Google advances data center push

Data centers are the story here — not just because Google wants another one, but because every new AI buildout now drags a giant electricity question behind it. That is the real stake. Communities want jobs and tax base, but they do not want higher utility bills, strained water systems, or a warehouse of servers dropped into town on vague promises. What changed is that Google is now advancing a specific Michigan project while also signaling more AI infrastructure ambition in India, which makes the scale of its expansion a lot harder to dismiss. ### What is Google actually building in Michigan? The Michigan piece is a proposed hyperscale campus in Van Buren Township, near Detroit Metro Airport. Google confirmed in March that it is the company behind the project long nicknamed “Project Cannoli.” The site under discussion is roughly 1.5 million square feet on vacant land, and the planned power draw is enormous — about 1 GW, which is in the range of a major city’s electricity use. (blog.google) ### Why are people skeptical? Because “we’ll build a giant data center, but don’t worry, it’s different” is exactly the kind of pitch that now gets challenged. Metro Detroit has already seen fights over data centers, especially around ratepayer risk, land use, and whether promised grid upgrades really stay isolated from everyone else’s bills. Google and DTE are trying to avoid that backlash by sending this one through a contested regulatory process instead of a faster, lighter-touch route. (michiganpublic.org) That gives outside experts and the public a formal chance to push on the assumptions. ### What is Google promising on power? Basically, Google’s argument is: we are not just plugging into the existing grid and hoping for the best. The company says the data center would be paired with 2.7 GW of new grid resources — solar, advanced storage, and demand flexibility — under a deal structure with DTE meant to add capacity rather than simply reshuffle scarcity. Google also says it will cover its own electricity and infrastructure costs, and it set up a $10 million Michigan Energy Impact Fund tied to affordability programs. (michiganpublic.org) ### Does that settle the issue? Not really. The catch is that “new clean resources” and “no burden on ratepayers” are promises that only matter if regulators accept the contracts and if the buildout happens as described. A 2.7 GW resource package sounds huge because it is huge — roughly a quarter of DTE’s current grid capacity by one public-radio estimate. But that scale is also why critics are not taking the company’s framing on faith. (blog.google) ### What happened in India? A separate signal came from New Delhi on May 8, when IT minister Ashwini Vaishnaw said Google is exploring investments in India across AI infrastructure and the manufacturing of servers and drones. That matters because it pushes Google’s India role beyond cloud services and software into the physical stack — the buildings, machines, and supply chain that AI actually runs on. (michiganpublic.org) ### Why pair these two stories together? Because together they show the same thing from two angles. Michigan is the local version of the AI buildout problem — who pays, who benefits, and where the power comes from. India is the global version — where Google wants more compute, more hardware capacity, and more control over deployment. Put them together and you get a clearer picture of hyperscaler demand staying high, not easing. (financialexpress.com) ### How big is the broader spending wave? Big enough that individual projects no longer look isolated. Alphabet lifted its 2026 capital-expenditure guidance to $180 billion to $190 billion in late April, up from the earlier $175 billion to $185 billion range, and said 2027 spending should rise again. More than half of Google’s machine-learning compute investment this year is expected to go to Cloud. So when Google advances one campus in Michigan and explores more infrastructure in India, that fits a much larger pattern. (blog.google) ### Bottom line? Google is not just adding a building. It is testing whether hyperscalers can keep scaling AI infrastructure while convincing local communities that the grid, the water, and the bill impacts are under control. Michigan is one proving ground. India may be the next expansion lane. (blog.google) (cnbc.com)