.co.jp) · NetEase (https://www.netease.com) · Electronic Arts (https://www.ea.com, private since 2026-08-04) · Take-Two Interactive (https://www.take2games.com) · Krafton (https://www.krafton.com) · Ubisoft (https://www.ubisoft.com) · Roblox (https://corp.roblox.com) · Epic Games (https://www.epicgames.com) · Valve (https://www.valvesoftware.com) · Embracer Group (https://embracer.com) · miHoYo/HoYoverse (https://www.hoyoverse.com)

Notable startups and smaller studios of consequence. The honest position is that this sector currently has very few venture-funded startups of scale — the 2022–23 funding wave into "metaverse" and web3 gaming collapsed and was not replaced. What matters instead are independent studios with outsized commercial results (the 2025 Steam cohort: 311 titles above $1m gross) and the casualties that tell you where the market is going: Polyarc (https://polyarcgames.com, shut down September 2026), Vertigo Games Amsterdam (closed by Embracer 2026-06-09), Schell Games (layoffs 2026-08-26), Intrepid Studios, Wildlight Entertainment, Strikerz, KO_OP, Star Stable Entertainment (−20% headcount 2026-09-07).

Active investors. Public Investment Fund (https://www.pif.gov.sa) and its Savvy Games Group — now the controlling shareholder of EA and the dominant esports financier · Silver Lake (https://www.silverlake.com) · Affinity Partners · Tencent (as strategic investor across Epic, Ubisoft subsidiary, Riot, Supercell) · Sony (minority strategic stakes) · a much diminished conventional venture presence.

Platforms and standards bodies. Apple App Store · Google Play · Steam · PlayStation Network · Xbox/Microsoft Store · Nintendo eShop · Khronos Group (OpenXR, Vulkan, https://www.khronos.org) · Metaverse Standards Forum (https://metaverse-standards.org) · IGDA (https://igda.org).

Regulators. European Commission DG COMP/DG CNECT, Digital Markets Act (https://digital-markets-act.ec.europa.eu) · Japan Fair Trade Commission, Mobile Software Competition Act (https://www.jftc.go.jp/en/) · China's National Press and Publication Administration (http://www.nppa.gov.cn) · Korea Fair Trade Commission and KCC · US District Court, N.D. Cal. (Epic v Apple, Epic v Google) · US FTC (children's privacy, dark patterns) · UK CMA and Ofcom (Online Safety Act) · Brazil's CADE · state attorneys-general in the US on child safety · PEGI (https://pegi.info) and ESRB (https://www.esrb.org) for content rating.

Research institutions. Niko Partners (https://nikopartners.com) · GameDiscoverCo (https://gamediscover.co) · Circana (https://www.circana.com) · Newzoo (https://newzoo.com) · Sensor Tower (https://sensortower.com) · academic centres including USC Games, NYU Game Center, Abertay University, University of Utah's EAE programme, and the DiGRA research network (http://www.digra.org).

Trade organisations. ESA (https://www.theesa.com) · ISFE (https://www.isfe.eu) · Ukie (https://ukie.org.uk) · game — Verband der deutschen Games-Branche (https://www.game.de) · SELL (France) · CESA (Japan, https://www.cesa.or.jp) · K-GAMES (Korea) · GDAC.

Consumer and civil-society groups, and labour. Communications Workers of America (https://cwa-union.org), which holds the Blizzard and ZeniMax bargaining units · SAG-AFTRA (https://www.sagaftra.org), whose Interactive Media Agreement governs AI use of performers · Game Workers Unite / IWGB in the UK · Fairplay and Common Sense Media on child safety · 5Rights Foundation on age-appropriate design.

5. Products, business models, technologies, customers

Major products. Platform hardware (Switch 2 at 23.68m units as of 2026-06-30; PS5 at ~93m as of 2026-03-30; Xbox Series X|S; Steam Deck; Quest 3/3S; Vision Pro; Steam Frame from $1,059). Software: the tentpole premium release (Grand Theft Auto VI, 2026-11-19), the annual sports franchise (EA SPORTS FC 27, Madden NFL 27, NBA 2K27), the evergreen live-service title (Fortnite, Delta Force, VALORANT, Honor of Kings, PUBG), and the sub-$20 independent hit that now defines the modal Steam success. Services: subscriptions, storefronts, engines, ad networks and creator-economy tooling.

How money is actually made today.

1. In-app purchases in free-to-play, still the largest pool. Mobile IAP is $121.1bn of Newzoo's 2026 model.
2. Recurrent consumer spending in premium titles — battle passes, cosmetics, currency. Now dominant even at premium publishers: 84% of Take-Two's Q1 FY2027 net bookings.
3. Premium unit sales, shrinking in share and, on PC, in price.
4. Subscriptions, which grew 20% in the US in 2025 but where the largest service is contracting (Xbox content and services −10% YoY in the quarter ended 2026-06-30).
5. Platform commission, the toll layer, now externally re-priced.
6. Engine royalties and seat licences.
7. Advertising, in-game and playable, growing from a small base.
8. UGC creator payouts as a cost line, and the platform's retained spread as the business.

How it is changing. Three simultaneous shifts. First, the toll layer is being re-priced downward by force of law across every major jurisdiction, which for the first time makes direct-to-consumer web checkout economically rational for mobile publishers — a 15–21 percentage-point spread between in-app (20–26%) and alternative distribution (5% Core Technology Commission) in the EU and Japan. Second, the subscription thesis has been tested and partly abandoned by its originator. Third, production cost structure is being altered at the asset layer by generative AI, which is now in shipped pipelines but concentrated in art, audio and localisation rather than in design.

Technologies that matter. Unreal Engine 5 and Unity as the production substrate; Vulkan, DirectX 12 and Metal at the graphics layer; upscaling and frame generation (DLSS, FSR, XeSS) which have become load-bearing for hitting frame targets; anti-cheat (a real and growing cost — 51% of developers report player and revenue losses to cheating, 69% call it significant); backend live-ops and telemetry platforms; generative models for 2D art, audio, voice and localisation; agentic coding assistants now shipping as first-party engine plugins (Unity's Claude Code plugin with 29 engine skills, 2026-09-10); age-assurance and facial age-estimation stacks, newly a core platform component rather than a compliance bolt-on.

Customer segments and what they buy on. Core console/PC players buy on IP, review consensus and social proof, and are increasingly price-sensitive at the top end. Mobile free-to-play payers are a small fraction of installs producing most revenue; they are bought, not attracted, which is why user-acquisition cost is the sector's real unit-economics driver. Youth UGC audiences do not buy games at all — they buy currency inside a platform, which is why age assurance hits monetisation directly. Subscribers buy optionality and breadth. Creators are a two-sided-market supply input who "buy" a platform on its cash-out rate, which is the number platforms are least keen to lead with.

6. Geography

Production. Historically concentrated in the US West Coast, Montreal, the UK, Sweden, Poland, Japan, Korea and China. Four years of contraction has been redistributing it: the 2026 closures cluster in exactly the high-cost metros (Ubisoft Winnipeg and Belgrade, TiMi Montreal, Vertigo Games Amsterdam, Bluepoint in Texas, Polyarc in Seattle), while Central and Eastern Europe, Southeast Asia and Latin America absorb co-development and outsourcing.

Capital. Increasingly Gulf sovereign rather than Western venture. PIF now controls EA outright (completed 2026-08-04) and, via Savvy Games Group, is the dominant financier of esports. Conventional games venture capital has not recovered from the 2022–23 collapse, and per the macro brief's capital-formation framing, what venture capital exists is being drawn overwhelmingly to the AI model layer.

Demand. Asia-Pacific is 47% of the global market ($100.7bn on Newzoo's 2026 model). China alone passed $50bn. North America is $56.9bn and Europe $38.5bn. MEA is the fastest growing (+10.3%) from a small base, which is not unrelated to where the capital is coming from.

Regulation. The most fragmented dimension of all, and now the most consequential. The EU (DMA), Japan (MSCA), the US (courts, plus state child-safety statutes), China (NPPA licensing, minors' playtime limits), Korea (in-app payment law) and Brazil (CADE) each impose a distinct regime, and a mobile publisher must now maintain a different fee agreement, distribution route and compliance posture in each.

Non-US market treated in depth: Japan. Japan is simultaneously a major demand market and the source of one of the two most consequential regulatory changes of the period. Adjust and Sensor Tower size Japanese mobile game IAP at $10.6bn, on 2.65bn annual app downloads and 114bn hours of app engagement (2025 base), with iOS holding an unusually high 67–68% share and 66% of game sessions — the highest iOS share of any major market, which is precisely why the Mobile Software Competition Act targets Apple most directly. H1 2026 growth was modest (app installs +4%, game installs +3%, game sessions +5%). Day-1 retention improved to 26% from 24% a year earlier and cost per install fell across most categories. Under the MSCA, from iOS 26.2 (2025-12-17) developers can distribute through alternative marketplaces and process payments outside Apple IAP in Japan; the resulting schedule is 21% commission plus a separate 5% Apple payment-processing fee (26% total for standard IAP), 10% for Small Business/Mini Apps/Video Partner participants and year-2+ subscriptions, 15%/10% on out-of-app links with a 7-day attribution window, and a 5% Core Technology Commission on alternative distribution. All developers had to accept the updated terms by 2026-03-17. Regional sources used: Nintendo's own Japanese-language IR disclosure, Apple's Japan distribution support page, the JFTC's MSCA materials, and the Adjust/Sensor Tower Japan market report. Sony's Japanese IR PDFs were not retrievable in this research pass (see §13).

7. Historical trend patterns

Over 25 years this sector has run an unusually legible cycle: a platform transition creates a land grab, a business-model innovation is over-extrapolated into a universal theory, capital floods the theory, and roughly 70% of it is destroyed. Knowing the specific past false positives is what makes the present readable.

Documented false positives in THIS sector:

• Motion control as the future of interaction (2006–2012). Wii sold 101m units and Kinect sold 24m in its first year. Both were treated as permanent shifts in how games are played. Neither produced a durable software ecosystem. Lesson: hardware unit sales are not platform adoption. This is directly applicable to XR today.
• Plastic instrument music games (2005–2010). Guitar Hero and Rock Band went from zero to $1.7bn annually and back to near-zero in five years. Lesson: accessory-dependent categories have brutal saturation dynamics.
• 3D television and 3D gaming (2010–2013). Near-universal industry commitment, zero durable adoption.
• Facebook/social games (2009–2013). Zynga's IPO was the peak; the platform changed its distribution rules and the category was destroyed in roughly 18 months. Lesson: a business built entirely on someone else's distribution surface has no terminal value. This is the single most relevant precedent for UGC-platform creators today.
• The first VR wave (2016–2018). Oculus Rift, HTC Vive and PSVR launched into predictions of tens of millions of units. Consumer VR did not reach them, and most first-wave studios closed. The second wave (2020–2024) is now resolving the same way — see §10.
• Blockchain/NFT gaming (2021–2023). Billions in disclosed funding, several billion-dollar valuations, essentially zero durable players. Axie Infinity's collapse was the marker. Lesson: funding volume and token price are not adoption. The contract's rule that search, funding and media volume cap out at velocity exists because of cases like this.
• The metaverse (2021–2023). See T-15-19. Still consuming capital at roughly $8.6bn of operating losses per half-year at Meta alone.
• Cloud gaming as a console replacement (2019–2022). Google Stadia is the cleanest failure case; the technology worked and the market did not exist at the price offered.

What actually cycled and persisted:

• The price of a premium game has been remarkably sticky in nominal terms ($50 → $60 → $70 across three decades) while the modal successful price has fallen — the 2025 Steam top ten contained three titles under $10.
• Platform commission at 30% persisted from the 1980s cartridge era through to 2021 and has now broken — the first structural change in four decades.
• Live-service monetisation has survived a full hype cycle, a downturn and sustained consumer hostility. It is the rare trend in this sector that is genuinely durable.
• Consolidation waves recur roughly every seven years (2007–08, 2014–15, 2021–23) and are followed by divestment and write-downs. Embracer is the current worked example.
• Labour cycles have amplitude: the 2020–22 hiring boom produced the 2023–26 contraction almost mechanically.

8. What is changing now (as of 2026-09-15)

The 30% commission is gone, everywhere, within nine months. This is the largest structural change in the sector and it happened through five independent mechanisms at once.

The EU also abolished the Core Technology Fee, the Initial Acquisition Fee and the Store Services Fee in favour of a single 5% Core Technology Commission. The unresolved US question — what Apple may charge on external-link purchases — is live before Judge Gonzalez Rogers after the Supreme Court refused Apple a delay on 2026-08-13 and Apple filed its proposal on 2026-08-14. Epic's reading of the Ninth Circuit's "necessary costs" standard is 0%. Both numbers are currently defensible and neither should be reported as settled.

Roblox has passed its engagement peak, and the metric substitution matters enormously. Q2 2026: revenue $1,469m (+36%), bookings $1,557m (+8%), DAU 123m against a 152m peak in Q3 2025, hours engaged 29bn against a 40bn peak, DevEx payouts $363m, net loss narrowed to $185m, free cash flow $294m (+66%). Q3 2026 guidance is bookings of $1,576–1,653m, a 14–18% year-on-year decline — the first in the company's history — with full-year guidance withdrawn in favour of quarterly-only. The 36% revenue growth is deferred bookings from earlier, higher-engagement quarters being recognised. Age-check penetration reached 57% globally and management explicitly cites monetisation pressure in younger cohorts. The safety architecture is working as designed and it is costing bookings.

Creator payouts are ~23% of bookings, not 70%. Roblox pays $363m of $1,557m bookings — 23.3% in cash — while documenting a 70% creator share of in-experience Robux sales. Both are true: the 70% is denominated in Robux and converts out at 10,000 Robux = $38 (~$0.0038) against a retail purchase price near $0.0125. The two-step conversion, not the percentage, is the economics.

Employment is in its fourth contraction year at a lower run rate. The community tracker records ~4,600 affected and 22 studios shut in 2026 through 2026-07-01 — well below the 2023–24 peak, but now hitting first-party platform studios (Bluepoint closed by Sony, TiMi Montreal by Tencent) and Epic itself (~1,000 on 2026-03-24). Closures continued after the tracker's cutoff (Polyarc, KO_OP, Star Stable). Two things are new in 2026: the first ratified Blizzard union contract covering AI, layoffs and remote work (2026-09-10), and evidence that substitution is moving upstream to the hiring decision — a Talent Signals survey reported 35% of games founders declining to hire because AI can do the work, which no layoff tracker can see.

Generative AI is in production, but shallowly. Steam's disclosure field — the sector's only hard denominator — showed 1,163 of 4,382 registered Next Fest demos disclosing AI use in June 2026 (26.5%, up from ~17% in October 2025); roughly 60% of disclosures concern visual assets. Code was exempted from disclosure in January 2026, so the series is not comparable across that boundary and the true rate is higher than disclosed. Named shipped uses are asset-layer: EA's generative commentator voice in NHL 27, Level-5's AI reveal material (for which its CEO apologised). Unity shipped an official agentic coding plugin on 2026-09-10.

Subscriptions reversed. Microsoft raised Game Pass Ultimate ~30% to $29.99 on 2025-10-01, then cut Ultimate and PC pricing on 2026-04-21 and removed day-one Call of Duty from future titles. Xbox content and services revenue fell 10% YoY in the quarter ended 2026-06-30, even as US subscription spending across all services grew 20% in 2025.

China normalised. NPPA approvals: 1,771 in 2025 (+25.1%), 779 through May 2026 (+19% YTD), projected >2,300 for the year. Imported titles remain scarce (25 of 779). The domestic market passed $50bn.

Ownership changed at the top. EA completed its take-private by PIF, Silver Lake and Affinity Partners on 2026-08-04 — the first top-five Western publisher to leave public markets, removing a quarterly disclosure the sector relied on.

Macro linkage. Per the macro brief, this is a sticky-inflation, rising-rate-risk environment. Three specific transmissions are visible here: (i) the EA LBO must service debt at 3.50–3.75% policy rates, unlike the 2021 LBO vintage; (ii) Sony raised console prices on 2026-04-02 citing the "global economic landscape", consistent with tariff and memory-cost pass-through; (iii) games venture funding is structurally starved because, as the brief records, >70% of Q2 2026 venture capital went to AI-focused companies.

9. The five lists

(Overlaps are flagged explicitly. T-15-01 appears on three lists because a global re-pricing of the sector's largest cost line is simultaneously the most important trend, the most business-relevant, and — indirectly, through pricing — consumer-relevant.)

Five most important current trends

1. T-15-01 — The flat 30% app-store commission dismantled in every major jurisdiction
2. T-15-02 — Recurrent consumer spending as the whole business (84% of Take-Two bookings)
3. T-15-17 — Roblox past its engagement peak, with the first guided bookings decline
4. T-15-08 — Generative AI in shipped production pipelines, at the asset layer
5. T-15-03 — A fourth consecutive year of industry job losses

Five fastest-growing signals

1. T-15-09 — Age assurance as platform architecture (57% Roblox penetration in ~8 months)
2. T-15-15 — Agentic coding assistants shipping as first-party engine features
3. T-15-12 — UGC content escaping the app store via standalone apps and 5% CTC routes
4. T-15-10 — Sovereign wealth taking outright control of Western publishers
5. T-15-13 — AI-substituted hiring, invisible to every existing employment tracker

Five trends most likely to affect businesses

1. T-15-01 — Commission re-pricing (also list 1)
2. T-15-02 — Live-service concentration reshaping the greenlight calculus (also list 1)
3. T-15-05 / T-15-18 — Subscription reversal invalidating a decade of strategic models
4. T-15-10 — EA's take-private and the take-private option now priced into every mid-cap
5. T-15-11 — Union contracts with enforceable AI clauses as a new cost of AI deployment

Five trends most likely to affect consumers

1. T-15-09 — Age assurance: identity checks to play, and account tiers by age
2. T-15-06 — Cheaper games, more expensive hardware
3. T-15-18 — Day-one subscription access narrowing; Call of Duty off Game Pass at launch
4. T-15-17 — Roblox's safety changes altering what under-16s can do and spend
5. T-15-08 — AI-generated content in shipped games, with disclosure but not consent

10. Overhyped / overlooked / cooling / reversing

Most overhyped

1. The metaverse as a general-purpose computing platform (T-15-19). The specific evidence that hype outruns substance: Meta Reality Labs lost $8,647m in operating losses in H1 2026 on $833m of revenue — a ratio of roughly 10:1 — and the largest actual virtual world, Roblox, is guiding to a bookings decline with DAU 29m below peak. There is no separately material "metaverse" revenue line in any credible market model. What survives is games, which were games before the label was applied.

2. Fully generative games and autonomous AI NPCs (T-15-20). The evidence: Steam's own disclosure data shows ~60% of AI use is visual assets, with audio, localisation and writing following — the least architecturally consequential uses. No commercially significant shipped title is built around generative NPCs. Nvidia's ACE has been demonstrated at CES every year since 2023 without converting into a shipped commercial title, which is a three-year disconfirming series. The trend that is real — asset and localisation tooling — is being systematically mislabelled as the trend that is not.

3. Honourable mention: cloud gaming. Microsoft flagged "upcoming changes to Xbox Cloud Gaming" in September 2026; cloud hours remain a minor feature of subscription tiers (5–15 hours per month by tier), not a distribution channel.

Most overlooked

1. The creator cash-out rate, not the revenue-share percentage (T-15-04). Every UGC platform leads with "70% to creators". The cash number is ~23% of bookings. Attention has missed this because it requires reconciling a disclosure in Robux against a disclosure in dollars, across two documents, and neither party has an incentive to do it. It is the single most useful piece of arithmetic in this dossier.

2. Apple's China commission cut (2026-03-15, 30%→25% and 15%→12%). This received almost no Western coverage and is arguably as significant as the DMA changes: it is the first unilateral, non-court-ordered reduction in Apple's standard rate in a major market, made "following discussions with Chinese regulators", and it establishes that the 30% rate is negotiable without litigation.

3. Age assurance as an economic variable rather than a compliance cost (T-15-09). Roblox has quantified what almost no other youth platform has: safety architecture has a measurable revenue cost, visible in a 14–18% guided bookings decline. Every other youth-facing platform faces the same obligation and none has priced it publicly.

4. AI substitution at the hiring decision (T-15-13). Layoff trackers cannot see a role that is never opened. If 35% of founders are declining to hire, the employment picture is worse than falling layoff counts suggest, and the damage is concentrated in junior roles.

5. The compliance barrier below the platforms. Regulation aimed at Apple and Google is simultaneously imposing per-jurisdiction fee agreements, age assurance, AI disclosure and child-safety architecture on developers. These are fixed costs. The pro-competitive remedy at the top of the stack may be anti-competitive at the bottom, and nobody is measuring it.

Cooling — with the indicator that turned

• Consumer VR content (T-15-16). Indicator: studio closures, not hardware announcements — Polyarc, Vertigo Games Amsterdam, Schell Games layoffs, ~100 Apple Vision Pro staff, Lynx into liquidation. Hardware roadmaps (Project Phoenix, Pico Space Pro, Galaxy XR) continue regardless, which is the definition of an announcement-versus-units gap. Valve's Steam Frame at $1,059 moves PC VR further from mass market, not closer.
• Roblox engagement (T-15-17). Indicator: DAU down 29m from the Q3 2025 peak, hours down 11bn, and the first guided bookings decline. Revenue growth of +36% is a lagging artefact.
• Day-one subscription (T-15-18). Indicator: Microsoft removed day-one Call of Duty on 2026-04-21 and cut the price it had raised six months earlier; Xbox content and services revenue −10% YoY.

May reverse — and the mechanism

• The commission cuts could be substantially undone (T-15-01). Mechanism: the US district court accepts Apple's 15%/10%/5% external-link proposal, restoring most of the commission on the one route that currently costs developers nothing; or a change in US enforcement posture reduces pressure to maintain the remedies. The EU and Japan changes are statutory and harder to reverse; the US ones are judicial and contingent.
• Subscriptions could re-accelerate. Mechanism: a post-GTA-VI competitive environment in which Microsoft reinstates day-one flagship access to defend share.
• Roblox bookings could re-accelerate. Mechanism: age-verified older cohorts monetise better than the under-13s they replaced, which is the explicit bet management is making; Q4 2026 is the confirming or disconfirming datapoint.
• China could re-freeze licensing. Mechanism: precedent exists (2018, 2021–22). Approval policy is discretionary, unannounced and has reversed twice in eight years.

11. Risks and major uncertainties

Sector-specific risks.

1. Single-title concentration risk. Take-Two's FY2027 guidance of $8.0–8.2bn net bookings and Newzoo's entire console growth forecast rest substantially on one release on 2026-11-19. A slip or a commercial disappointment moves a sector aggregate.
2. Regulatory whiplash on commissions. Six jurisdictions, six different schedules, all changed within twelve months, several still contingent on pending rulings.
3. Child-safety liability. State attorney-general actions, platform bans (Kuwait, Indonesia, Bahrain; Russia's ban lifted 2026-06-10) and statutes like California AB 1709 create both compliance cost and existential market-access risk for youth platforms.
4. AI copyright exposure. Training-data litigation is unresolved. A studio that has shipped generative assets has no certainty about their provenance status.
5. Labour cost re-pricing. Enforceable AI clauses in ratified contracts change the cost of deployment, not just its optics.
6. Debt service on the EA LBO in a 3.50–3.75% rate environment with hawkish dissent.
7. Hardware input-cost inflation from tariffs and memory pricing, already visible in Sony's 2026-04-02 price increase.
8. Cheating and integrity costs: 51% of developers report player and revenue losses; an esports pivot to betting and prediction markets raises the stakes on match integrity.
9. Talent-pipeline depletion from four years of contraction plus suppressed junior hiring.

Genuine unknowns — and the distinction matters.

Things we don't know but could find out:

• XR unit shipments. No manufacturer publishes them. This is a disclosure choice, not a measurement impossibility, and it is why every XR unit figure in circulation is an estimate.
• Game Pass subscribers since February 2024. Microsoft stopped disclosing; the number exists.
• Fortnite Creative/UEFN per-creator payout distribution. Epic publishes a pool, not a distribution.
• Actual generative-AI use rates as opposed to disclosed ones.
• Whether developers are materially using the new link-out and alternative-distribution routes. Apple and Google know; nobody else does.

Things nobody can know:

• Whether generative AI compresses production cost enough to change what games get made, or only shifts which roles exist. The mechanism is plausible; the magnitude is unknowable in advance.
• Whether a consumer XR market exists at any price point, or whether the category is permanently enterprise and niche.
• Whether age-verified older cohorts monetise better than the younger ones displaced — this is a genuine natural experiment currently running at Roblox.
• Whether China's licensing normalisation is a durable policy or the current phase of a cycle.
• Whether GTA VI resets the premium price anchor upward or proves to be the last title that could have.

12. Scenarios to 2030

Base case — grinding single-digit growth with a re-priced toll layer. The market grows mid-single-digits from Newzoo's $213.9bn 2026 base; mobile leads; live-service concentration persists; commission settles in a 15–26% band with meaningful link-out leakage; employment stabilises at a structurally lower level with a hollowed-out junior tier; generative AI becomes standard at the asset and localisation layer without transforming design; consumer XR remains a sub-$5bn niche. Falsifiable early indicator: Take-Two's FY2028 guidance. If the post-GTA-VI year guides above ~$7bn, the premium tentpole model is healthier than this scenario assumes; below ~$6bn and the concentration is worse.

Upside — the toll cut converts into supply. Sustained 4–10 point margin relief plus AI asset-cost reduction makes mid-budget development viable again; the number of titles clearing $1m gross rises materially above the 311 recorded in 2025; new-title share of playtime recovers. Indicator: the 2027 and 2028 Steam cohort counts above $1m gross. A rise to 400+ would be the clearest possible confirmation; stagnation near 300 falsifies it.

Downside — the commission relief is clawed back and demand stalls. The US district court grants Apple 15% on link-outs; other jurisdictions soften under trade pressure; consumer spending flattens in a sticky-inflation environment; hardware prices rise further on memory and tariff costs; UGC platform engagement continues to decline under safety architecture. Indicator: the district court's ruling on Apple's proposal — the single most falsifiable near-term event in this sector, and it is pending now.

Disruption — agentic production collapses the cost curve. Engine-embedded agentic assistants (of which Unity's 2026-09-10 plugin is the first instance) move AI from assets into systems and content generation, cutting production time for a mid-budget title by a large factor and inverting the discovery problem from too few games to catastrophically too many. Indicator: Steam annual release counts. A step-change well above the current trajectory, combined with AI disclosure rates passing ~50% of releases, would confirm it; disclosure plateauing in the high twenties would not.

Regulatory — child safety becomes the binding constraint, not commission. Age assurance generalises from Roblox to every youth-facing platform via the UK Online Safety Act, the EU DSA, Australian rules and US state statutes; identity verification becomes a precondition of play; youth monetisation is structurally impaired. Indicator: whether a second major platform (Fortnite, Minecraft or a mobile publisher) adopts mandatory age assurance at Roblox's level of rigour by end-2027, and whether it discloses a bookings impact.

Failure — the tentpole model breaks in public. GTA VI slips again or disappoints; console hardware economics deteriorate further; one or more platform holders retreat from first-party development; the sector consolidates into live-service annuities plus a very long tail with nothing in between. Indicator: whether any Western publisher other than Take-Two greenlights a new $200m+ single-player premium title in 2027. If none does, the middle is gone.

13. Data gaps and limitations

Discovery capability was materially reduced in this research pass. The shared 200-call session WebSearch pool was exhausted after 8 of this agent's ~22 allocated searches. Per the macro brief's standing instruction, research continued via WebFetch on primary sources, which worked well for company IR, Apple and Google developer documentation, SEC filings and trade bodies, but eliminated the ability to discover sources not already known. This sector should be prioritised for a re-run with full search budget. The specific consequence is that consensus-checking of contested figures, non-English source discovery and coverage of smaller regional markets are thinner than they should be.

Sites that blocked or failed retrieval (and therefore are gaps rather than omissions): gamesindustry.biz, eurogamer.net, newzoo.com, theverge.com, circana.com press releases, sony.com IR PDFs (all formats attempted), pocketgamer.biz tag pages, unity investor relations, krafton.com IR, theesa.com/resources index, and Roblox's own JavaScript-rendered IR pages.

Specific unverified or single-sourced items, flagged in the records:

• Annual layoff totals for 2022–2025 could not be retrieved; only the 2026 year-to-date figure (~4,600, 22 studios, to 2026-07-01) was obtained. The 2023 and 2024 figures quoted in §7 as ~10,500 and ~14,600 are from prior knowledge and were not verified in this pass.
• EA take-private transaction value. EA's own completion announcement does not state it. The widely reported ~$55bn figure is not verified from a primary source here and is recorded as unverified in the entity record.
• PS5 cumulative units (93m as of 2026-03-30) and the 2026-04-02 US price increase rest on a Tier C chronology because Sony's IR site was unreachable. Re-verify against Sony IR.
• Game Pass subscriber figures (~30m, July 2026) and the 2026-04-21 price reduction are Tier C sourced, corroborated only in direction by Microsoft's Tier A −10% revenue line.
• Fortnite Creative/UEFN creator payouts could not be obtained at all. Epic publishes an engagement-payout pool but the per-creator distribution is not disclosed, so no cross-platform comparison with Roblox's DevEx is possible. This is a real, structural gap.
• XR unit shipments. No manufacturer publishes them; IDC and Counterpoint estimates were not retrievable. The only hard XR number in this dossier is Meta's Reality Labs segment revenue, which bundles smart glasses with VR and is therefore not a clean proxy.
• Esports revenue and prize-pool aggregates are absent. Only qualitative trade coverage was obtainable, which is why T-15-14 is scored low on evidence.
• Korea beyond Krafton, and India, Brazil and MENA markets are thin. Krafton's record quarters are captured; NCSoft, Nexon, Netmarble and Kakao Games are not.
• Non-English primary sources were used indirectly (Nintendo's IR, Tencent's interim report, Apple's Japan pages, NPPA counts via Niko Partners) but the NPPA batch lists themselves were not read in Chinese. Recorded as a regional-source limitation.

Conflicting figures recorded rather than resolved (full detail in the contradictions arrays of the trend records):

1. US external-link commission: 0% (Epic's reading) vs 15%/10%/5% (Apple's proposal). Undecided by the court as of 2026-09-15.
2. 2026 layoffs: ~4,600 actual YTD vs ~14,259–14,666 modelled full-year forecasts. These are different objects and are routinely quoted interchangeably.
3. Roblox growth: +36% revenue vs +8% bookings vs −14% to −18% guided bookings, same company, same quarter. An accounting distinction, not a source disagreement.
4. Steam AI disclosure: 26.5% of registered demos vs ~20% of all Next Fest entries. Two valid populations, routinely conflated.
5. Subscription direction: US subscription spending +20% (2025) vs Xbox content and services −10% (quarter to 2026-06-30). Category grew; market leader shrank.
6. Reality Labs: +16.5% segment revenue growth vs consumer VR studio closures. Smart glasses growth masking a declining VR content business inside one reported line.
7. NPPA 2026 forecast revised from ">2,000" (January) to ">2,300" (June) by the same firm.

Where a Phase 2 licensing budget would have to go: Circana's granular US series, Niko Partners' full China and Asia models, Sensor Tower/Appfigures mobile estimates, and the ISFE/GSD European panel. All four are the authoritative measurement in their domain and all four publish only headlines free.

14. Ranking scorecard

15. Sources

1. "Apps in the European Union — fee schedule and Core Technology Commission", Apple Developer, https://developer.apple.com/support/apps-in-the-eu/, 2026-08-18 — A
2. "Adjustments to the China storefront", Apple Developer News, https://developer.apple.com/news/?id=dadukodv, 2026-03-12 — A
3. "App distribution in Japan (Mobile Software Competition Act)", Apple Developer, https://developer.apple.com/support/app-distribution-in-japan/, 2025-12-17 — A
4. "Service fees", Google Play Console Help, https://support.google.com/googleplay/android-developer/answer/112622, effective 2026-06-30 — A
5. "Apple proposes to take a 15% cut of purchases made outside the App Store", TechCrunch, https://techcrunch.com/2026/08/14/apple-proposes-to-take-a-15-cut-of-purchases-made-outside-the-app-store/, 2026-08-14 — B
6. Roblox Corporation Form 8-K, Q2 2026 results, SEC/Roblox, https://www.stocktitan.net/sec-filings/RBLX/8-k-roblox-corp-reports-material-event-8934bc65536a.html, 2026-07-30 — A
7. "Roblox Q2 2026 slides: revenue rises 36% as bookings growth stalls", Investing.com, https://www.investing.com/news/company-news/roblox-q2-2026-slides-revenue-rises-36-as-bookings-growth-stalls-93CH-4826381, 2026-07-30 — B
8. "Earn on Roblox", Roblox Creator Documentation, https://create.roblox.com/docs/production/earn-on-roblox, accessed 2026-09-15 — A
9. "Dedicated Video Game Sales Units", Nintendo Co., Ltd. Investor Relations, https://www.nintendo.co.jp/ir/en/finance/hard_soft/index.html, data as of 2026-06-30 — A
10. "Take-Two Interactive Reports Results for Fiscal First Quarter 2027", Take-Two Interactive, https://www.take2games.com/ir/news/take-two-interactive-software-inc-reports-results-fiscal-first-6, 2026-08-07 — A
11. Microsoft FY26 Q4 earnings press release, Microsoft Investor Relations, https://www.microsoft.com/en-us/investor/earnings/fy-2026-q4/press-release-webcast, 2026-07-29 — A
12. "Meta Reports Second Quarter 2026 Results", Meta Platforms Investor Relations, https://investor.atmeta.com/investor-news/press-release-details/2026/Meta-Reports-Second-Quarter-2026-Results/default.aspx, 2026-07-29 — A
13. Tencent Holdings 2026 Interim Report, Tencent Holdings Ltd., https://www.tencent.com/wp-content/uploads/2026/08/E700_IR.pdf, 2026-08-13 — A
14. "NetEase Announces Second Quarter and Interim 2026 Unaudited Financial Results", NetEase, Inc., https://ir.netease.com/news-releases/news-release-details/netease-announces-second-quarter-and-interim-2026-unaudited, 2026-08-20 — A
15. "EA Announces Completion of Acquisition by PIF, Silver Lake, and Affinity Partners", Electronic Arts, https://www.ea.com/news, 2026-08-04 — A
16. "2025 U.S. Consumer Spending on Video Games Nears Pandemic-Level Peak at $60.7 Billion", ESA / Circana / Sensor Tower, https://www.theesa.com/2025-u-s-consumer-spending-on-video-games-nears-pandemic-level-peak-at-60-7-billion-second-highest-on-record/, 2026-02-11 — A
17. "2026 Essential Facts About the U.S. Video Game Industry", Entertainment Software Association, https://www.theesa.com/resources/essential-facts-about-the-us-video-game-industry/2026-data/, 2026-08-19 — A
18. "Newzoo: The Gaming Market in 2026", Newzoo via GameDev Reports, https://gamedevreports.substack.com/p/newzoo-the-gaming-market-in-2026, 2026-08-25 — B
19. "China's NPPA approved 1,771 video games in 2025", Niko Partners, https://substack.nikopartners.com/p/chinas-nppa-approved-1771-video-games, 2026-01-05 — B
20. "China approves 158 games in May as licensing pace accelerates in 2026", PocketGamer.biz / Niko Partners, https://www.pocketgamer.biz/china-approves-158-games-in-may-as-licensing-pace-accelerates-in-2026/, 2026-06-03 — B
21. Video Game Industry Layoffs tracker, community-maintained, https://publish.obsidian.md/vg-layoffs, updated 2026-07-01 — C
22. "The AI Disclosure Report — Steam Next Fest June 2026", Llama & Griffin (GameDiscoverCo/SteamDB data), https://www.llamagriffin.com/Data/SteamNextFestJune2026/part-2.html, 2026-06-30 — B
23. "Are Steam players of particular genres less accepting of AI?", GameDiscoverCo, https://newsletter.gamediscover.co/p/are-steam-players-of-particular-genres, 2026-08-18 — B
24. "How have the genres of hit PC games changed since 2021?", GameDiscoverCo, https://newsletter.gamediscover.co/p/how-have-the-genres-of-hit-pc-games, 2026-08-25 — B
25. Game Developer news (Steam Frame pricing; Polyarc shutdown; EA NHL 27 AI voiceover; Rockstar Glasgow tribunal; Kojima/Xbox), Informa, https://www.gamedeveloper.com/, 2026-09-14 — B
26. PocketGamer.biz industry news (Blizzard union contract; Unity Claude Code plugin; Talent Signals founder survey; China market >$50bn; Roblox Everywhere; California AB 1709; Anybrain cheating survey), Steel Media, https://www.pocketgamer.biz/, 2026-09-15 — B
27. Krafton press room — Chief AI Officer appointment; Q1 and Q2 2026 results, KRAFTON, Inc., https://www.krafton.com/en/news/, 2026-07-29 — A
28. "Epic Games v. Google", Wikipedia, https://en.wikipedia.org/wiki/Epic_Games_v._Google, accessed 2026-09-15 — C
29. "Epic Games v. Apple", Wikipedia, https://en.wikipedia.org/wiki/Epic_Games_v._Apple, accessed 2026-09-15 — C
30. "Xbox Game Pass", Wikipedia, https://en.wikipedia.org/wiki/Xbox_Game_Pass, accessed 2026-09-15 — C
31. UploadVR (VR studio closures; Vision Pro staff reductions; Project Phoenix; Lynx/EssilorLuxottica), https://www.uploadvr.com/, 2026-09-13 — B
32. "Steam isn't slowing down in 2026 — 42,318,602 concurrent users", PCGamesN, https://www.pcgamesn.com/steam/peak-concurrent-user-count-march-2026, 2026-03-16 — B
33. "Adjust & Sensor Tower: Japan's Mobile Market in 2026", via GameDev Reports, https://gamedevreports.substack.com/p/adjust-and-sensor-tower-japans-mobile, 2026-09-03 — B
34. "PlayStation 5" (cumulative units; April 2026 price increase), Wikipedia citing Sony IR, https://en.wikipedia.org/wiki/PlayStation_5, accessed 2026-09-15 — C
35. Roblox Newsroom and "Roblox" (age assurance rollout, account tiers, regulatory actions), https://about.roblox.com/newsroom, accessed 2026-09-15 — C
36. "Introducing the App Store Mini Apps Partner Program", Apple Developer News, https://developer.apple.com/news/?id=xcz1s7cz, 2025-11-13 — A
37. "Changes to iOS in Brazil", Apple Developer News, https://developer.apple.com/news/?id=dhwadr2x, 2026-06-18 — A
38. Esports Insider (Riot prediction markets; Data.Bet; PGL/1xBet; Tier-2 ecosystem), https://esportsinsider.com/, 2026-09-11 — B
39. Macro Context Brief, Phase 1 research programme, internal, 2026-09-15 — A (framing only; figures not re-reported)

Source-tier distribution: 17 Tier A, 15 Tier B, 7 Tier C. No claim in this dossier rests on Tier C alone except where explicitly flagged as single_source in the trend records (PS5 units, Game Pass subscriber history, Epic layoff headcount) with the gap recorded in §13.

More research in this sector · Explore trend records