Alphabet’s second-quarter earnings release on Wednesday (July 22) provided a clear answer to one of the central questions surrounding generative AI. It’s about whether companies are moving from testing technology to paying at scale fast enough.
For Google, there is growing evidence from enterprise customers that demand for compute power, Gemini models, and AI applications contributed to an 82% increase in Google Cloud revenue and pushed its cloud backlog to $514 billion.
The scale of that demand has strengthened executives’ view of the AI opportunity over the past year. During an analyst Q&A, CEO Sundar Pichai was asked whether Alphabet’s expectations for AI-generated profits had changed, and CEO Sundar Pichai said they had. He specifically pointed to conversations with business executives, saying many companies are “barely beginning the early stages of what they can do here.”
Management has been discussing supply constraints for several quarters. CFO Anat Ashkenazi acknowledged that demand continues to outpace the capacity Alphabet has added.
As a result, Google Cloud Platform (GCP) grew faster than the cloud as a whole, with Google Cloud revenue reaching $24.8 billion. AI infrastructure, core GCP products, and AI solutions all contributed. Cloud operating profit more than tripled to $8.8 billion.
Search remains the clearest test of whether Google can deploy generative AI as a tailwind to its core business. Search and other revenue grew 17% to $63.3 billion, with AI Overview and AI Mode increasingly working as part of a single search experience. Pichai said Google is seeing an increase in the total number of queries as users ask longer and more complex questions and people use AI capabilities for searches they might not have done before. The company is also extending its AI mode to more commercial queries, bringing products closer to shopping and purchasing decisions.
Gemini expands into search, advertising and commerce
Gemini is becoming a bigger part of the company’s proposition. Almost 90% of Fortune 100 companies use Gemini Enterprise. According to Alphabet, over the past year, approximately 500 cloud customers have each processed more than 1 trillion tokens, and more than 2,000 companies have consumed more than 100 billion tokens.
Customers are deploying Gemini within cybersecurity, data analytics, and other applications, with the model operating as one component of a broader system. “Models are just one element of these solutions,” Pichai told analysts.
The same strategy is playing out in Google’s consumer business, particularly at the intersection of search, advertising and shopping.
Retail made the largest contribution to search and other advertising revenue, with financials also contributing significantly. YouTube’s ad revenue increased 13% to $11.1 billion, led by direct response and branded ads. Overall ad revenue increased 14%.
The company also uses Gemini across its advertising systems, including query interpretation, advertiser tools, and advertising with an AI-powered search experience. According to the company, Gemini has improved the visibility of relevant shopping ads by 20%. More than half of Google’s small business advertising customers now use AI to create or optimize their ad creative.
Commerce is moving closer to these search and advertising products. Target and Steve Madden are now running on the open source Universal Commerce Protocol, executives said on the conference call. We also announced Universal Cart, which allows consumers to add items from multiple retailers to a single cart across Google services and complete a single checkout.
YouTube is developing a similar relationship between content, advertising, and transactions. Google has expanded its shoppable ad formats and introduced Buy with Google Pay. This allows connected TV viewers to complete purchases directly from their TV in just two clicks. We also use affiliate partnerships and YouTube Shopping commissions to connect creators more directly with sales.
Gemini itself is also accumulating consumer scale. The Gemini app has 950 million monthly active users, and daily active users have tripled over the past year. Alphabet’s model application programming interface (API) processes approximately 22 billion tokens per minute, up from more than 16 billion a quarter ago. More than 9 million developers build with Google models every month.
Overall, Alphabet’s revenue rose 24% to $119.8 billion, and operating profit rose 30% to $40.8 billion.
The cost of meeting AI demand remains part of the equation. Capital spending reached $44.9 billion in the quarter, and Alphabet raised its 2026 capital spending forecast from $180 billion to $190 billion to $195 billion to $205 billion. Management said the increase reflects faster deployment of computing power, but that third-party capabilities will temporarily supplement Google’s infrastructure. Investors likely took note of the expansion, sending the stock down 3.5% in after-hours trading.
Asked about the return on additional computing investments in 2027, Pichai cited long-term customer contracts, renewals and continued demand.
“We are seeing strong demand indicators, including long-term contracts,” he said. “If anything, the dynamics look healthier than they did about a year ago. That’s giving us the confidence to take on these investments.”
