The Soaring Cost of Intelligence
Artificial intelligence runs on two things: data and immense computational power. Training and running advanced models like Google's Gemini requires building and operating massive, energy-hungry data centers filled with specialized servers. This hardware
has a limited lifespan and needs constant investment. As a result, Alphabet, Google's parent company, is in the midst of an unprecedented spending surge. For 2026, the company has guided capital expenditures (CapEx) to a jaw-dropping range of $180 billion to $190 billion, nearly double what it spent in 2025. This firehose of spending is a massive bet on future growth, but in the short term, it raises a critical question for investors: can the revenue from AI grow fast enough to offset these costs and protect the company's famously healthy profit margins?
Clue #1: Follow the Capital Expenditures (CapEx)
The most direct signal of AI-related spending is found in the Capital Expenditures, or CapEx, line on the cash flow statement. This number represents spending on long-term physical assets, and for Alphabet, it has become the primary indicator of its AI infrastructure build-out. CFOs and savvy investors watch the CapEx guidance that management provides during earnings calls like a hawk. When Alphabet raises its full-year CapEx forecast, as it has been doing, it signals immense confidence in future AI demand. However, it also means the costs are hitting the books now, long before the corresponding revenue fully materializes. This spending directly impacts free cash flow—the cash a company generates after accounting for these large capital outlays—which is a key metric of financial health. In the first quarter of 2026, Alphabet's free cash flow fell significantly year-over-year, absorbing the record AI spending.
Clue #2: Scrutinize 'Cost of Revenue'
While CapEx shows the investment in building the AI factory, the 'Cost of Revenue' on the income statement reveals the cost of running it. This line item includes expenses directly tied to generating revenue, such as data center operations, energy consumption, and the depreciation of all that new server hardware. As billions in CapEx from previous years turn into active data centers, the depreciation expense starts to climb, pushing the cost of revenue higher. Alphabet doesn't break out AI-specific operational costs, so analysts look at the growth rate of the total Cost of Revenue relative to revenue growth. If costs are growing faster than sales, it's a clear sign of margin pressure. This is where the efficiency of its custom TPU chips and the profitability of AI-powered searches become critical variables in the long-term financial picture.
The Tell: Watch Google Cloud's Performance
The clearest signal of AI monetization is Google Cloud. This segment provides the strongest evidence that AI spending is producing direct commercial returns. Unlike consumer AI features where the return on investment can be murky, enterprise customers pay directly for AI models, tools, and computing capacity. In the first quarter of 2026, Google Cloud revenue soared 63% to $20 billion, with its operating margin expanding to nearly 33%. Analysts watch Cloud's revenue growth and, just as importantly, its backlog—the value of signed contracts to be fulfilled in the future. A rapidly growing backlog, like the one Google has reported, suggests that its massive CapEx spending isn't just speculative; it's being done to meet concrete, contracted demand from customers like Anthropic, which has reportedly committed to spending billions on Google's cloud infrastructure.
