Google now displays an AI assistant’s answer to a consumer question at the very top of its search results, above the paid links. The placement is designed to make Google the go-to AI assistant ahead of rivals such as ChatGPT, yet it carries a trade-off: a strong answer can mean fewer clicks on Google’s own ads. Alphabet (GOOGL、GOOG) earns revenue from search and YouTube advertising and cloud computing; rival Meta Platforms (META) sells digital ads across Facebook, Instagram and WhatsApp.
AGI refers to artificial intelligence capable of matching a skilled worker at most jobs. As machines approach that level, both companies must answer the same question: whether either advertising business holds up once AI agents begin making choices for people.
Both companies rely on digital ad auctions, but differing advertiser goals leave a gap in how well revenue withstands AI. Alphabet advertisers pay to reach people actively searching for something; an AI agent booking a dinner table can handle the whole task without opening a results page or viewing the ads on it.
Meta advertisers pay to reach people scrolling Facebook and Instagram or chatting on WhatsApp. Users open those apps to check on friends, watch videos and pass the time, and no one hands that activity off to an AI agent, so Meta’s ads keep reaching real people. About 98% of Meta’s 2025 revenue came from advertising. More than 3.5 billion people used Meta’s apps each day as of late 2025, an audience that gives the company genuine pricing power; in the second quarter of 2026, the average price per ad across its apps rose 12% year over year.
Power and memory supplies remain tight, and both companies are building massive data centers and custom chips. Alphabet edges ahead on who uses its silicon. Meta focuses largely on internal needs, building giant computing clusters such as Prometheus and Hyperion and designing custom Iris chips mainly for its own apps. Outside customers rent Alphabet’s custom TPU chips along with accompanying software, and nine of the top 10 AI labs use Google Cloud, which brought in $20 billion of revenue in the first quarter of 2026.
Alphabet is extending its chip business beyond its own data centers: in the second quarter of 2026 it began shipping TPU systems directly into customer facilities and teamed with investment giant Blackstone on a TPU-based cloud venture. A customer moving AI work to other chips must first retune its software, a switching cost investors flag; high switching costs make customers sticky, giving Alphabet’s chip business a moat against rivals.
Meta looks better built for AGI because its ads reach people in the act of keeping up with friends and family, an activity users want to handle themselves, leaving no AI agent between them and the ads. Alphabet, by contrast, depends on searches AI assistants can answer and act on independently. The case for Alphabet rests on cloud computing: if outside developers paying for TPU and cloud capacity cannot secure enough rival chips and power, cloud gains could outweigh the pressure on search advertising.
