Daniel Hertzberg The corporate arms race for dominance in artificial intelligence is on. This year America’s large tech firms will spend [about $400 billion](https://www.economist.com/leaders/2025/09/11/what-if-the-3trn-ai-investment-boom-goes-wrong) on AI infrastructure. [Nvidia](https://www.wsj.com/market-data/quotes/NVDA) CEO Jensen Huang [estimated](https://techcrunch.com/2025/10/10/the-billion-dollar-infrastructure-deals-powering-the-ai-boom/) that $3 trillion to $4 trillion will be spent on AI infrastructure by the end of the decade. Some [predict](https://fortune.com/2025/03/18/google-deepmind-ceo-demis-hassabis-agi-outsmart-human-workers-job-replacement-ai/) that artificial general intelligence, a stage of development when AI can match or exceed the cognitive abilities of human beings, is only a few years away. The largest AI companies are vying to reach that milestone first—and [reap](https://www.economist.com/leaders/2025/09/11/what-if-the-3trn-ai-investment-boom-goes-wrong) unimaginable returns. But the focus on artificial general intelligence obscures where the AI revolution is moving even faster: advertising. Google, [Meta](https://www.wsj.com/market-data/quotes/META) and [Amazon](https://www.wsj.com/market-data/quotes/AMZN) are spending hundreds of billions on AI to defend and further build their advertising businesses. At their core, Google, Meta and to a lesser extent Amazon are advertising behemoths. Last year Google earned 78% of $346 billion in revenue from advertising, and Meta 98% of [$164 billion](https://www.sec.gov/Archives/edgar/data/1326801/000132680125000017/meta-20241231.htm). Amazon earned [about $56 billion](https://www.sec.gov/Archives/edgar/data/1018724/000101872425000004/amzn-20241231.htm) from direct advertising. Together, the three companies raked in nearly half of the $1 trillion in ad spending worldwide. AI allows these companies to improve the services—search, social networking and e-commerce—around which their advertising businesses are built. It also positions the companies to enter the next phase of their dominance, making advertising itself smarter, faster and more automated—a shift that’s already transforming how ads are created and delivered. Google, Meta and Amazon have built AI systems that integrate the creation, testing, optimization and placement of advertising. Google Performance Max, Meta’s Advantage+ and Amazon’s AI Creative Studio take basic assets—images, videos, ad copy, headlines, sounds and voices—and combine them to create hundreds of ad variations. The AI systems compare the ads’ performance, select the best performing mix for a campaign and place the ads across their properties. No competitor can offer what these platforms provide: AI-powered ad creation integrated with access to billions of users and decades of data. The result: Advertisers get better campaign performances. Nielsen tracked more than one million performance campaigns and found, for example, that AI-powered campaigns on YouTube delivered [17% higher returns.](https://www.nielsen.com/insights/2025/google-mmm-case-study/) Google [reports](https://blog.google/products/ads-commerce/google-ai-max-for-search-campaigns/) that advertisers using AI-powered search campaigns typically see 14% more conversions at a similar cost. Improved results lock advertisers into these platforms, which in turn learn and improve with every campaign. Ad creation has traditionally been the purview of creative workers and ad agencies. The new AI systems create ads from text prompts and by combining digital assets. They resemble the first textile mills and assembly lines—more productive, faster and cheaper than human creations. But unlike those early factories, AI advertising systems don’t only produce. They are integrated into distribution and optimization systems that no stand-alone ad agency can match. This soup-to-nuts integration is immensely valuable. Smaller companies—like restaurants, shops and mom-and-pop businesses—can afford to produce ads they couldn’t before. Large advertisers often need smaller ad teams and can run multiple new ad campaigns. A majority of Google’s advertisers [reportedly](https://mycodelesswebsite.com/performance-max-statistics/) use Performance Max. But this integration has a cost. By automating ad creation at scale, these platforms are displacing creative workers and agencies. [WPP](https://www.wsj.com/market-data/quotes/UK/XLON/WPP), the world’s largest advertising holding company, has begun [using AI](https://www.freelanceinformer.com/news/wpp-slashes-freelancer-workforce-by-25-as-ai-agents-replace-external-talent-and-sector-looks-to-consolidate/) for many tasks and eliminated 25% of its freelancers. Actors, photographers, videographers, artists, animators, voice actors, sound designers, set designers, copywriters, editors and translators all stand to lose their work. Emily Blunt says she is terrified of Tilly Norwood, the [digital “actress”](https://variety.com/2025/film/global/ai-actress-tilly-norwood-talent-agents-zurich-summit-1236533454/) that has the cheekbones of Natalie Portman and the presence of Scarlett Johansson. SAG-Aftra [condemns](https://www.npr.org/2025/09/30/nx-s1-5558119/tily-norwood-ai-hollywood) her creation. But Tilly is far less a threat to Hollywood than actors who earn a living in car commercials and insurance ads. Her natural home is TikTok, which last year [introduced](https://www.nytimes.com/2024/06/17/style/tiktok-ads-ai-influencers.html) AI-generated avatars, free for businesses that spend $20 to thousands of dollars a day on TikTok ad campaigns. Actor Scott Jacqmein [sold](https://www.nytimes.com/2025/08/17/business/tiktok-ai-avatars.html) his likeness to TikTok for $750 and receives no royalty when his avatar appears in ads. Similarly, the demand for real pictures is falling. Photographers [struggle](https://www.wsj.com/articles/the-last-stock-photographers-await-their-fate-under-generative-ai-822d1e6a?mod=article_inline), selling fewer pictures and earning far less in royalties when they do. Getty and [Shutterstock](https://www.wsj.com/market-data/quotes/SSTK), which announced a [merger](https://newsroom.gettyimages.com/en/getty-images/getty-images-and-shutterstock-to-merge-creating-a-premier-visual-content-company) this year, used to be the leaders in stock photography. Now they sell AI-created images with the guarantee that these don’t infringe copyrights. [Adobe](https://www.wsj.com/market-data/quotes/ADBE) Firefly creates “commercially safe” AI generated images, video, audio and vector graphics. Companies like Adobe, Midjourney and Eleven Labs generate billions of digital “assets” that can be fed into systems that create, test and display ads. The economics of integrated AI advertising systems are compelling. These systems are still in their infancy. As they improve, they will consolidate and automate more advertising work. Google’s Veo and OpenAI’s Sora will no doubt be developed into full-fledged systems for creating video ads for pennies. The biggest winners will be companies like Google, Meta and Amazon, which can serve as one-stop shops for advertisers. Individually, creative professionals are unlikely to stop the AI engine that is creating advertising wealth. The companies that benefit most from these systems need to integrate creative workers into their workforce. Regulators need to help level the playing field so that creative professionals, whose work trains these systems, can get a stake in the wealth they produce. Critics worry that spending hundreds of billions on AI is a gamble on superintelligence that may never arrive. But Google, Meta, and Amazon aren’t making only this bet. They are methodically consolidating the trillion-dollar advertising industry—and displacing the creative workers who once powered it. *Mr. Kohli is a professor at Columbia Business School.* Copyright ©2025 Dow Jones & Company, Inc. All Rights Reserved. 87990cbe856818d5eddac44c7b1cdeb8 Appeared in the December 16, 2025, print edition as 'AI Is About to Empty Madison Avenue'.