- Cloud Growth: Google Cloud’s revenue has grown 82% year over year in the April-June period.
- Demand Signals: Customer acquisition more than doubled, while Gemini processing reached about 22 billion tokens per minute.
- Infrastructure Bill: Alphabet spent $44.9 billion in the quarter and raised its annual capital-expenditure range to $195 billion-$205 billion.
- Capacity Test: Supply constraints slow backlog conversion, while Alphabet’s future free cash flow will test whether spending can keep pace.
Google Cloud more than doubled its rate of new-customer acquisition year over year, while existing customers exceeded their commitments by more than 50%. Its revenue grew 82% year over year in the April-June period. Google parent Alphabet faced higher AI infrastructure costs at the same time.
During the April-June period, Google Cloud generated $24.8 billion in quarterly revenue and $8.8 billion in operating income. Its $514 billion backlog represents contracted business not yet recognized as revenue.
Google remained supply constrained even as demand increased. Signed contracts do not become recognized revenue until Google can provide the purchased cloud capacity.
Alphabet raised its annual capital-expenditure range to $195 billion-$205 billion, up $15 billion per end. Capital expenditure covers spending on data centers, chips, networking, power systems, and related compute capacity.
Demand Is Rising Across the Cloud Stack
Google Cloud’s revenue growth accelerated by nearly 20 percentage points from the previous quarter. AI services have contributed to the increase substantially.
AI chip sales added another growth source. Hardware sales and cloud rentals are separate revenue streams that draw on the same computing ecosystem. Chips sold generate hardware revenue, while cloud workload result in service revenue as customers consume computing capacity.
Google parent Alphabet now oversees a cloud business far larger than its second-quarter operation in 2023, when the company was already heavily investing in AI. T
In the last quarter, existing customers exceeded their commitments by more than 50%. New contracts expand future business, while commitment overages show that current customers are consuming more than they reserved and may require capacity sooner. The customer measures describe acquisition and consumption rather than revenue booked during the quarter.
Google says that Gemini services processed about 22 billion tokens per minute, up from 16 billion in the preceding three-month period. Tokens are the units AI models process when handling prompts and generating responses.
Google is also turning its Tensor Processing Unit (TPU) AI chips into a broader commercial business. Its push to sell TPU systems beyond Google Cloud could add hardware customers without requiring them to move every workload onto the platform. Anthropic’s recently expanded TPU commitment is such an example of outside demand for Google’s custom accelerators and the computing power those customers require.
Capacity Turns Growth Into a Cash Test
Alphabet spent $44.9 billion on capital expenditure during the quarter. That spending can cover chips, servers, networking, land, buildings, cooling, and power systems needed for future AI workloads. Quarterly expenditure records money deployed over three months, while annual guidance describes the expected 2026 spending envelope rather than cash already spent.
Google’s earlier compute-capacity constraints illustrate how installing servers does not create usable capacity by itself. Earlier in 2026, Alphabet’s AI infrastructure spending plan treated servers, data centers, networking, and power as linked parts of the buildout. Chips alone cannot turn signed cloud demand into usable customer capacity, when the rest of the required infrastructure has not been made available yet.
Alphabet’s free cash flow was negative for the quarter. Free cash flow is the cash left after operating and capital costs, while segment operating income measures only Google Cloud’s operations. Heavy AI construction can weigh on parent-company cash without turning the cloud unit’s positive operating income into a loss.
Supply limitations will set the pace at which Google Cloud’s backlog becomes revenue. Alphabet’s next quarterly free-cash-flow figure will test whether cash generation is catching up with the buildout.
