Driven by strong growth in its cloud business and YouTube, Alphabet's total annual revenue surpassed $4000 billion for the first time, representing a year-on-year increase of 15%. In addition to record revenue, Google CEO Sundar Pichai revealed in a conference call that the monthly active users (MAU) of its AI assistant, Gemini, have officially exceeded 7.5 million, indicating that Google's counterattack in the AI arena has been effective since the launch of the Gemini 3 model.
Gemini 3's performance is a huge boost, with user numbers rapidly approaching those of ChatGPT.
According to financial data, the number of monthly active users of the Gemini App surged from 650 million in the previous quarter to 750 million in just one quarter, mainly thanks to the Gemini 3 model launched last November.
Sundar Pichai described the Gemini 3 as the "fastest adopted" model in Google's history. In comparison, Meta AI currently has about 5 million monthly active users, while ChatGPT's estimated figure for the end of 2025 is between 8.1 million and 9 million, meaning the user gap between Google and OpenAI is rapidly narrowing.
Furthermore, Google's API usage is astounding, currently processing over 100 billion tokens per minute. Sundar Pichai also mentioned a future collaboration with Apple to drive a new version of the Siri digital assistant service using a customized version of the Gemini 3 model, which is sure to drive another surge in usage.
YouTube and Google Cloud become new cash cows
In addition to AI, traditional businesses also performed well:
• Youtube:YouTube's annual revenue (advertising + subscriptions) has surpassed $600 billion. According to Nielsen market research data, YouTube continues to hold its position as the leading streaming media platform.
• Google Cloud:The annual revenue run rate reached $700 billion.
• Subscription services:Google One and YouTube Premium together have more than 3.25 million paid subscribers.
Capital expenditures to double: $1800 billion to be spent on AI by 2026
While Google earns a lot, it doesn't hold back on spending. To support the massive computing demands of its AI, Alphabet expects its capital expenditures in 2026 to fall between $1750 billion and $1850 billion, almost double its spending in 2025.
This huge sum will primarily be used to expand its chip inventory (including its self-developed Ironwood TPU and NVIDIA GPUs) and continue building data centers. Despite the impressive financial figures, this aggressive spending plan has raised market concerns, causing Alphabet's stock price to dip slightly by 2% after the earnings release.
Analysis of viewpoints
The financial report shows that Google's AI strategy has shifted from "defense" to "offense" and has begun to generate revenue.
In the past, the market worried that AI would erode Google's search advertising revenue, but data shows that Google has successfully stabilized its position through "hybrid monetization" (Google Cloud enterprise services + Google One/YouTube personal subscriptions + search ads). In particular, the success of Gemini 3 proves that Google still has a strong foundation in underlying algorithm technology and is not being constantly overwhelmed by OpenAI.
It's worth noting that Google has doubled its capital expenditures and continues to purchase and expand its chip computing power. With Microsoft, Meta, and Amazon all aggressively expanding their AI infrastructure, Google's decision to follow suit signifies that the AI arms race has entered a phase of "competing on the depth of one's resources," where whoever possesses the most computing power will be able to train the next generation of models.
While Google spends heavily, it also has its own self-developed TPU chips (such as the latest Ironwood), which may give it an advantage in cost control over competitors who rely solely on NVIDIA.
The next key focus will be on the development of "Agentic AI." Google's announcement of adding a "checkout function" to Gemini signifies that AI will no longer be just a chatbot, but a super assistant that can truly help you "complete tasks," which is the next holy grail for AI commercialization.
