A surprisingly strong quarterly report from Alphabet (GOOGL) has brought not relief but a significantly larger bill.
According to a research report from Barclays, following the release of Alphabet's Q2 2026 earnings, the bank has dramatically raised its capital expenditure forecast for the company to $350 billion for 2027 and further to $500 billion for 2028. These figures exceed 100% of projected operating cash flow for the same periods, indicating that Alphabet is expected to be in a state of negative free cash flow for the next two consecutive years.
Core Driver of the Spending Surge
The primary driver behind this surge in capital expenditure is Alphabet's effort to build up computing power reserves for the large-scale training of its Gemini 4 model. Barclays believes Alphabet is currently renting external GPU capacity at a market premium to provide "transitional" computing support for some Google Cloud Platform (GCP) clients, thereby freeing up its own in-house capacity for Gemini 4 training. This arrangement is estimated to cost around $1 billion per month and is expected to pressure cloud business margins in Q4 and Q1 2027.
Latest earnings showed Google Cloud revenue grew 82% year-over-year in Q2, with cloud backlog increasing by $50 billion in the quarter to $514 billion. Search advertising revenue grew 17%, and YouTube ad growth accelerated to 13%. However, during the earnings call, the dual expansion of operational costs and capital expenditures driven by the full-scale integration of AI emerged as a central focus for the market. The CFO indicated that AI infrastructure investments will continue to pressure the income statement and cash flow.
Cloud Growth Accelerates Amid Cost Pressures
Barclays identified the standout highlight of the quarter as the continued acceleration of Google Cloud. In Q2 2026, Google Cloud revenue reached $24.768 billion, an 82% year-over-year increase, with the cloud business operating margin improving to 36%. The massive $50 billion quarterly increase in cloud backlog demonstrates strong enterprise demand for GCP.
However, this rapid growth in the cloud business is accompanied by significant cost increases. The report notes that Alphabet has entered into a GPU leasing agreement costing approximately $1 billion per month to provide transitional capacity for some GCP customers, expected to last about two quarters. Barclays views this arrangement as essentially a move to free up Alphabet's own first-party computing capacity for Gemini 4 training.
The premium rent is considered "eye-catching" by Barclays, but given the current urgency of compute supply and demand and the temporary nature of the arrangement, the decision is seen as somewhat justified. The report also suggests this setup could theoretically help Alphabet establish a foundation for future load capacity in its orbital data center racks.
Gemini 4 Training Fuels "Explosive" Capex Growth
Barclays explicitly linked its significant capex forecast revision directly to the training demands of the Gemini 4 model. The bank raised its 2027 capex expectation from $256.1 billion to $350.2 billion, a 37% increase, and its 2028 capex expectation from $327.4 billion to $500.3 billion, a roughly 53% increase.
Based on Barclays' projections, Alphabet's capital expenditures from 2026 to 2028 are forecast to be approximately $200.5 billion, $350.2 billion, and $500.3 billion, respectively. This would result in capex exceeding operating cash flow for two consecutive years, with free cash flow turning negative by $6.5 billion and $135.2 billion in those years.
The report also noted that Alphabet's management expressed clear confidence in the development pace of Gemini 4 and its overall AI models during the earnings call. Current Gemini API traffic has reached 22 billion tokens per minute, up significantly from 16 billion last quarter and representing a 3-4x year-over-year increase. Barclays estimates its market position is now tied for second with Anthropic.
Full-Stack AI Strategy Builds Moat at High Cost
Barclays characterizes Alphabet as "the company at the highest global AI frontier," with a comprehensive full-stack AI strategy spanning the application layer, model layer, infrastructure layer, and raw compute layer. This breadth is expected to become more pronounced as cloud business acceleration continues.
However, the cost of competing on all fronts is substantial. The report points out that Alphabet faces competition from Anthropic and OpenAI at the model layer, pressure from AWS and Azure at the infrastructure layer, while also maintaining operations for 13 products with over 1 billion monthly active users each. This "multi-front war" is the fundamental reason behind the simultaneous expansion of operating expenses and capital expenditures.
Regarding profit forecasts, Barclays raised its 2027 revenue and operating income estimates by 5%, mainly reflecting upward revisions for Google Cloud. However, due to the impact of the GPU leasing agreement, EPS estimates for Q4 2026 and Q1 2027 were lowered, while estimates for subsequent periods were raised.
Barclays increased its price target for Alphabet from $405 to $425. The valuation is based on an average of a 25x P/E multiple on 2027-2028 estimated EPS and a 15x multiple on estimated EBITDA. The bank maintains an Overweight rating.
From a longer-term perspective, Barclays believes that in terms of capturing value across the full AI stack, Alphabet currently has few comparable rivals. The strategic value of its cross-layer positioning is expected to be gradually realized as AI demand continues to expand.
Comments