Amazon and Google Lead the AI Capital Expenditure Race — But Where's the Prize?

AI Capital Expenditure Race Heats Up

In 2026, as the AI wave sweeps the globe, tech giants' wallets are shrinking at an alarming rate. Amazon has announced plans to invest up to $200 billion in capital expenditure (capex), primarily for AI infrastructure, while Google is not far behind, planning to spend between $175 billion and $185 billion. This is not just a numbers game; it is a life-and-death struggle for future AI supremacy.

In 2026, Amazon plans to spend $200 billion in capital expenditure. Google follows closely with $175–185 billion. That's a staggering amount! (TechCrunch)

The atmosphere of the capex race is suffocatingly intense. Amazon, through its cloud computing arm AWS, is accelerating the deployment of NVIDIA H100 and its own Trainium chip clusters; Google, leveraging TPU v5 and the Gemini model, is building the world's largest AI training network. In comparison, though Microsoft and Meta also have substantial budgets, they clearly lag behind these two frontrunners.

Amazon: The Cloud Empire’s AI Ambitions

As the leader in the global cloud computing market, Amazon's $200 billion capex is not unfounded. AWS already holds about 32% of the cloud services market, and AI demand is driving rapid revenue growth. In 2025, AWS's AI-related business contributed billions in revenue, expected to double by 2026.

Amazon's investment priorities include building dozens of new data centers, procuring millions of AI accelerators, and developing the next-generation Inferentia inference chips. These initiatives aim to reduce the cost of AI model training and provide enterprise customers with one-stop AI services. The infrastructure foundation laid during the Bezos era is helping Amazon thrive in the AI capex race.

Google: From Search to AI All-Rounder

Google's budget of $175–185 billion reflects the ambition of the Alphabet group. Although Google Cloud holds only an 11% market share, it has caught up in AI applications thanks to DeepMind's AlphaFold and the Gemini large model. The 2026 capex will be used to expand supercomputing clusters across the United States, Europe, and Asia, with total computing power expected to exceed 10 EFLOPS (exaflops).

Google's advantage lies in vertical integration: its self-developed TPU chips reduce reliance on NVIDIA, while Waymo's autonomous driving and YouTube's recommendation systems provide massive training data. In this investment race, Google is transforming from a follower into a leader.

Industry Background: A Trillion-Dollar Bet Under the AI Boom

Looking back at AI history, ChatGPT ignited a global AI race in 2023, leading to chip shortages and a scramble for data center land. NVIDIA's market capitalization soared to $3 trillion, making it the biggest winner. But the capex frenzy of 2026 stems from a deeper logic: training cutting-edge large models requires trillion-parameter computing resources, with a single training run costing over $100 million.

According to Morgan Stanley, global AI infrastructure investment will reach $500 billion in 2026, accounting for 70% of total tech giant capex. Microsoft's $100 billion supercomputer project with OpenAI, Meta's AI-specific data centers, and Apple's private AI cloud are all intensifying the competition. Chinese companies like Alibaba and Tencent are also accelerating their local deployments, trying to grab a slice.

However, this race is not a zero-sum game. Power shortages, supply chain bottlenecks, and geopolitical risks are testing each company's execution. The U.S. Federal Energy Regulatory Commission warns that AI data centers could consume 10% of the nation's electricity, sparking environmental controversy.

Where’s the Prize? Monopoly or Bubble?

The question is: what do these astronomical investments actually buy? In the short term, it's monopoly rights over computing resources. Whoever owns the most GPUs can launch AGI (Artificial General Intelligence) faster and seize the trillion-dollar AI SaaS market. Amazon and Google aim to build an "AI moat," forcing competitors to pay for their infrastructure.

In the long run, the prize could be a new business empire. Imagine AWS becoming the global AI brain, and Google Gemini driving intelligence in everything. This would reshape advertising, e-commerce, and entertainment industries, creating trillions of dollars in value. But the risks are equally huge—if the AI bubble bursts or regulations tighten (e.g., the EU AI Act), these capex could become sunk costs.

Editor's Note: A Rational View of the AI Investment Frenzy

As an AI tech news editor, I believe this capex race is a double-edged sword. On one hand, it accelerates technological progress, pushing humanity toward an intelligent era; on the other, excessive investment could lead to a 2000 internet bubble 2.0. Amazon and Google's lead comes from first-mover advantages, but Microsoft's OpenAI ecosystem and emerging dark horses like xAI should not be underestimated. Investors should focus on ROI (return on investment) rather than simply the pace of spending. In the future, whoever can efficiently convert computing power into products will laugh last.

In short, the winner in the AI capex race takes all, but the prize is far from easy to grasp. The tech world stands at a historic crossroads—let's wait and see.

This article is translated from TechCrunch, by Russell Brandom, original date February 6, 2026.