The AI CapEx Ranking: Who’s Betting the Most Per Dollar of Revenue

Amazon Is Burning Through Cash at 3x Meta’s Rate to Win the AI Infrastructure War

Amazon delivered zero free cash flow on $181.5 billion in revenue this quarter—a financial impossibility that reveals the most aggressive AI infrastructure — as explored in the economics of AI compute infrastructure — bet in corporate history. While Meta generated $26.8 billion in net income to self-fund its AI ambitions, Amazon is raising $53.4 billion and torching every dollar of operational cash flow to build the data center empire that will determine AI’s future.

According to The Business Engineer’s earnings analysis series covering all seven tech giants, the AI capital expenditure intensity rankings reveal a stark divide between companies building the infrastructure and those riding on others’ investments. Google escalated capex from $106 billion to $190 billion while adding $31.1 billion in new debt. Microsoft’s Azure re-accelerated to 40% growth, proving enterprise AI demand remains insatiable. Meanwhile, Apple—sitting on the world’s largest cash pile—is spending the least per revenue dollar, instead partnering with Anthropic and Google for AI capabilities.

TSMC reported that AI-related revenue now comprises 15% of total sales at $23.5 billion quarterly revenue, while Qualcomm’s AI chip revenue doubled year-over-year to $8.7 billion. These numbers expose the supply chain reality: chip makers are printing money while cloud providers hemorrhage cash building capacity.

The most telling metric isn’t absolute spending—it’s spending per dollar of revenue. Amazon’s 29.4% AI capex intensity dwarfs even Google’s 23.7% rate, despite Google’s larger absolute numbers. This suggests Amazon believes whoever controls the physical infrastructure layer wins everything above it. They’re essentially betting the company on becoming the AWS of the AI era.

Meta’s position is uniquely sustainable at 18.1% intensity because their $26.8 billion quarterly profit fully funds AI investments without external financing. Mark Zuckerberg is building AI capabilities using Instagram and Facebook’s money machine—a luxury Amazon can’t afford as they sacrifice all free cash flow.

Apple’s 8.9% intensity reveals Tim Cook’s calculated decision to let others build expensive infrastructure while Apple focuses on consumer AI integration. Their partnerships with Anthropic and Google for AI capabilities represent the ultimate asset-light strategy—but it also means Apple controls none of the underlying AI value chain.

The chip companies tell the real story. TSMC’s 15% AI revenue mix on $23.5 billion quarterly sales, combined with Qualcomm’s doubled AI chip revenue to $8.7 billion, proves AI demand is genuine and growing. These companies sit in the profitable middle—selling shovels to gold rush participants.

Microsoft’s Azure re-acceleration to 40% growth after concerns about slowing proves enterprise AI adoption isn’t a bubble. Companies are paying premium prices for AI capabilities, validating the massive infrastructure investments.

The bold prediction: Amazon’s zero free cash flow strategy will either create the next AWS-scale monopoly or represent the most expensive infrastructure bet in corporate history. By 2025, either Amazon Web Services dominates AI infrastructure globally, or Google’s more balanced approach of building both infrastructure and models wins. Meta emerges as the dark horse—their profit-funded AI development creates sustainable competitive advantages without financial risk.

Apple’s partnership strategy looks smart short-term but dangerous long-term. In five years, they’ll either be praised for capital efficiency or scrambling to build AI capabilities they should have developed internally. The companies burning cash today are building moats. Those preserving cash might find themselves permanently locked out.