The trend: AI investment is shifting away from servers and chips toward energy, real estate, and infrastructure control. 

SoftBank agreed to acquire infrastructure giant DigitalBridge in a $4 billion deal, per CNBC, and Alphabet is acquiring data center company Intersect for $4.75 billion in cash.

As AI is constrained by power and regulatory approvals, not software, advertisers will face rising cost disparities and uneven access to AI tools depending on which platforms secure capacity first.

Why it’s worth watching: The trend reframes AI from a software and model race into a physical supply-chain challenge.

The number of global hyperscale data centers has more than doubled over the past five years, rising from 584 sites in Q4 2019 to 1,189 as of the end of Q1 2025, per Synergy Research Group. Even that expansion is falling behind AI demand, exposing power and government approval as chokepoints.

SoftBank and Alphabet are effectively hedging their bets ahead of grid bottlenecks that could take years to unwind, but this strategy carries risk. New AI data centers are expensive, slow to monetize, and vulnerable to overbuild if efficiency gains reduce compute needs or AI demand cools. 

Bigger picture: Firms of various backgrounds are getting into the AI infrastructure race.

• BlackRock: The global asset manager is backing large-scale data center platforms as long-duration assets, betting that compute and power could rival traditional commercial real estate in strategic value.
• Prologis: The world’s largest logistics real estate owner plans to invest about $8 billion to develop AI data centers on warehouse-adjacent land, with capacity coming online between 2026 and 2029.
• Chevron: The energy corporation is developing power plants designed for co-located AI data centers, targeting up to 4 gigawatts of capacity by 2027.

These buildouts indicate AI is becoming industrial. Companies with no previous data center interests could rival hyperscalers like Amazon Web Services (AWS), Microsoft Azure, and Google Cloud. 

Industry takeaway: Future dominance will be defined by timelines and access to power and land—not just algorithms. Companies controlling data centers and pipelines for AI can dictate price and determine access and how fast the technology evolves.

As power, land, and grid access tightens, platforms with secured infrastructure will deliver faster models, lower latency, and more predictable pricing, while others face throttling, higher costs, or usage caps.