Google struggling to make AI Mode searches pay

Shares in Google's parent company, Alphabet, dropped nearly ten percent on Wednesday after the tech giant admitted that AI-powered searches generate less revenue than traditional search engine ads. Alphabet reported its Q3 results, including news that its year-over-year growth in search queries is the highest in the past five years.

This growth includes users adopting Google's "AI Mode" – the company's latest system powered by large language models (LLMs). Google's search revenue rose 12 percent year-over-year to $63.25 billion, slightly below the 14.5 percent growth reported in the previous quarter. During Alphabet's earnings call, chief business officer Philipp Schindler explained the slowdown was due to "lower monetization we were seeing in AI Mode."

This slowdown is concerning as Google is investing heavily in infrastructure. In Q3, $51.4 billion was spent on servers and networking equipment, with the rest funding data center builds. This investment seems to have affected profitability, as Alphabet reported $26.17 billion in net income for the quarter. In contrast, last year's Q3 generated $35 billion in net income from Alphabet's $102.35 billion in revenue.

Schindler expressed optimism about AI search, stating that many AI Mode queries are "upper funnel," allowing the company to capture new, high-value commercial intent. Additionally, he claimed that AI Mode queries are creating more monetization opportunities over time. Google attempted to address the lack of AI Mode revenue by enabling ads alongside these searches worldwide, after initially testing them in the US.

However, AI Mode searches still lag behind Google's existing search business, which accounts for just under half of Alphabet's $125.8 billion quarterly revenue. Google's cloud division, the company's second-largest earner, generated $24.75 billion during the quarter. However, Google faces challenges in this area as it doesn't have sufficient capacity to serve all its customers.

The company has started offloading some customer workloads to third parties, but this move is expected to reduce future margins. Thomas Kurian, Google Cloud's CEO, revealed that over 100 companies utilize Google's Tensor Processing Unit (TPU) silicon, with all TPU system inventory associated with committed contracts. This means Google can generate some revenue from these capex investments, and the company anticipates TPU sales to third parties in upcoming quarters.

Google CEO Sundar Pichai remains optimistic about future AI revenue. He highlighted that Google's Gemini app has attracted a billion users faster than any other product in the company's history and praised the performance of its models. Pichai emphasized that Google is moving swiftly to develop agentic AI tools, but will initially focus on business customers due to the high security, safety, and governance standards required.

He added that Google intends to bring these features to Search and its billion users through the Gemini app. While Pichai did not disclose when agentic AI will be available for search or if it will avoid the monetization issues experienced with AI Mode, the company has maintained its $195-$205 billion capital expenditure forecast for the year and plans to spend "significantly" more in 2027.