Author | Shuyan For more financial information | BT Finance Data Pass The full text is 2,444 words and is expected to take 9 minutes to read.

Google, Meta, Microsoft, and Amazon plan to collectively spend over $600 billion on AI infrastructure this year. This figure exceeds the annual GDP of most developed countries in Europe. Meanwhile, financial forecasts circulating on Wall Street indicate that OpenAI's CFO warns: If growth does not accelerate, the pressure from computing bills will continue to rise. Together, these two events constitute the core contradiction in the AI industry in 2026—money is still pouring in, but no one has clearly explained how it will be earned back.

1

What Are the $600 Billion Bets on?

Let's first clarify the numbers for these four companies.

Google's parent company, Alphabet: Capital expenditure guidance for 2026 is $175 billion to $185 billion, primarily for TPU chip clusters and cloud infrastructure (Source: Alphabet earnings report, February 2026).

Meta: Capital expenditure guidance for the full year 2026 is $115 billion to $135 billion, up about 60% to 90% from 2025, with over 80% allocated to AI infrastructure (Source: Meta earnings call, February 2026).

Amazon: Capital expenditure for 2026 is approximately $200 billion, with CEO Andy Jassy stating in a shareholder letter that most will gradually monetize from 2027 to 2028 (Source: Amazon shareholder letter, April 2026).

Microsoft: Capital expenditure for fiscal year 2026 is about $100 billion, with about two-thirds allocated to short-cycle assets like GPUs and CPUs (Source: Microsoft earnings report, February 2026).

The combined total for the four companies exceeds $600 billion, marking the largest single-year capital expenditure in the global tech industry in recent years (Source: CME Group analysis, February 2026).

Each company's betting logic differs. Google is betting on cost optimization through self-developed TPUs—while competitors are still affected by third-party GPU price hikes, Google's computing costs have begun to differentiate from its revenue. Meta is betting on the continuous conversion ability of advertising cash flow: advertising revenue covers the speed of infrastructure expansion. Amazon is betting on the time difference—spending money first and waiting for orders to monetize from 2027 to 2028. Microsoft is betting on Azure's cloud market share.

However, these four companies' bets share a common underlying assumption—that the speed of AI commercialization can keep up with the pace of capital investment.

This assumption has not yet been fully proven.

2

Where Do the Three Monetization Paths Stand?

There are three main paths for AI commercialization: advertising monetization, enterprise SaaS services, and Agent-based fees. The progress of these three paths varies significantly.

▍Advertising Monetization—Most Mature, But Growth Slowing

Meta is the fastest on this path. Driven by AI, according to Meta's earnings call, Instagram's ad conversion rate has increased by about 5%, Facebook's by about 3%, and Meta AI assistant's monthly active users have exceeded 1 billion. In fiscal year 2025, Meta's operating profit margin reached 43%, with advertising cash flow remaining the most stable monetization engine. However, according to research by analysis firms, its incremental operating profit margin has shown signs of stabilizing at a high level, with marginal revenue growth slowing. Google's situation is similar—search ads remain strong, and plans to introduce ads to the Gemini App are still under evaluation, with management stating in the latest earnings call that "plans will be shared when the time is right."

▍Enterprise SaaS—Rapid Growth, But Significant Size Differences

Anthropic is a representative case on this path. Its annualized revenue grew from $9 billion at the end of 2025 to about $30 billion in April 2026, with over 1,000 enterprise clients spending more than $1 million annually (Source: Anthropic official disclosure, April 7, 2026). Claude Code holds about 54% of the market share in AI programming tools, with single-product annualized revenue exceeding $2.5 billion. This path of precisely targeting high-net-worth enterprise clients is currently highly efficient in AI commercialization.

▍Agent-Based Fees—Greatest Potential, But Earliest Stage

AI Agents are seen as the next monetization engine and a key betting direction for this round of capital expenditure. However, as of now, the commercial scale of Agent services is still in its early stages, with no predictable and stable revenue source yet formed.

3

The Time Difference Between Expenditure and Monetization

There is a time difference that is difficult to quickly compress between capital expenditure and commercialization revenue. The computing power of data centers typically takes six to twelve months from procurement to deployment; another period is needed from deployment to generating stable revenue.

Amazon has directly stated this in its shareholder letter: The capital expenditure in 2026 will gradually monetize from 2027 to 2028.

Institutions like Morgan Stanley and Citigroup have reached a relatively consistent judgment on this earnings season: The first signal for judging AI investment returns is revenue acceleration, not the scale of capital expenditure itself. Whoever can continue to expand investment while accelerating revenue growth simultaneously will have a more solid fundamentals.

Anthropic Annualized Revenue (ARR): About $30 billion (Source: Anthropic official disclosure, April 7, 2026)

OpenAI Annualized Revenue: About $25 billion (Source: Financial forecasts circulating on Wall Street, February 2026)

Google's incremental operating profit margin is about 53.20%; Microsoft's is about 58.90% (Source: TradingKe research report, February 2026)

Amazon's capital expenditure for 2026 is about $200 billion, with most monetizing from 2027 to 2028 (Source: Amazon shareholder letter, April 2026)

The $600 billion AI bet is waiting for an answer—when will it pay off?

This week, the first-quarter earnings reports of Google, Microsoft, Amazon, and Meta will be released one after another. What the market truly wants to see is not just how high the revenue figures are, but whether management can provide a verifiable monetization path.

This article is for information sharing and industry analysis only and does not constitute any investment advice, investment analysis opinion, or trading solicitation. The market has risks, and investment should be cautious. Anyone making investment decisions based on this article's content bears the risks and gains or losses themselves, with the author and publishing platform not assuming any legal responsibility.

Information Sources

1. CME Group: Tech Giants' Earnings Season Analysis (February 2026)

2. TradingKey: The Endgame of $600 Billion—2026 Global AI Giants' Capital Efficiency Audit Report (February 2026)

3. Anthropic Official Disclosure: Annualized Revenue Exceeds $30 Billion (April 7, 2026)

4. Sina Finance: Anthropic's Annualized Revenue Surpasses OpenAI's for the First Time (April 7, 2026)

5. Zhitong Finance: The AI Bull Market Narrative Reaches Its "Moment of Truth" (April 29, 2026)

6. 21 Economic Report: OpenAI Raises $110 Billion in Funding (February 2026)

This article is an original piece by BT Finance and may not be used, reproduced, disseminated, or adapted without permission. Legal action will be pursued for infringement.