Alphabet’s deal for Intersect is less about owning solar farms than owning something scarcer: a team, a development pipeline — and a faster path through America’s grid bottlenecks.
Alphabet’s agreement to buy Intersect for $4.75bn in cash (plus assumed debt) looks, at first glance, like another chapter in Big Tech’s long romance with renewable energy. But read the fine print and the motivation is clearer: this is not a trophy purchase of operating wind and solar farms. It is a bid to control time — and to reduce the risk that the next decade of AI growth gets stranded in an interconnection queue.
The transaction, announced December 22, 2025 and expected to close in the first half of 2026, brings Alphabet what it says it wants most: Intersect’s “world-class team” and “multiple gigawatts” of energy and data-centre projects that sit in development or construction through an existing partnership with Google. Intersect will keep its brand and leadership, and will work closely with Google’s infrastructure group — including on a co-located data-centre and power site in Haskell County, Texas.
What Alphabet is not buying may matter even more. Intersect’s existing operating assets in Texas, and its operating and in-development assets in California, are explicitly excluded and will remain with an independent company backed by existing investors. In other words, Alphabet is buying a capability and a pipeline — not the bulk of the steel already in the ground.
That distinction tells you what problem Big Tech is trying to solve.
The AI boom has turned electricity into a strategic constraint
Data centres can be built quickly by infrastructure standards. The energy system that powers them cannot. The International Energy Agency notes that a data centre can be operational in two to three years, while grids and generation often require longer lead times and large upfront investment.
That mismatch is now colliding with the fastest load growth many power markets have seen in decades.
In the IEA’s base case, global electricity use by data centres doubles to ~945 TWh by 2030, rising about 15% a year from 2024 to 2030. In the United States alone, the IEA projects data-centre consumption increases by roughly 240 TWh by 2030 versus 2024.
US government analysis is even more blunt about the near-term squeeze. A DOE report estimates data centres consumed about 4.4% of total US electricity in 2023, and could reach 6.7% to 12% by 2028 — with total usage rising from 176 TWh (2023) to 325–580 TWh (2028).
That surge is now visible in national demand forecasts. The US Energy Information Administration expects power consumption, after a record 4,198 billion kWh in 2025, to rise to 4,256 billion kWh in 2026 and 4,364 billion kWh in 2027, citing data centres among the key drivers.
The electricity system is not only being asked to supply more power. It is being asked to supply it in very specific places — Northern Virginia, parts of Texas, Arizona, Ohio, the Chicago area — where data centres cluster like industrial districts. That local concentration is where politics begins.
If you want a preview, look at what is happening in PJM, the giant grid covering 13 states and Washington, DC. The White House has been pushing measures that would force data centres to finance new generation and even “bring your own generation”, as policymakers try to stop consumer bills from absorbing the cost of the AI buildout.
Alphabet’s Intersect deal lands in this exact moment: AI demand is rising, the grid is straining, and the public mood is shifting from “innovation” to “who pays?”.
(For more on that debate — and why “pay your way” is becoming the new social contract — see our related feature: AI data centers are arriving like small cities — and turning electricity into politics.)
The real choke point is not turbines. It is interconnection
For years, corporate climate strategy in tech was built around a simple idea: buy clean electricity through long-term contracts and claim progress. Those deals still matter. BloombergNEF estimates that 2,916 corporate PPAs signed between 2015 and 2024 helped enable 301 GW of renewable projects.
But PPAs solve a procurement problem more than a physical one. They are a way to finance new clean capacity somewhere on the grid. They do not guarantee that power will show up where your data centre is, when it needs it — or that the project will connect on schedule.
That is where the interconnection queue becomes the plot.
As of the end of 2024, Lawrence Berkeley National Laboratory counted about 10,300 projects seeking grid interconnection in the US, representing 1,400 GW of generation and about 890 GW of storage. Behind those enormous numbers are two quieter, more consequential ones:
- The median time from interconnection request to commercial operation has risen to over 4 years for projects built in 2018–2024 (more than double the duration for projects built in 2000–2007).
- Only 13% of capacity that submitted interconnection requests from 2000–2019 had reached commercial operation by the end of 2024; 77% had been withdrawn.
In plain English: the queue is long, slow, and full of projects that never get built.
For a hyperscaler racing competitors in AI, that is intolerable risk. If your data centre opens before your power plan clears studies and upgrades, you either delay the entire project — or you plug in anyway and let the local grid figure out the consequences. Both options are expensive. One is financially expensive; the other is politically expensive.
Owning the developer is a way to reduce both.
Why buy Intersect, specifically?
Intersect is not just a solar-and-storage shop. Alphabet describes it as an infrastructure provider that co-locates energy-intensive loads with dedicated gas and renewable power generation — essentially an “industrial park” model built around firm supply.
On its portfolio page, Intersect boasts $15bn of infrastructure operating or under construction, with 10.8 GW expected to be in operation or construction by late 2028. It lists utility-scale projects across Texas and California, including a large build in Haskell County, Texas — the same county Alphabet highlights for its co-located site — with 840 MW of solar PV and 1.3 GWh of battery storage under construction.
That matters for three reasons.
First: development is now a competitive moat. In many markets, the rare asset is not land or modules. It is a viable interconnection path, a community that will tolerate construction, and a team that knows how to navigate permitting, transmission upgrades and offtake structures. Alphabet is, in effect, buying organisational muscle memory.
Second: the new product is “power + speed”, not “green energy”. Intersect has marketed a “power-first” approach, where data-centre load is anchored to new generation so that the campus comes online alongside its own supply, easing strain on the grid. In a 2024 release about its partnership with Google and TPG, Intersect said the model was designed to catalyse $20bn in renewable power investment by the end of the decade; it also announced a funding round of more than $800m led by Google and TPG Rise Climate. The logic is simple: synchronise the two construction timelines — data centre and power plant — instead of hoping the queue cooperates.
Third: this is a political hedge. When policymakers start floating special tariffs and emergency auctions to make data centres finance new generation, owning the supply pipeline can look less like a climate gesture and more like compliance-by-design.
This is where the deal becomes bigger than Alphabet. It points to a new pattern: hyperscalers drifting towards quasi-utility behaviour, whether they admit it or not.
From “buying clean” to “building the grid around us”
For a decade, Big Tech’s energy story was written in contracts: long-term PPAs, renewable energy certificates, and, more recently, 24/7 “carbon-free energy” matching. Those tools were designed for a world where the grid could absorb new load without a fight.
That world is ending.
The grid is now a contested space where transmission is constrained, permitting is slow, and communities increasingly ask why they should accept higher bills, more substations and more water use so that someone else can train models.
In that environment, the next competitive advantage is physical: who can build firm capacity faster, where it is needed, without shifting costs onto everyone else?
Alphabet’s own press release frames the acquisition in exactly those terms: unlocking “abundant, reliable, affordable energy supply” to build data-centre infrastructure “without passing on costs to grid customers.” That sentence is doing a lot of work. It is both a promise and a warning: the public is watching, and the old, opaque utility-deal model is losing legitimacy.
(We’ve been tracking the downstream consequences — including why old fossil “peaker” plants are being kept alive by AI-era power prices — in The peaker paradox.)
The climate trade-off: faster renewables, or more gas?
There is a temptation to read this as good news for decarbonisation: a cash-rich company buying a developer should mean more wind, more solar, more storage, faster.
Sometimes it will. Intersect’s portfolio is heavy on solar and batteries, and co-location can reduce the need for long-distance transmission by placing load nearer generation.
But the AI power crunch is also reviving an “all of the above” energy strategy. The IEA’s analysis of electricity physically consumed by data centres (not contractual claims) suggests that to 2030, renewables meet nearly half of additional demand — but natural gas and coal together still meet over 40% of the additional electricity demand from data centres.
In the US, the IEA estimates natural gas already supplies a large share of data-centre electricity and adds significant incremental generation through 2030.
Intersect’s model explicitly includes gas alongside renewables. That does not mean Alphabet is abandoning clean energy ambitions. It does mean the company is acknowledging what grid operators have been saying for months: data centres do not run on annual averages. They run on hourly reliability.
This is why “owning the pipeline” matters. It gives a hyperscaler freedom to choose a portfolio that works for its uptime requirements — even if the portfolio looks less pure than the marketing slogans.
The climate question is therefore not whether Big Tech will buy clean energy. It will. The question is whether its need for firm power accelerates a new buildout of gas infrastructure — and whether carbon capture, long-duration storage and advanced geothermal arrive fast enough to keep that buildout from locking in emissions.
A quieter shift in market power
There is another implication that is easy to miss: buying developers could change who controls the future shape of the grid.
Utilities and independent power producers have historically been the key actors in deciding what gets built and where. If hyperscalers start buying development pipelines — or building their own “energy parks” and interconnection assets — the centre of gravity shifts.
Three consequences follow.
1) The queue may become even more unequal. If the scarcest resource is a viable interconnection path, the richest players will buy it. Smaller industrial loads and ordinary customers will not. That is politically combustible.
2) The cost-allocation fight gets sharper. Policymakers are already pushing to prevent grid-upgrade costs from being socialised across ratepayers. If Big Tech owns more of the upstream build, it will have more leverage — and more scrutiny.
3) Community impact becomes central, not peripheral. Intersect’s partnership language emphasises coordination with “grid planners, operators, and communities” and points to local economic development, jobs and tax revenue. That is not charity. It is project risk management.
What happens next
Alphabet’s Intersect deal is unlikely to be the last of its kind. Once one hyperscaler demonstrates that owning a pipeline reduces delay risk, competitors will ask why they should pay a premium in the market when they can internalise the capability.
Watch four signals.
First, more “capability acquisitions”. Deals that look expensive against the cash flow of operating assets — because they are not priced on cash flow. They are priced on speed.
Second, more co-location. Expect more projects that pair data centres with dedicated generation and storage, especially in markets where transmission upgrades are slow.
Third, a new regulatory bargain. If states and grid operators adopt “pay your way” tariffs or emergency procurement structures, tech companies will respond by building (and owning) more of the supply chain.
Fourth, a carbon-market squeeze. As hyperscalers expand power and emissions footprints, they are also racing to lock up durable carbon removal. That creates price pressure and scarcity for everyone else — a trend we explored in The Data Center vs. the Planet.
In the end, the Intersect deal is a reminder that the AI era is not only a story about chips and models. It is a story about steel, substations and permits — and about whether the world can build clean power as quickly as it is building compute.
Alphabet is betting $4.75bn that it can.
