A hillside above the site has already been stripped and terraced. But mounting opposition from environmentalists and residents, including litigation over water supplies and proximity to a wildlife sanctuary, is creating friction the company and its state backers didn’t anticipate at this scale.
Reading that story alongside Capacity’s own reporting on Central Texas’s hyperscale water pressures produces an uncomfortable sense of déjà vu. Different continent, different regulator, different water system, but the underlying pattern looks remarkably similar. Worth asking directly: has community pushback stopped being an occasional hazard of hyperscale development and become a predictable, structural cost of it?
The same regulatory lag, twice
In Texas, Dr Margaret Cook of the Houston Advanced Research Center told Capacity that the core problem isn’t resource scarcity alone, it’s a mismatch between the scale of what’s being built and the regulatory apparatus around it. Projects with the footprint of major industrial developments are being approved with what she called the mentality of a commercial building, without the community advisory structures that typically accompany industrial-scale projects elsewhere. Her clearest example: a Texas county judge who testified he didn’t know a hyperscale data centre was under construction in his own county until roads needed closing.
The Andhra Pradesh case follows a near-identical shape. Public interest litigation now before the Andhra Pradesh High Court alleges the project was fast-tracked without adequately weighing risks to water supplies and wildlife, a claim the state government denies. Separate reporting has cited satellite imagery suggesting construction began before environmental clearance was fully granted, and three further cases are reportedly pending before India’s National Green Tribunal. In both geographies, the complaint isn’t really about whether data centres should exist, it’s about approval processes moving faster than the oversight built to match them.
Water is the first flashpoint everywhere
Kamil Cook of Public Citizen told Capacity that water is almost always residents’ first concern in Texas, given how much of the state sits under drought conditions, closely followed by a sense of powerlessness over decisions already made. The same framing has surfaced verbatim in Visakhapatnam, where activists have marched holding banners reading “We cannot drink DATA,” a slogan that could have come from either state’s protest movement.
The specifics differ. Andhra Pradesh’s contested site sits near the Kambalakonda Wildlife Sanctuary, home to leopards and pangolins, with litigation focused on noise and construction impact at close range. Central Texas’s flashpoints, according to Cook’s research, are less about a single sanctuary and more about diffuse hyperscale growth landing in unincorporated counties such as Williamson, Caldwell and Hays, specifically because those areas carry lighter regulatory oversight than incorporated cities. Different mechanisms, same underlying dynamic: development gravitating toward wherever oversight is thinnest.
The jobs promise, and the scepticism that follows it
Google’s Andhra Pradesh project is projected to create up to 188,000 jobs, a figure state officials have used to frame the investment as transformational. Wider reporting on India’s data centre boom, including coverage of Google and Microsoft projects elsewhere in the country, has documented similar friction between big tech’s investment framing and how affected communities, particularly farmers whose land borders these developments, actually experience the trade-off. That echoes a pattern familiar from Texas, where Kamil Cook notes that tax abatements offered to attract data centres often become a further source of resident frustration once a company drawing millions of gallons of water annually is also seen receiving a tax break.
A pattern showing up well beyond these two sites
Neither Texas nor Andhra Pradesh is an isolated case. Capacity’s reporting on water rights and permitting risk found that more than 40% of planned and existing US data centres sit in areas already classified as high or extremely high water scarcity, and that roughly $156 billion in US projects were delayed or cancelled in 2025 alone as a direct consequence of local resistance. More than 300 data centre-related bills were introduced across 30 US states in the first six weeks of 2026, most pivoting away from tax incentives and towards mandatory water disclosure and permitting requirements. Maine has already moved to pause new construction until November 2027, and Seattle’s mayor has said the city is exploring a similar block. The same report found 84% of proposed UK data centre developments are planned in water-stressed areas. Andhra Pradesh isn’t a one-off flashpoint, it’s the latest entry in a list that’s growing on multiple continents at once.
The industry’s counter-argument
It’s worth acknowledging the industry’s pushback on how this issue tends to get framed. As Capacity has previously reported, operators and vendors increasingly argue that raw consumption figures obscure the more relevant question of efficiency, and that newer closed-loop cooling architectures reuse water rather than simply consuming it. That’s a fair technical point, and it’s consistent with what Cook told Capacity about dry cooling and water reuse being the fastest-scaling mitigation levers in Texas. But it doesn’t resolve the trust deficit driving community opposition, which, as both Cooks and the wider reporting above suggest, is rooted less in the physics of water use and more in how little communities are told before ground is broken.
What technology can and can’t fix
On the mitigation side, Capacity’s Texas reporting found real momentum behind cooling technology as a partial answer. Reza Azizian of cooling start-up Ferveret pointed to the shift from air cooling toward direct-to-chip liquid cooling, and ultimately to phase-change approaches like his own, as a way to materially cut water and energy overhead. But Cook’s research carries an important caveat that applies equally to any market chasing a quick technical fix: switching to dry cooling can raise energy consumption by as much as 20%, and if that additional power comes from a water-cooled plant, total water consumption can rise rather than fall. Technology alone doesn’t resolve the trust problem, because the trust problem is fundamentally about transparency and timing, not engineering.
The pattern, not the incident
Google’s public response in India, offering sound-dampening measures and describing itself as aiming to be “a quiet, unobtrusive neighbour,” reads as a reasonable mitigation step but not a structural answer to the underlying complaint. Cook’s central lesson from Texas, drawn from Corpus Christi’s water crisis, is that public trust and resource planning have to be built years ahead of a shortage, not assembled reactively once opposition has already organised. Andhra Pradesh’s state government is now managing exactly that reactive dynamic in real time.
None of this means hyperscale AI investment is slowing down, in either country. It does suggest that the industry’s current default, moving fast, securing state-level political backing, and addressing community and environmental concerns after ground has broken, is producing a recognisable and repeatable pattern of friction rather than a series of unrelated local disputes. Whether operators start treating community engagement and transparent resource reporting as a cost of entry rather than a reputational afterthought may end up mattering more to the pace of the AI build-out than any single cooling technology.
RELATED STORIES
ITW Asia 2026
ITW Asia brings together the whole connectivity and digital infrastructure industry to get business done. Join 1700+ leaders from carriers, MNOs, cloud solution providers, hyperscalers, content service providers, data centres, satellite operators, investors, regulatory authorities and more, to define the future of connectivity in Asia.
