Alphabet’s $4.75bn Intersect deal shows AI race is now a power and concrete problem

Alphabet’s decision to acquire Intersect for $4.75bn is less about expanding its corporate portfolio and more about confronting a hard reality of the AI boom: intelligence now depends on electricity, land and speed of construction as much as algorithms.

For years, the competitive edge in artificial intelligence was framed around research talent and model architecture. That has changed. Today, the bottleneck is infrastructure. Training and running large models requires vast data centres, enormous amounts of power and the ability to bring both online quickly. Alphabet’s move makes clear that the fight with rivals such as OpenAI is shifting decisively into this physical layer.

Intersect sits at the intersection of two constraints that have begun to collide. It develops data centre sites while also building or coordinating the energy generation needed to power them. That dual role is increasingly valuable as utilities struggle to keep up with demand from hyperscalers and as local opposition to new grid upgrades intensifies. Alphabet’s argument is straightforward: owning this capability reduces friction and shortens timelines.

The deal also reflects how extreme the AI arms race has become. OpenAI and its partners have announced infrastructure commitments running into the hundreds of billions of dollars, spanning data centres, chips and energy projects. Alphabet, despite its size, cannot assume its existing footprint will be enough. Buying Intersect is a way of accelerating capacity without relying entirely on third-party developers or slow-moving utilities.

There is also a defensive logic at work. Energy has become a reputational and political issue for big tech. Data centres are increasingly blamed for rising electricity prices, grid congestion and emissions. By bringing energy infrastructure in-house, Alphabet gains more control over how power is generated, where it is sourced and how new capacity is presented to regulators and communities. That control may prove just as important as raw megawatts.

The transaction is structured to limit disruption. Intersect will operate independently, and several assets are explicitly excluded from the deal. That suggests Alphabet wants the expertise and pipeline more than a wholesale integration. It is effectively buying optionality: the ability to co-design power and compute projects when it needs to, without fully absorbing the operational risk of every asset.

Texas looms large in this strategy. Alphabet has already committed tens of billions of dollars to the state, and Intersect’s co-located power and data centre projects there offer a template for how future builds might look. Regions with looser planning rules and abundant land are becoming critical battlegrounds in the AI race.

More broadly, the acquisition underlines a shift in how investors and executives should think about AI competition. Model performance still matters, but it is no longer sufficient. The winners will be those who can reliably secure land, power, cooling and grid access at scale. That favours companies with deep balance sheets and a willingness to invest years ahead of demand.

Alphabet’s Intersect deal is therefore not just a capacity play. It is a signal that AI leadership is being redefined, from who has the smartest software to who can build and power the biggest machines fastest.

The Recap

• Alphabet agreed to buy Intersect, a data center provider.
• Deal aims to bring data center and generation capacity online.
• Announcement posted on Alphabet's Investor Relations site for details.