AI-driven search results could come with a price from Google

The cost of AI services may push leaders in the sector to use subscription services

According to reports, Google intends to charge for AI-enhanced search functionality, which would be a major departure from its current income model. Given the high expenses of delivering the service, this change is deemed necessary. As a result, experts predict that in order to control these expenses, all of the industry’s key companies will eventually implement a subscription model.

Google’s plans, as reported by the Financial Times, would limit access to its new search function to subscribers to its premium plans. Utilizing artificial intelligence assistants in other Google products, such as Gmail and its office suite, already requires these services.

Similar to ChatGPT and other rivals, Google is presently beta-testing a search interface that leverages its generative AI to deliver direct answers to searches in a conversational manner.

“AI search is more computationally expensive than Google’s traditional search methods,” said Heather Dawe, chief data scientist at UST. Therefore, Google hopes to recoup at least some of these expenses by charging for AI search.”

The spotlight in AI often falls on the significant costs associated with the computing power required to train advanced generative models. Engineer James Hamilton revealed that Amazon conducted a single training run costing $65 million (£51 million) last year, and he anticipates the company will soon exceed the $1 billion mark.

Recently, OpenAI and Microsoft unveiled intentions to construct a $100 billion data center for AI training. Additionally, in January, Mark Zuckerberg stated his aim to invest at least $9 billion solely in Nvidia GPUs.

The expense of training AI, however, only makes up a tenth of the industry’s overall costs, according to analyst Brent Thill of the investment company Jefferies. Thill stated in a briefing note that “The majority of AI compute spend today is directed to the running, not training, of models, and 90%+ of AI compute spend today is being directed towards inferencing [the process by which an AI model is queried], as inferencing spend has been growing much faster than training as more models and tools get put into production.”

He went on to say, “Some businesses have started charging more for new Generation AI features, with the goal of offsetting consumption expenses. Some have chosen per-usage charging as a way to protect themselves from price swings. Some have also included these elements into their current plans in an effort to encourage user expansion.”

Rivals in the AI search field provide comparable subscription options. Perplexity, an AI-driven search engine, doesn’t display advertisements but offers a $20 monthly “pro” level that grants access to more advanced AI models and unlimited usage.

However, some competitors still offer their services at a loss. The AI capabilities in Microsoft’s Bing are free to use but require the company’s Edge browser. Arc, a browsing and search startup, offers its services to users for free and plans to generate revenue in the future by charging companies for business-related features.