AINEWS
Google's parent beat expectations with 24% revenue growth and 82% Cloud growth, but rising AI infrastructure spending is the test for margins.
Alphabet reported its strongest quarter in years on Wednesday: $119.8 billion in revenue, up 24% year-on-year, with diluted earnings per share of $9.11. Google Cloud, the unit that sells AI infrastructure and software to enterprise customers, grew 82% to $24.77 billion. The print easily beat consensus. The stock still fell.
The single number that explains both halves of that sentence is the new Google Cloud backlog: $514 billion of contracted future revenue, not yet recognized. That figure is the AI bet, in dollars. The question on Wednesday's earnings call was no longer whether the demand exists. It is how much Alphabet has to spend to feed it.
Backlog is the load-bearing term here. When a customer signs a multi-year Google Cloud contract, the deal shows up in backlog until Google delivers the service and recognizes the revenue. A growing backlog means future quarters are already partly sold. Alphabet's $514 billion figure, disclosed in the Q2 earnings release, is roughly a year of total Google Cloud revenue forward-booked at current run-rate.
On the same call, the company described how that demand is being absorbed. Gemini Enterprise, Google's family of AI products for businesses, is now used by nearly 90% of the Fortune 100, with named customers including PepsiCo, Intel, HSBC, and Macy's. The Gemini app, Google's consumer-facing chatbot, passed 950 million monthly active users, and daily active users tripled over the past year. Roughly 9 million developers build on Google AI models each month, and the company's AI APIs process about 22 billion tokens per minute, up from 16 billion last quarter. A next-generation Gemini 4 model is in development.
The bear case writes itself: capital expenditure is climbing fast. AI infrastructure requires chips, data centers, and power, and Alphabet is signing long-term commitments on all three. If next quarter's capex guidance rises again, and the backlog number flattens or contracts, the "demand is signed for" framing collapses. That is the live tension: not whether AI is paying off, but whether the cost of feeding the build-out stays inside the margin.
The bull case is on the same page. The Cloud backlog and 82% growth show that enterprise customers are committing to multi-year AI spend before the underlying models are fully built. That is unusual. In most software cycles, customers wait for product maturity before signing long contracts. The fact that the Fortune 100 is doing it now suggests the buyers have decided AI infrastructure is a 2026 procurement decision, not a 2028 one.
The Google blog framed the quarter as a transition: "We are at the beginning of a long cycle of AI investment." That phrase is now the watch item. If capex guidance climbs again in Q3, the cycle is not yet long. If it holds or moderates, the market will start pricing the backlog as visible revenue rather than a deferred promise.
For the next quarter, three numbers will matter more than the headline beat. First, Cloud backlog trajectory. A second consecutive jump above $500 billion would lock in the demand story. Second, capex guidance, which will be set when Alphabet reports Q3. Third, Gemini Enterprise seat growth, both the Fortune 100 percentage and the named accounts that follow PepsiCo and Intel.
Alphabet's Q2 print is not a contradiction. It is a snapshot of a company whose AI product is selling faster than its AI capacity can be built. The market's job on Wednesday was to price the gap between the two. Investors who believe the backlog will translate into recognized revenue are buying the dip. Those who think the capex bar will keep rising are selling it. The next earnings call will say who was right.