Big Tech's AI agents: who will pay for the costly computing

what happened

Technology companies are heavily promoting always-on AI agents that carry out tasks for users in the background. According to The Wall Street Journal, Meta Platforms, OpenAI, SpaceX and the startup Instinct have already released such products, but they require huge computing resources, and without lowering those costs it is hard to earn money from them.

To contain expenses, some services are available only to paying subscribers or limited to a region. Meta initially launched its free Muse agent only in the United States, while Google had to build a new platform for its Gemini agent. Mark Zuckerberg said the company expects to improve Muse's cost efficiency by more than tenfold over the next year.

A Deutsche Bank analyst note estimates the capital cost of each user's agent at $25 to $50. Elon Musk said the cost of serving Grok Bot was unsustainably high relative to revenue.

why it matters

For the market, the key question is whether the industry can turn AI agents into a profitable business. Demand is growing faster than capacity, and after voters backed data center moratoriums in several states, building new facilities is getting harder, so investors are watching how efficiently existing chips are used.

market impact

Large tech companies: higher computing costs pressure the economics of AI agents

instruments

on polymarket