What computing power demands differentiate AI agents from search engines or chatbots?

Question

Accepted Answer

Microsoft says AI agents need far more computing power than search engines or chatbots
Microsoft says running its new AI agent, Copilot Cowork, demands vastly more computing power than a search engine or a chatbot, and that the amount of computing power used can vary widely from user to user. The pricing shift to pay-as-you-go is directly tied to these higher and more variable compute costs, with each task billed separately based on computing power consumed.
Copilot Cowork runs tasks independently, which drives higher compute use
Copilot Cowork is described as an “agent” that can independently carry out office tasks such as drafting documents, building spreadsheets and sending emails. It can also be given an assignment and run with it on its own, sometimes for several hours, which Microsoft links to the need for far more computing power than a typical search or chatbot interaction.
AI agent usage varies widely across users and tasks
Microsoft says different users consume widely varying levels of computing power, which is part of why it moved away from a single fixed subscription structure. Because task workloads can differ, the compute footprint of running an agent can also differ, making usage-based billing more aligned with actual resource consumption.
The company ties the pricing change to compute cost differences
Microsoft’s executive vice president for Copilot and agents, Charles Lamanna, said the shift is prompted by the soaring cost of artificial intelligence and that “the only way to make the model work” is a structure that matches those costs. He compared the new approach to “filling up your gas tank at the pump,” reflecting that computing power use is measured per task rather than covered by one overarching license.
The comparison is stated at a high level, without technical breakdown
The information provided draws a clear contrast in computing power demand at the level of products: AI agents versus search engines or chatbots. It does not specify the exact technical differences in underlying model architectures or processing steps that create the disparity, but it does state that agents require “vastly more computing power” and that usage can vary widely.

Reading history

Hide →

This history is cleared when the page is closed.

What computing power demands differentiate AI agents from search engines or chatbots?

Loading

Was this article helpful?

Helpful

Not helpful

Report an issue

1

0

0

Share with friends