Google has set an internal target to grow AI capacity 1,000 times by 2030, even as CEO Sundar Pichai publicly warns about market irrationality. According to CNBC, infrastructure VP Amin Vahdat presented the plan at a November 6 all-hands meeting. The directive requires doubling AI serving capacity every six months to support what the company calls the age of inference.
Aggressive Expansion Amid Market Skepticism
The mandate contrasts sharply with Pichai’s recent BBC interview, where he acknowledged elements of irrationality in AI market valuations. But employees heard a different message internally. Vahdat warned that AI infrastructure competition is the most critical and expensive part of the race. Pichai admitted Google missed opportunities with its Veo video tool due to hardware limits.
Alphabet raised its 2025 capital spending forecast to $93 billion, with a significant increase planned for 2026. When employees questioned this strategy during Q&A, Pichai defended the approach by citing the company’s balance sheet. He argued Google is better positioned to withstand misses than rivals. The logic frames underinvestment as existential but overinvestment as merely expensive.
Custom Chips Drive Efficiency Strategy
The plan relies on custom silicon to prevent costs from spiraling. Google recently made its Ironwood TPU chips generally available, claiming 10x peak performance improvement over the v5p generation. The seventh-generation chip also offers twice the performance per watt compared to the previous Trillium design.
Vahdat outlined a brutal requirement: deliver 1,000 times more capability for the same cost and energy level. Google is offloading standard tasks to new Arm-based Axion CPUs to free power for AI workloads. Engineers integrate software directly with hardware through co-design, allowing the company to squeeze gains from custom architectures.
Industry Context and Competitive Pressure
The Big Four tech companies plan to spend over $380 billion on infrastructure this year. Nvidia CEO Jensen Huang rejected bubble claims this week, citing tangible demand. OpenAI faces its own challenges, with a leaked memo from Sam Altman referencing rough vibes and economic headwinds.
Google’s advantage lies in vertical integration and custom chips. The strategy aims to weather a potential price war better than competitors relying on Nvidia hardware. Recent launches like Gemini 3 Pro drive inference demand. The outcome depends on whether premium features generate revenue faster than hardware depreciates. Leadership framed 2026 as an intense year of ups and downs ahead.
