Sundar Pichai on the AI Spending Boom: Why Google Is Investing Up to $190 Billion

Exploding Demand for AI

Google also backed its argument with usage data showing extraordinary growth in AI workloads.

According to company figures shared around I/O, the amount of AI tokens processed across Google’s systems has grown from 9.7 trillion per month two years earlier to more than 3.2 quadrillion per month today.

That explosive increase reflects demand across:

• AI‑enhanced Search experiences
• Gemini assistants and agents
• Developer APIs and enterprise cloud workloads
• Consumer products such as YouTube and productivity tools

Pichai’s point: infrastructure investment is rising because usage is rising even faster.

Gemini 3.5 Flash: A Case Study in Efficiency

One example highlighted during the event was Gemini 3.5 Flash, a faster and cheaper AI model introduced to expand enterprise and developer adoption.

Lower‑cost models like this demonstrate the efficiency improvements Pichai believes will eventually reshape the economics of AI. By reducing inference costs and increasing speed, such models allow the same infrastructure to support far more applications and users.

The Real Constraints: Power and Compute

Despite defending the investment strategy, Pichai acknowledged that the AI boom faces significant physical constraints.

Among the biggest challenges:

• Power consumption for massive AI data centers
• Compute demand growing faster than hardware supply
• Infrastructure scaling for global AI services

These factors are becoming central issues for the industry as companies race to build enough computing capacity to support AI growth.

A Shift From His 2025 Tone on the AI Boom

Pichai’s comments at I/O 2026 also contrasted with his more cautious remarks a year earlier.

In 2025 interviews, he acknowledged that the AI investment surge contained “elements of irrationality” and warned that no company would be immune if an AI bubble burst.

By 2026, his message had shifted. Rather than focusing on bubble risks, he emphasized:

• unprecedented AI demand
• the inevitability of efficiency improvements
• the strategic necessity of building infrastructure early

He did not deny the risks—but argued that the long‑term trajectory of AI makes the investment rational.

The Big Picture

Pichai’s defense of Google’s spending reflects a broader industry bet. Nearly every major tech company is pouring capital into AI infrastructure, from specialized chips to massive data centers.

Google’s view, as articulated at I/O 2026, is that this phase of intense spending is the price of building the next computing platform—and that, as in previous tech cycles, the payoff will come when efficiency gains make AI dramatically cheaper and more ubiquitous.