Google plans to insert ads in its Gemini multimodal model, following the recent addition of advertisements in the AI Overviews portion of its search engine as it seeks to recoup the high cost of processing AI workloads.
“We do have very good ideas for native ad concepts,” said Google CEO Sundar Pichai, in a call with analysts to discuss fourth-quarter and year-end earnings on Tuesday (Feb. 4). “But you will see us lead with the user experience” and first making sure it works at scale for billions of people.
Gemini ads will not come in 2025. Google said it would focus on offering a free and paid version of Gemini, which competes with ChatGPT.
Google already began inserting ads in its AI Overviews, or AI-powered summarized answers to user search queries, since October.
Google CBO Philipp Schindler said that thus far, AI Overview ads are monetizing at “approximately the same rate” as ads in traditional Google Search instead of cannibalizing revenue.
Schindler added that this bodes well for Google to experiment and “innovate even more.”
Google’s core business, search, is under fire after the tech giant lost an antitrust court case last year accusing it of monopolizing online search and advertising markets by entering into exclusive partnerships. The remedy phase of the lawsuit is yet to be finalized.
Morningstar analyst Malik Ahmed Khan wrote in a research note that Google parent Alphabet’s priority is to preserve its strong search ad business — and using AI to keep it in the lead.
“We see the firm’s investments in AI as a continuation of this effort to safeguard its core product, Google Search,” he wrote. “We believe that by leveraging generative AI, Google can not only improve its own search quality via features such as AI Overviews, but also improve its advertising business by augmenting its ability to target customers with relevant ads.”
Khan said the majority of Alphabet’s revenue comes from search. Over the years, Google has invested “considerably” to improve its search capabilities, to ensure that its search engine remains people’s preferred choice.
Google Weighs In on DeepSeek
Pichai said that DeepSeek proves something the company has known for a long time — that costs, latency and performance will continue to improve.
The CEO pointed to the recently released Gemini 2.0 Flash and Flash Thinking models as “some of the most efficient out there, including compared to DeepSeek’s V3 and R1” models.
Pichai said it was possible for Google to continue to increase efficiency while improving performance due to its “full stack” approach to AI — encompassing AI infrastructure, research and applications and platforms — focus on optimization, and “obsession” with cost per query.
Also, the proportion of AI workloads has been shifting from training frontier or cutting-edge foundation models to inferencing when these models are actually used. Typically, training costs are higher than inferencing costs, although the latter can add up over time.
Frontier model developers like Google pay for training costs; clients of Google Cloud pay for inferencing costs.
“That trend is good,” Pichai said. The recent trends towards reasoning, or thinking, AI models is good because these use inferencing.
Asked if agentic AI will affect Google search, Pichai said it will only expand the market. “Plenty of room for many new types of use cases to flourish.”
By the Numbers
Alphabet reported net income of $26.5 billion, or $2.15 per share, in the fourth quarter ending Dec. 31, beating the consensus estimate of $2.12 per share. Revenue came to $96.5 billion, slightly below consensus of $96.7 billion, as compiled by Zacks Investment Research.
Quarterly revenue in the quarter grew by 12%, which was its slowest pace since 2023. Investors did not like the news, selling off shares by 7% in after-hours trading on Tuesday.
Alphabet expects to spend $75 billion in capex this year, mainly towards data centers, servers and networking infrastructure.
Khan continues to be bullish on Alphabet.
“On the antitrust front, we don’t foresee a material deterioration in Google’s search business resulting from governmental or judicial intervention,” he wrote. “While there is a range of possible outcomes depending on what remedial steps are imposed, we think it is likely that Google will maintain its leadership position in search and text-based advertising in the long term.”