← Module 7/The economics of the industry
RU
Module 7 · The present (2018–26)

The economics of the industry

The largest contraction in the history of games (2023–26, ~45k jobs) is not an accident but a capital cycle: cheap COVID money inflated hiring and AAA budgets, rates went up — the bubble burst. The survivors consolidated into a handful of giants and sovereign funds, and workers started to unionize.
~17 min💰 macro + 🤝 labor
The gist in 30 seconds
The 2023–26 contraction (the largest in history, ~45,000 jobs from 2022 to mid-2025) is a classic capital cycle: overhiring during the COVID boom and cheap money → rising rates + inflated AAA budgets ($100–200M+) + live-service saturation + early GenAI → mass layoffs (Embracer cut from 15,701 to 7,873, roughly half). Consolidation: Microsoft (Activision, $69B), Sony (Bungie, $3.6B), Tencent (stakes everywhere), Saudi PIF ($35B+, one of the few buyers during the crunch) — power contracted toward subscription platforms and sovereign capital; the mid-tier got squeezed out (a barbell, as in indie). Unionization — the largest wave in history: Activision QA (2021), ZeniMax (2023), the SAG-AFTRA strike over AI protections, settled July 9, 2025 (consent for AI replicas + disclosure + a 15% raise). GenAI is double-edged (productivity vs displacement). Plus broken discovery (50k games/year, 80% of players are in games more than 5 years old) and the burst crypto bubble (a Ponzi). For a developer/founder in 2026: a buyer's market for seniors, indie-as-refuge, more protections, AI clauses, and a brutal attention market.

The mechanism: the cycle, consolidation, labor

The capital cycle — why it blew up now

The contraction is not "games are over" but a cheap-capital cycle, the same one running through all of tech. The COVID boom (2020–21): a surge in engagement plus near-zero rates → studios overhired and inflated budgets. Then the reversal (2022–25): rates rose (capital got expensive), engagement normalized, AAA costs swelled to $100–200M+, live service saturated, and early GenAI tooling gave an excuse to "do the same with fewer hands". The result: layoffs across >100 companies, ~45,000 jobs from 2022 to mid-2025. Embracer makes it vivid (a debt-financed acquisition empire collapsing):

15701→7873 (−8000,~half; ≈⅕ of all losses)

The drivers read like a textbook financial cycle: cheap money → overinvestment → capital gets expensive → correction. Games simply went through it more sharply because of AAA cost inflation and the bet on live service.

Consolidation and sovereign capital

Those who survived bought up the weakened, concentrating power:

The result is a barbell at industry scale: platform giants and sovereign funds at the top, cheap indie at the bottom, the middle layer squeezed out (like mid-budgets in the indie lesson, but at the scale of the whole industry).

Unionization — labor's counter-move

Decades of a non-union, contract-based norm gave way to the largest union wave in the history of games (2022–26): Activision QA (Raven, Dec 2021 — the first major US union), ZeniMax/Bethesda QA (2023, with Microsoft staying neutral post-Activision), Sega of America (2023), all under the CWA umbrella. The peak was the SAG-AFTRA strike (voice actors, 2024–25) over protection from AI voice: 11 months, settled July 9, 2025 (95% in favor), winning consent for AI digital replicas plus disclosure and +15.17% on pay. Why now: crunch culture hit its limit, the layoff wave made a union feel like insurance, big-tech unionization (Amazon/Google) softened the ground, and Microsoft's neutrality opened the door at the largest publisher. AI clauses in contracts now determine what can be shipped at all.

GenAI — double-edged

Generative AI is both a driver of layoffs here (productivity → fewer hands in art/QA/writing — it was among the causes of the contraction) and a lever for indie (a small team producing AAA volume; compare scope and Meshy pipelines). Plus legal and quality risks and labelling: Steam rewrote its AI-disclosure policy, and ~8% of games carry an AI disclaimer. For you as an AI engineer this is the central question: does the tool make a worker more productive or redundant — and union AI clauses set the frame for the answer directly.

Broken discovery and the crypto bubble

Demand is skewed too: 50k+ games a year, 80% of players in 2024 were in games more than 5 years old (Fortnite, GTA Online, FFXIV, Minecraft), finding a new game is brutally hard, and virality (TikTok) is the only reliable channel (Lethal Company, Palworld). A separate instructive bubble is crypto/NFTs (2021–22): Web3 pulled in $4B+, Axie peaked at ~$1.3B/month → the Ronin hack ($620M) plus the Terra/Luna and FTX collapses gutted the sector. The lesson for a designer: when a player economy rests on an inflow of new money rather than on gameplay value, it is a Ponzi, not a game.

🕹 What to read — and what to notice

You "play" industry data here — the numbers and events are visible in trackers and deals.

Layoff trackers the human scale of the contraction

Public trackers (Wikipedia's "2022–2026 video game layoffs", industry roundups) show ~45k jobs across >100 companies — not an abstraction but people and shuttered studios.

🎮 Read: open a layoff tracker and notice the rhythm: the 2023–24 peak, who and when. Line it up against the macro (the 2022 rate rise) — you will see a capital cycle, not "a few games failed".

Microsoft–Activision ($69B) consolidation → redundancy

The largest deal in the history of games (closed Oct 2023) and what followed it: integration layoffs, studio closures, content pulled under Game Pass. A textbook "acquire for a content moat → redundancy".

🎮 Read: the chronicle of the deal and the cuts that followed. Notice the logic: the giant buys volume, locks content into the subscription, trims the "duplicates". Power concentrates — that is consolidation live.

SAG-AFTRA / Concord / Axie labor, saturation, bubble

Three signals of the era: the SAG-AFTRA win on AI voice (labor organizes), the failure of Concord (~2 weeks, live-service saturation) and the collapse of Axie/NFTs (a crypto Ponzi). Each is a facet of one macro fracture.

🎮 Read: take apart one of the three. In SAG-AFTRA notice the AI clauses (consent for replicas) — how labor answers AI; in Concord, the price of betting on a saturated live-service market; in Axie, how "play-to-earn" turned out to be a pyramid built on an inflow of new money.

Deep end · macro: the capital cycle, AAA cost inflation and the barbellskippable

The mechanics of the cycle

Cheap capital (low rates in 2020–21) makes hiring and speculative bets cheap → companies overhire against demand that looks permanent (the COVID engagement surge). Rising rates (2022+) make capital expensive → investors demand profitability now → cuts to growth hiring and unprofitable lines. Games went through it more sharply because of two amplifiers: AAA cost inflation (budgets of $100–200M+ require a top-10 hit to break even — there is almost no room for moderate success) and live-service saturation (everyone chased the same Fortnite, and the attention market is not elastic — Concord as the symbol).

The industry barbell

The same barbell as in indie production, but at the level of the whole sector: at the top, platform giants (Microsoft/Sony/Tencent) and sovereign capital (Saudi PIF as one of the few buyers during the crunch); at the bottom, cheap indie; and the middle layer (mid-budget studios) squeezed out. Consolidation follows directly: in a crunch the strong buy the weak, and power and IP concentrate.

GenAI in the cost equation

GenAI hits the cost of production from both sides: it lowers costs (fewer hands on assets/QA/text — part of the layoff driver) and at the same time hands indie a lever (AAA volume from a small team). The net effect on employment is disputed; the effect on the cost structure is downward, which strengthens both the barbell and the pressure on the middle.

Deep end · labor: why unions now and what AI clauses changeskippable

Unionization is a rational response to specific conditions, not a fashion.

  • Crunch reached its cultural limit (burnout postmortems, Indie Game: The Movie as an early symptom at small scale).
  • The 2023–24 layoff wave turned a collective agreement into "insurance": an individual contractor is defenceless against a cut.
  • Big-tech unionization (Amazon, Google) normalized unions in tech and supplied the infrastructure (CWA).
  • Microsoft's neutrality post-Activision opened the door at the largest publisher (ZeniMax QA — the first wall-to-wall recognized union).

AI clauses as the new front

The SAG-AFTRA strike was explicitly about AI: consent for digital replicas plus disclosure of AI voice use. That changes what can be shipped: training a model on an actor's voice without consent is now a breach of contract. For an indie and for an AI engineer this is a concrete frame: AI assets require provenance and consent, otherwise you carry legal and reputational risk. Labor turns "technically possible" into "legally permitted".

Analogy
The industry went through a classic gold-rush boom and bust. Cheap money from the COVID boom inflated hiring and AAA budgets like a gold-rush bubble; rising rates punctured it; the survivors bought up the claims of the ruined and merged into a handful of large holders (plus sovereign-fund investors), while the prospectors began organizing into unions for protection. GenAI is the new steam shovel: it makes every worker more productive and reduces how many workers are needed. Rush, bust, consolidation into a few large hands, and organizing labor — the recognizable pattern of any capital-fueled sector.
Why it matters
Understanding the macroeconomics of the industry is mandatory for anyone building a career or a company in it (yourself included). The 2023–26 contraction is not an accident but a capital cycle (cheap money → overinvestment → correction) on top of structural pressures (AAA cost inflation, live-service saturation, GenAI, broken discovery). Consolidation compressed power, unions rose as a counterweight, AI is reshaping labor. Reading those forces — and hype cycles (crypto, VR, cloud) — is the difference between betting a studio on the peak of inflated expectations and building for the real market.
🔁 Beyond games — where this transfers
The lesson is the macro capital cycle, consolidation, and labor's answer to automation — directly about your own industry.

ML / AI (your domain): the boom-bust in games is literally the tech/AI industry pattern: cheap capital → hiring boom → overinvestment → correction (the same 2022–23 tech layoffs, and the current AI capex boom — where are we in the cycle?). GenAI as a layoff driver is the central question of your work: does the tool make a worker more productive or redundant? — and union AI clauses (SAG-AFTRA: consent for replicas) set the governance frame for what you build. Hype-cycle literacy (crypto→crash, VR→niche, metaverse→hangover) applies to AI hype directly: is this the plateau or the peak? Consolidation/sovereign capital ⇄ the concentration of compute and capital in AI (a handful of labs plus hyperscalers plus sovereign funds); the barbell ⇄ the model market (frontier labs or small open-weights, with the middle squeezed out). This is the most career-applicable lesson in the course for an AI engineer in games.

Finance / strategy: the rate cycle → hiring/cuts; M&A as consolidation during a crunch; do the due diligence on "where are we in the cycle" before you grow or borrow.

Labor / society: automation → labor organizing; provenance and consent for data (voice/face) as a legal frame for AI; unions and regulation as a counterweight.

Principle: industries breathe in capital cycles and hype waves; consolidation and automation concentrate power, while labor and regulation push back. Read where you are in the cycle and on the hype curve — it determines both a career and a company's bets.

🔧 Run it and poke at it — on your home machine
What to look at is above (🕹). Here — trying on the macro lens.
🔧 Tinker (macro breakdown) ~40 min
Take a 2022–26 layoff tracker and overlay it on a chart of Fed rates: you will see the boom→bust correlation. Separately, take apart one deal (MS–Activision) as consolidation: what was bought, what was closed, why the content went into Game Pass. State where the industry is right now in the cycle and what that means for hiring and for indie.
🧪 Test (career/company) ~15 min
For yourself as an engineer/founder: is a buyer's market for seniors an opportunity or a risk? Is indie-as-refuge worth it? Which AI clauses will touch what you want to ship? Run a "hot" AI technology through the hype cycle (peak/trough/plateau) — is it worth betting a project on right now?
Checklist: overlaid layoffs on interest rates (the capital cycle); took apart one deal as consolidation; located the industry on the cycle; assessed AI clauses and the hype cycle for your own project.
Connections
foundation
Indie production — the barbell and AAA cost inflation that squeezed out the middle layer and made indie a refuge.
foundation
Distribution and visibility — broken discovery (50k games/year, 80% in old games) is the demand side of the same crisis.
adjacent
VR/AR and the cloud — hype cycles as a special case: where a technology did not deliver on its promises (crypto being the purest bubble).
next
The capstone (Module 12, coming) — how to build a project and a career while reading these macro forces (including your own Novgorod).
Questions worth asking
Why did the contraction happen in 2023–26 if games are more popular than ever?
Because the cause is macro-financial, not "demand fell". A classic capital cycle: cheap money plus the COVID engagement surge (2020–21) → studios overhired and inflated budgets against demand that looked permanent. Rising rates (2022+) made capital expensive → investors demanded profitability now → a correction to growth hiring. Games went through it more sharply because of AAA cost inflation (you need a top-10 hit to recoup $150M) and live-service saturation. The popularity of games is beside the point — what was cut was the excess hiring of the boom, not a response to a shrinking audience. All of tech went through the same cycle.
Is GenAI really destroying game jobs — or is that panic?
The truth is more complicated than yes/no. GenAI genuinely was among the drivers of the contraction: art/text/QA tools made it possible to "do the same with fewer hands", and that accelerated the cuts. But it is also a lever: a small team can carry AAA volume (indie amplification, Meshy pipelines). The net effect on the number of jobs is disputed and depends on whether the volume of games produced grows faster than the labor cost of each one falls. What is not disputed: GenAI shifts labor (less routine asset production, more curation and integration) and provokes a legal response (SAG-AFTRA's AI clauses). For you as an AI engineer this is not an abstraction but the direct subject of your work: design tools that amplify people rather than simply replace them.
Why do the giants buy studios during a crunch — isn't it more expensive to buy at the peak?
The opposite — a crunch is the buyer's moment. In a contraction assets get cheap (studios in debt, investors exiting), while giants with strong balance sheets (Microsoft) and sovereign funds (Saudi PIF) have capital when nobody else does. That is why PIF became "one of the few buyers". The motive is strategic: Microsoft takes Activision for $69B for the Game Pass content moat (exclusives lock in subscribers), not for immediate profit. Consolidation during a crunch concentrates IP and power with the survivors — that is exactly the mechanism that turns a downturn into a redistribution of the market in favor of the strongest.
Why did unions rise now, after decades of absence?
Several conditions coincided. Crunch culture reached its limit (years of burnout postmortems). The 2023–24 layoff wave turned a collective agreement into insurance — an individual contractor is defenceless, a union gives leverage. Big-tech unionization (Amazon/Google) normalized it in tech and supplied the infrastructure (CWA). And Microsoft's post-Activision neutrality removed the main barrier at the largest publisher. On top of that, AI gave a new reason to organize — protecting voice and likeness from AI replicas (SAG-AFTRA). So it is not "they suddenly felt like it" but a rational answer to a specific combination of threats (layoffs plus AI) and opportunities that opened up (neutrality plus infrastructure).
Is the industry dying — or is this a healthy correction?
More a painful correction and redistribution than a death. Audience and revenue are as large as ever; what collapsed was the excess hiring and budgets of the boom and an unviable middle layer. The result is not disappearance but concentration (power moving to platforms and sovereign capital) and polarization (giants plus indie, the middle squeezed out). For a worker that is hard (a buyer's market for seniors, instability), but it opens windows too: cheap talent, indie-as-refuge with a GenAI lever, new union protections. "Dying" is the wrong frame; the right one is "restructuring expensively into a more concentrated and polarized shape". Reading that soberly matters more than falling into apocalypse or denial.
Further reading