Alphabet Lifts Capital Spending Plan to $205 Billion, Stock Slides

Alphabet has attached a bigger invoice to the artificial intelligence boom, and shareholders flinched at the total. On its July 22 earnings call, the company raised its 2026 capital spending guidance to a range of $195 billion to $205 billion, up from $180 billion to $190 billion previously and well above the roughly $186 billion analysts had modeled, according to CNBC. Shares fell about 5 percent in extended trading despite a second quarter that beat expectations on nearly every line.

The results themselves were formidable. Revenue climbed 24 percent year over year to $119.8 billion, 9to5Google reported, with operating income of $40.8 billion producing a 34 percent margin. Google Cloud supplied the showpiece figure, growing 82 percent to $24.8 billion against consensus expectations near $22.46 billion. Bloomberg's assessment distilled the evening: a clear cloud beat, overshadowed by a spending plan that deepened anxiety about the cost of the AI buildout.

The reaction extends a pattern that has hardened across recent big tech earnings seasons. Beating estimates no longer guarantees a rally; the market now grades the invoice as closely as the income statement, and it treats each upward revision in infrastructure budgets as a claim on future margins that management must justify rather than simply announce.

Strong Numbers, Skeptical Tape

The advertising franchise showed no sign of the AI-driven erosion that bears have predicted for two years. Search and other revenue rose 17 percent to $63.3 billion, and Google Services overall grew 15 percent to $94.5 billion. The strain surfaced elsewhere: quarterly capital expenditure reached a record $44.9 billion, and the outlay pushed free cash flow negative for the period, according to the earnings call transcript published by Investing.com. A company that has historically functioned as a cash machine briefly spent more than it generated in order to buy compute.

Chief executive Sundar Pichai kept the focus on demand. "Cloud revenue grew 82%, powered by strong demand for AI infrastructure and AI solutions," he said in remarks accompanying the results. The argument, implicit but unmistakable, is that Alphabet is capacity constrained rather than demand constrained, and that every dollar of deferred spending is revenue surrendered to Microsoft or Amazon.

Backlog Worth Half a Trillion Dollars

The strongest evidence for that argument sits in the order book. Google Cloud's backlog swelled to $514 billion, up from roughly $460 billion a quarter earlier, a contracted revenue pipeline exceeding four years of the division's current sales. Usage statistics point the same way. The Gemini app has reached 950 million active users, Google's systems now process 22 billion API tokens per minute, up from 16 billion in the prior quarter, and nearly 90 percent of the Fortune 100 uses Gemini Enterprise, according to earnings call highlights compiled by Yahoo Finance.

The token figure repays attention as an operating metric. A sequential jump of nearly 40 percent in processing volume within a single quarter describes inference demand compounding faster than almost any revenue line in the company, and it explains why management talks about capacity in the language of shortage. It also carries a caution: a substantial share of that consumption is free-tier usage that monetizes indirectly, if at all, which is precisely the gap between activity and revenue that skeptics keep pointing to.

Finance chief Anat Ashkenazi told analysts the higher guidance reflects "an acceleration in the delivery of capacity," per the Investing.com transcript, language that describes data centers arriving faster rather than costs inflating on a fixed plan. She also signaled that spending would step up substantially again in 2027, a disclosure the market arguably found more consequential than the current-year revision.

Where $205 Billion Actually Goes

The composition of the budget clarifies the strategy. Roughly 60 percent of technical infrastructure investment flows to servers, primarily AI accelerators including Alphabet's in-house TPUs alongside Nvidia GPUs, with the remaining 40 percent funding data center construction and networking, Business Standard reported. The TPU share is the quiet advantage: every workload Alphabet lands on its own silicon loosens its dependence on Nvidia's pricing and widens a unit-cost gap over competitors buying accelerators at market rates.

Depreciation is the slow-moving consequence of all of it. Accelerators bought at record volume today become expense lines for years to come, and if AI revenue compounds more slowly than the hardware ages, margins compress with mechanical certainty. A 34 percent operating margin gives Alphabet substantial room to absorb that drag. Whether shareholders will wait through it is the open question, and Wednesday's selloff suggests the answer is no longer automatic.

Hyperscaler Spending Ratchet

Alphabet's revision also resets the competitive baseline. In the current cycle, each hyperscaler's capital budget functions as a signal to the others, because conceding capacity during an AI demand surge reads as conceding future market share. The $514 billion backlog is management's rebuttal to the charge of speculative overbuilding: the capacity being financed is, to a large degree, already sold. What the market is discounting instead is execution, the timing of converting backlog into recognized revenue, the margins embedded in those contracts, and the possibility that enterprise AI budgets cool before the concrete cures.

The bull and bear cases now share the same facts and disagree only on duration. Bulls see an 82 percent cloud growth rate, a half-trillion-dollar order book and an advertising business still expanding at double digits, and conclude the spending is rational. Bears see negative free cash flow, a 2027 budget already flagged to rise, and a return profile that remains a promise rather than a disclosure. Both positions were visible in a single after-hours session that repriced one of the world's most profitable companies downward on its best cloud quarter to date.

October Is the Next Test

The next checkpoint arrives with third-quarter results in the autumn, when investors will want cloud margin expansion to travel with cloud growth and clearer disclosure on the returns the 2026 program is producing. For now, Alphabet is behaving like a company convinced that the binding constraint on its future is megawatts and silicon rather than demand. The share price carries the counterargument: the market is done funding conviction on faith and wants the yield on all this capital spending made visible, quarter by quarter.