News
Alphabet CFO Ruth Porat discusses hybrid strategy of building and leasing data centers for AI infrastructure
Thursday, March 12, 2026 at 10:48 PM
Alphabet CFO Ruth Porat emphasized the company's full-stack approach to AI infrastructure, noting that while Alphabet primarily constructs its own data centers, it also utilizes leased facilities to support its operations.
Context
Alphabet CFO Ruth Porat recently detailed Google's hybrid infrastructure strategy, emphasizing a "full stack approach" that combines building proprietary facilities with leasing third-party capacity. During recent investor communications, Porat stated that the company operates "a set of data centers, which we mostly build ourselves, but we also lease data centers," highlighting a flexible model designed to scale AI workloads rapidly. This shift follows Alphabet's massive $4.75 billion acquisition of Intersect in late 2025, a move intended to accelerate the deployment of energy and data center capacity.
The hybrid strategy is critical as Alphabet navigates a projected 60% year-over-year increase in capital expenditures for 2026, following a record $91–$93 billion spend in 2025. By balancing self-built sites optimized for internal TPU v7 (Ironwood) silicon with leased facilities, Google aims to mitigate bottlenecks in power and land availability. This infrastructure blitz is essential to support the 750 million monthly active users on the Gemini app and to maintain a competitive edge against rivals like OpenAI and Microsoft.
Sources (8)
Alphabet Investor Relations - 2024 Q2 Earnings Call
Alphabet to acquire data center and energy infrastructure company IntersectThe billion-dollar infrastructure deals powering the AI boom | TechCrunchGoogle TPUv7: The 900lb Gorilla In the RoomPresentations | Morgan Stanley
Google Q3 2025: $93 Billion CapEx Marks the Moment AI Became InfrastructureCanadian Technology Magazine: Why Google’s Full-Stack AI Push Changes EverythingVertical Integration: Google’s AI Compute Edge? – HyperFRAME Research
Related Companies
Google
GOOGL