The Hidden Price of AI: What Google and Amazon Reveal About Technology’s True Cost

Amazon will spend over $100 billion on artificial intelligence infrastructure in 2025, with the vast majority directed toward AWS, CEO Andy Jassy revealed during fourth-quarter earnings—a $17 billion increase from 2024 that signals an unprecedented capital race among cloud giants. Microsoft, Amazon, and Google are collectively projected to spend $325 billion on AI in 2025, a 40% jump from $230 billion the prior year, primarily for data center buildouts, GPU clusters, and custom silicon. Yet beneath this aggressive expansion lies a troubling shift: after two decades of declining cloud prices, both Amazon and Google have begun raising rates sharply, ending an era of cost predictability for enterprises and signaling that artificial intelligence’s true operational burden is far steeper than the market has priced in.

The $100 Billion Bet That Changes Everything

Amazon’s commitment to spend over $100 billion on AI infrastructure in 2025 represents the most ambitious capital allocation among hyperscalers and reflects the intensity of competition for AI workloads. The increase from $83 billion in 2024 underscores how rapidly cloud providers must expand capacity to meet surging demand for compute-intensive model training and inference. AWS, Amazon’s cloud division, will absorb the vast majority of this spend, positioning the company as the primary infrastructure backbone for generative AI deployment globally.

This capital intensity extends across the entire cloud industry. Microsoft, Amazon, and Google combined are deploying $325 billion toward AI infrastructure in 2025, a 40% year-over-year increase that dwarfs spending in traditional enterprise IT. Amazon has guided 2026 capex to approximately $200 billion, with cumulative capex through 2027 reaching $344 billion—a trajectory that implies $10 billion per year in free cash flow deficits during peak buildout. Google has raised its 2025 capex guidance to $91 billion to $93 billion, up from $75 billion, citing robust demand for cloud products and heavy investment in global infrastructure.

The Price Hikes That Signal a New Era

The most striking signal of AI’s true cost came in mid-2026, when AWS and Google Cloud simultaneously raised prices—moves that shattered a two-decade trend of declining cloud costs. AWS increased hourly rates for EC2 Capacity Blocks for ML by approximately 20% effective July 1, 2026, with new pricing reaching $14.04 per hour for P6-B300 instances and $12.355 per hour for P6-B200 instances. This marked the second price increase in 2026 alone, following a 15% hike in January, reflecting the tight supply of GPU capacity and operating costs exceeding $1 million annually to power H100 clusters.

Google Cloud’s price adjustments were even more dramatic. Effective May 1, 2026, Google raised prices for CDN Interconnect and AI infrastructure, with North America CDN rates jumping 100%—doubling from $0.04 per gigabyte to $0.08 per gigabyte. The company attributed these increases to heavy investment in global infrastructure, acknowledging that the era of perpetually falling cloud costs has ended. For enterprises that have built business models around declining compute expenses, these hikes represent a structural shift in the economics of AI deployment.

Strategic Partnerships Reveal the Real Economics

Behind the capex headlines, major AI companies are locking in long-term commitments that expose the true scale of infrastructure needs. OpenAI signed a $38 billion, 7-year AWS spend commitment in November 2025, later updated by Barclays analysts to $138 billion when contingent investment tied to commercial milestones is included. This contract secures AWS infrastructure for GPT model training and chat queries, cementing Amazon’s role as OpenAI’s primary cloud provider and guaranteeing revenue streams that justify massive capex outlays.

Amazon has simultaneously positioned itself as a key investor in AI model creators. The company committed up to $25 billion to Anthropic, the creator of Claude, with an initial $5 billion tranche and $20 billion contingent on commercial milestones, following $8 billion in prior rounds. Google responded with a $40 billion investment in Anthropic announced in 2026, including a $10 billion upfront payment that valued the company at $350 billion. Analysts note that these investment deals are fundamentally about securing compute contracts—the actual economic engine driving hyperscaler capex decisions.

Market Dominance Masking Narrowing Leads

AWS maintains the largest share of global cloud infrastructure at 28% in the fourth quarter of 2025, with first-quarter 2025 revenue reaching $29.27 billion and year-over-year growth of 16.9%. The division claims approximately 100% AI growth, reflecting strong demand for machine learning workloads on AWS infrastructure. However, AWS’s competitive position is tightening as rivals capture high-value AI segments.

Microsoft Azure holds 33% market share while Google Cloud has captured 30% or more of AI growth, according to industry tracking. Amazon’s dual monetization model—earning revenue externally through AWS sales while deriving internal benefits through retail and logistics efficiency—gives the company an edge competitors like Meta (reliant on advertising revenue) or Microsoft (split between Azure and internal uses) cannot match. Wall Street responded to Amazon’s $125 billion capex announcement with an 11% stock surge, reflecting investor confidence that the company’s infrastructure investments will generate outsized returns.

Two Decades of Price Declines Meet Infrastructure Reality

The price increases announced by AWS and Google Cloud in 2025 and 2026 represent a fundamental break from cloud computing’s foundational economics. Since the early 2000s, cloud providers competed primarily on cost, with prices declining steadily as infrastructure scaled and efficiency improved. That dynamic has reversed. The massive capex requirements of AI—driven by the need for specialized GPUs, custom chips like Trainium and Inferentia, and power-intensive data centers—have overwhelmed the traditional cost reduction curve.

The shift reflects supply constraints and operational realities that money alone cannot immediately solve. Building data centers capable of housing thousands of GPUs, securing reliable power supplies in an era of energy scarcity, and developing custom silicon all require time and capital beyond what traditional cloud economics could accommodate. AWS’s decision to raise prices twice in six months, and Google’s 100% CDN rate increase, signal that demand for AI compute far exceeds available supply.

What Comes Next for Enterprise Buyers and Investors

The trajectory of capex spending and pricing suggests that AI infrastructure costs will remain elevated through 2027 and beyond. Amazon’s $200 billion 2026 capex guidance and Google’s increased spending targets indicate that hyperscalers view the current buildout phase as multi-year. Enterprise customers should expect further price increases as demand continues to outpace supply, and cost optimization will become a critical priority for companies deploying AI at scale.

Investors tracking Amazon, Google, and Microsoft must recognize that the AI infrastructure race carries substantial financial risk alongside opportunity. The $325 billion annual spend, rising prices, and massive capex commitments represent a bet that AI adoption will accelerate rapidly enough to generate returns justifying these outlays. The next critical milestones—AWS and Google Cloud’s pricing announcements in 2026 and 2027, along with earnings reports revealing AI revenue growth relative to capex—will determine whether this bet pays off or whether the industry has overestimated demand and underestimated the true cost of artificial intelligence.