Google Amazon AI Cost Warning | AI Infrastructure Spending Concerns

Google and Amazon have issued stark warnings about the true cost of artificial intelligence through their latest financial disclosures and sustainability reports, revealing that the AI gold rush carries a far heavier price tag than Wall Street has priced in. The twin tech giants are exposing painful financial and environmental realities—from ballooning depreciation expenses to surging carbon emissions and the first cloud price hikes in two decades—that investors and policymakers are only beginning to grasp.

The Depreciation Time Bomb

The most immediate warning comes from the accounting realm. Bank of America analyst Justin Post has projected that Alphabet, Amazon, and Meta will face combined depreciation and amortization expenses of $43.5 billion, $95.9 billion, and $40.7 billion respectively by 2027—figures that significantly exceed Wall Street consensus estimates. The gap is particularly stark for Alphabet, where Bank of America forecasts a $7 billion shortfall versus Street estimates in 2027 alone.

This matters because when companies spend billions on long-term assets like data centers and graphics processing units, they don't count the full cost upfront. Instead, they spread it over the asset's "useful life." If that useful life is revised downward—as Amazon recently did when it changed some server and networking assets from six years to five years due to rapid AI innovation—depreciation costs accelerate, eating into reported profits.

The underlying problem is that GPUs powering AI workloads may have shorter lifespans than traditional enterprise hardware. High-intensity workloads and rapid technological advancement could render these expensive assets obsolete faster than expected. This creates a dangerous disconnect: while Alphabet, Meta, and Amazon are projected to grow revenue at 13% and 12% annually through 2027, their combined depreciation expenses are expected to grow at 33% and 30% over the same periods.

The AI Infrastructure Arms Race

The scale of spending is staggering. In 2026 alone, Amazon projects $200 billion in capital expenditures, Alphabet $175-185 billion, and Microsoft approximately $190 billion. Collectively, the major hyperscalers are on track to spend roughly $725 billion this year on AI infrastructure.

For context, Goldman Sachs estimates that Amazon, Microsoft, Google, and Meta will collectively spend $5.3 trillion on AI infrastructure between 2025 and 2030. This represents a fundamental shift from the asset-light, capital-efficient business models that supported premium tech valuations over the past decade. The median capex-to-sales ratio for Big Tech companies has more than doubled, rising from around 10% to over 20% as of late 2025.

Microsoft's June 2026 stock rout—shares slid to a 52-week low of $349.20—served as a stark warning. The selloff was driven largely by investor scrutiny of the company's roughly $190 billion AI capital spending plan, with roughly two-thirds going toward short-lived assets like GPUs and CPUs that depreciate faster and tie more directly to near-term revenues. Microsoft Cloud gross margin guidance of 64% for the fiscal fourth quarter represents a 4% year-over-year decline, as new capacity costs outpace revenue generation.

First Cloud Price Hikes in 20 Years

Perhaps the clearest indicator that AI costs are real and mounting came in January 2026, when Google Cloud announced price increases across its services—breaking a 20-year industry convention of declining cloud prices. Data transfer prices doubled in North America, rose 60% in Europe, and increased 42% in Asia. Just days earlier, Amazon Web Services had raised its EC2 machine learning capacity block prices by approximately 15%.

These price hikes signal that the era of subsidized AI is ending. For years, cloud providers absorbed infrastructure costs to capture market share. Now, the sheer expense of building and operating AI-capable data centers is forcing them to pass costs to customers. Google Cloud explicitly framed its price increases as necessary to match the value and performance of its services.

The Environmental Toll

Google and Amazon's recent sustainability reports reveal another dimension of AI's true cost: surging carbon emissions. Google's emissions increased 25% year-over-year, while Amazon's rose 16%, largely due to AI-driven energy demand.

The primary culprit is Scope 3 emissions—the pollution generated by companies' supply chains and product use. For both companies, this includes GPU manufacturing, data center construction, and the semiconductor supply chain. The steel and cement used in data center construction are both carbon-intensive industries. Semiconductor fabrication consumes enormous amounts of energy and uses potent greenhouse gases thousands of times more warming than carbon dioxide.

While both companies continue purchasing renewable energy to power operations, they are increasingly relying on fossil fuels to meet AI's voracious energy demands. Even more concerning is that their Scope 3 emissions—which arise from building and equipping data centers—are growing at a rate that renewable energy purchases cannot offset.

The Monetization Challenge

The fundamental question now facing the industry is whether AI revenues can grow fast enough to justify the massive investment. Uber burned through its entire 2026 Claude Code budget in four months. Walmart shifted from unlimited AI token access to fixed per-employee allocations. OpenAI CEO Sam Altman recently acknowledged that companies are pulling back on AI usage as costs become a "huge issue."

A Bain & Company study found that 44% of large firms funding AI investments are basing them on savings from previous spending that "haven't yet materialized for some of them." This creates a dangerous cycle: companies invest in AI expecting cost savings, but those savings don't materialize, making future investment harder to justify.

What This Means for the Industry

The warning from Google and Amazon is clear: AI's real cost extends far beyond the sticker price of GPUs. It includes accelerating depreciation that erodes profits, environmental damage that threatens climate goals, and the first cloud price hikes in two decades that will pass costs to customers. The era of subsidized AI experimentation is ending, replaced by a period of financial reckoning where technology must prove its economic value.

The outcome of this spending cycle is far from certain. As one expert put it, "The future of AI will not be defined by who spends the most on AI. It will be defined by who can turn AI into an enduring organizational capability that creates measurable business advantage." For Google, Amazon, and their peers, the warning signs are flashing that the path to AI profitability will be longer and more expensive than anyone anticipated.