Google faces an unusual problem for a tech giant: even with nearly $205 billion committed to building artificial intelligence computing capacity in 2026, the company still cannot construct data centers fast enough to meet customer demand. This supply shortage has forced Google to pre-sell chunks of its scarce computing power to outside customers, including Anthropic, the AI startup behind the Claude language model. The arrangement reveals both Google's dominance in cloud computing and the intense competition for the physical infrastructure that powers modern AI systems.

Why Is Google Struggling to Build AI Capacity?

During Google's second-quarter earnings call in July, CEO Sundar Pichai acknowledged the constraint directly. "We continue to be supply constrained, a sign of momentum and rapid adoption," he stated. This admission is striking because Google, as Alphabet, is one of the world's largest technology companies with virtually unlimited access to capital and engineering talent. Yet the demand for AI computing power has outpaced even their ability to build it.

Sundar Pichai

The reason is straightforward: building data centers takes time. Google must design the facilities, manufacture or procure specialized chips called tensor processing units (TPUs), install power infrastructure, and connect everything together. Even with aggressive spending, these projects take years to complete. Meanwhile, customers want access to computing power now, not in 2027 or 2028.

How Is Google Monetizing Its Supply Shortage?

Rather than hoard its scarce capacity, Google has turned the shortage into a business advantage by locking in long-term contracts with major customers. The most significant example is Anthropic, the AI company that competes with Google's own AI efforts. Google has committed to providing Anthropic with multiple gigawatts of computing power, with approximately 5 gigawatts coming online starting in 2027. In exchange, Google secured a massive investment opportunity: the company committed up to $40 billion to Anthropic, with $10 billion invested immediately and as much as $30 billion more tied to performance milestones.

This arrangement benefits Google in several ways. First, it converts future computing capacity into contracted revenue years in advance. Second, it gives Google a financial stake in Anthropic's success, aligning their interests. Third, it demonstrates Google's confidence in its ability to build the infrastructure Anthropic needs, which strengthens Google's position as a cloud provider.

What Does This Mean for Google's Cloud Business?

Google Cloud, the division that sells computing services to outside customers, is experiencing explosive growth. The segment's revenue grew 82 percent year over year in the second quarter, reaching $24.8 billion, up from 63 percent growth in the first quarter. More importantly, profits are growing even faster than revenue. Operating income more than tripled year over year, from $2.8 billion to $8.8 billion, giving the segment an operating margin of approximately 36 percent, compared to about 21 percent a year earlier.

The backlog of signed contracts tells an even more compelling story. Google Cloud's contracted future revenue reached approximately $514 billion in the second quarter, up from $462 billion at the end of the first quarter. To put this in perspective, that backlog is more than four times the revenue Alphabet's entire business produced in a single quarter. This massive pipeline of committed work suggests that Google's cloud business will remain a growth engine for years to come.

Steps to Understand Google's AI Infrastructure Strategy

• Supply Constraint as Pricing Power: When demand exceeds supply, companies can charge premium prices and lock in long-term contracts. Google's inability to build capacity fast enough actually strengthens its negotiating position with customers like Anthropic, allowing it to secure multi-year commitments at favorable terms.
• Capital Spending Outpacing Operating Cash Flow: Google is spending more on infrastructure than its operations generate in cash, requiring the company to raise capital through stock sales and debt issuance. In the second quarter alone, the company collected $49.6 billion from stock sales and issued $20.3 billion in debt to fund its expansion.
• Anthropic's Dependency on Commercial Success: According to a Broadcom securities filing, Anthropic's access to the 5 gigawatts of computing power is contingent on the company's continued commercial success. Anthropic reported a run rate revenue exceeding $30 billion in April 2026, up from approximately $9 billion at the end of 2025, demonstrating the extraordinary growth required to justify Google's investment.

"We continue to be supply constrained, a sign of momentum and rapid adoption," said Sundar Pichai.

Sundar Pichai, CEO at Alphabet

The arrangement between Google and Anthropic illustrates a broader trend in the AI industry: the companies building the most advanced AI models need access to enormous amounts of computing power, and the companies that can provide that power hold significant leverage. Google's strategy of pre-selling capacity to Anthropic while simultaneously investing in the company represents a sophisticated approach to managing both supply constraints and competitive dynamics.

For investors and industry observers, the key takeaway is that Google's cloud business is not just growing; it is becoming increasingly profitable and locked in through long-term contracts. The company's willingness to invest $40 billion in Anthropic, a potential competitor, suggests confidence that the overall market for AI computing services is large enough to support multiple winners. At the same time, Google's supply constraints underscore just how resource-intensive building and operating modern AI systems has become.