Video game experts say Google’s Project Genie isn’t an industry killer. Investors don’t seem convinced.

Video game stocks have been getting repeatedly KO’d in recent weeks, following the launch of Google’s new generative-AI tool, Project Genie.

The tool, a “general-purpose world model” available to Google AI Ultra subscribers who pay $125 a month in the US, lets users create interactive, “playable” worlds with just a text or image prompt.

Gaming giants have shed billions of dollars since Project Genie launched on January 29, with companies including Unity Software, Take-Two, and Roblox experiencing sharp sell-offs. The moves appear to be in line with a broader trend Bloomberg’s Matt Levine calls the “AI Whateverpocalypse,” wherein newly announced AI applications trigger steep losses in whatever market they happen to focus on (be it enterprise software, financial data, or trucking). Combined with high memory costs weighing on hardware makers (also due to AI, in a different way), gaming is off to a rough start in 2026.

It’s clear that investors see Google’s world model as a potential silver bullet for the gaming industry, set to dramatically lessen the effort and costs required to make immersive worlds — and sharply increase the supply of interactive games (some of which could rip off popular intellectual property).

“Right now, too many investors treat gaming like a horse track, betting on the next breakout hit title rather than understanding how franchises are patiently built over decades,” said Joost van Dreunen, CEO of analytics firm Aldora and a gaming strategy professor at NYU. “After the postpandemic cooldown and tighter global trade dynamics, capital has crowded into familiar IP and scaled incumbents, so something like Project Genie feels disruptive by default.”

Industry executives, analysts, and employees have worked to dispel that idea and clarify Project Genie’s actual, current capabilities, but the market has proved tough to convince.

Currently, Project Genie can only generate one-minute-long experiences at 720p resolution and 24 frames per second. mBank analyst Piotr Poniatowski noted that the experiences lack depth, as users can’t perform complex actions fundamental to standard gameplay like crouching, climbing, dodging, or shooting. Poniatowski called Project Genie a “one-minute-long walking simulator generator,” and said the market’s sell-off was unjustified.

Raymond James upgraded “Grand Theft Auto” maker Take-Two in a note last week, with analyst Andrew Marok writing that the current market sentiment is “overdone.” From Marok:

“We understand world models as an additional tool in the content-creation stack, not a substitute for what modern game engines can do — particularly at the AAA scale. That matters for TTWO, where the value is tied less to ‘can you create a game’ and more to ‘can you ship this game to massive audiences,’ which entails cinematic quality, deterministic gameplay, live services, and multiplayer-first monetization...

It seems as if investors have jumped from ‘world models can generate environments’ to ‘therefore engines are obsolete.’ In our
view, that skips the hard part: shipped games are built on the deterministic end of this spectrum for reasons that are structural, not cosmetic.”

Gaming execs have also taken time to address the sell-off, in part by highlighting their own companies’ Genie-like AI efforts. Earlier this month, Roblox appeared to answer Project Genie by launching the open beta of its “4D” AI creation tool. Roblox’s tool can generate interactive objects, like a drivable car with moving wheels.

Unity, which makes the gaming engine used by popular titles like “Hollow Knight” and “Pokémon Go,” said AI-driven authoring is a major focus for the company in 2026. In the company’s earnings call February 11, CEO Matthew Bromberg said Unity will unveil a beta version of its upgraded Unity AI next month, which will “enable developers to prompt full casual games into existence with natural language only.”

“We believe world models are going to be a source of inspiration and assets for creators, but that they’re not in any way going to replace game engines. They are complementary, not duplicative,” Bromberg told investors.

Lost in much of the chatter around AI-generated games is the broader gaming community’s fierce rejection of AI thus far. In December, game publisher Running With Scissors scrapped a planned game over negative reactions to AI-generated graphics in its trailer. In the same month, the Indie Game Awards revoked its Game of the Year award for “Clair Obscur: Expedition 33” after it discovered the developer, Sandfall Interactive, had launched the game with AI-generated textures.

While Project Genie has inspired some experimentation around the limits of its copyright-violating capabilities, it would stand to reason that AI-generated games would need to prove they can overcome the near-certain audience backlash before becoming profitable to a degree that would threaten companies the size of Take-Two or Roblox.

“What the execs may be too polite to say is that it’s naive to assume the current generation of world-building engines poses a real threat to the existing supply side,” van Dreunen said. “Sustainable value in games comes from proprietary engines, live operations, and deeply rooted player communities — systems that take years to mature and can’t be conjured by models generating low-grade worlds without a durable business backbone.”