The $22 billion AI gaming windfall that doesn't add up
Morgan Stanley published a note on Tuesday that has been widely picked up. In this, the bank argues that advanced AI tools could cut video game development costs by 44%, unlocking about $22 billion in annual profits. The report is directionally interesting, but the number is not to be trusted.
The chain of reasoning runs as follows. Morgan Stanley estimates global consumer spending on video games will total $275 billion in 2026. It assumes roughly 20% of that — about $55 billion — is reinvested in game development and operations. Apply a 44% AI-driven cost reduction to that $55 billion total, and you arrive at approximately $22 billion in savings, which the report frames as additional profit for the sector.
However, each of the links in this equation is doing a lot of heavy lifting.
The $275 billion market figure is broadly defensible, though it depends on exactly what you count, but the 20% reinvestment rate is a blended average that quietly collapses a huge range of cost structures. Triple-A console games routinely reinvest 30 to 50% of eventual revenue into development. Mobile casual games, which now account for perhaps half of total consumer spending, operate on far leaner budgets.
Treating the whole market as a single 20% figure produces a tidy calculation but obscures the reality underneath it.
AI ≠ savings
The 44% cost reduction is the most aggressive assumption and the one that deserves the most scrutiny. Most measured productivity studies of AI coding and content tools show efficiency gains in the range of 10 to 30% on specific, well-defined tasks under favourable conditions.
Applying a 44% reduction to the entire development budget, which includes creative direction, level design, narrative, audio, playtesting, polish, project management, server infrastructure and marketing, is not a forecast. It is a best-case scenario applied uniformly to every line item, including the ones AI currently has little purchase on.
Savings ≠ profits
Even accepting the numbers at face value, the step from “cost savings” to “unlocked profits” is not automatic. If AI lowers the cost of making mid-scale games, more mid-scale games probably get made. Prices face downward pressure. Alternatively, publishers may reinvest savings into larger scope, higher production values or more aggressive marketing rather than banking the margin.
Morgan Stanley itself argues that smaller public companies such as Playtika and Netmarble will face greater competitive pressure as AI reduces barriers to entry, which is precisely a mechanism that erodes the profit upside the report is simultaneously celebrating.
Incumbents ≠win
The more consequential problem with the report is its conclusion about who benefits. Morgan Stanley identifies large publicly-owned platforms such as Tencent, Sony and Roblox as the key winners, followed by publishers like Take-Two, EA and Ubisoft. The logic is that companies with proprietary data, IP, live operations and distribution scale will capture the most value.
That is a reasonable description of how AI has tended to reinforce incumbency in mature, slowly-moving industries. It is a much weaker argument in an environment where AI is compressing the cost and time required to build games.
The whole premise of the report, that development costs could fall by nearly half, is also the premise under which a small team with no legacy infrastructure, no expensive middle management and no multi-year release pipeline can build and ship something competitive.
The incumbents’ advantage is their IP, their catalogue, their distribution and their capital. Their disadvantage is that they are carrying massive organizational weight and processes built around a legacy cost structure.
History suggests when a technology dramatically lowers the cost of production in a creative industry, the main beneficiaries are not the existing large companies but the new entrants who build natively around the new tools.
The early mobile era is instructive: the conventional wisdom in 2009 was that major publishers – EA, Take-Two, Ubisoft etc – would dominate the App Store because of their IP and marketing budgets. Supercell, Zynga, King and a handful of others with no such advantages ended up writing the decade’s defining commercial story instead.
AI may not follow the same pattern exactly; platform control matters more now, and the distribution chokepoints are different. Good news for Roblox, which is already leaning heavily into AI.
But a report that confidently assigns the gains to today’s listed companies is making a bet on incumbency that the disruptive potential of the technology puts in question.
If it materialises at all, the $22 billion seems unlikely to land where Morgan Stanley expects.
