Kevin O’Leary has agreed to shrink the footprint of his planned Utah data center project, a move that signals how quickly large-scale AI infrastructure proposals can be reshaped once local politics, environmental concerns, and public scrutiny collide.
The update centers on Project Stratos, a massive facility O’Leary—best known as a star investor on Shark Tank—had previously outlined as spanning roughly 40,000 acres in and around the Locomotive Springs Waterfowl Management Area. According to a letter he sent to Utah Senate President J. Stuart Adams, O’Leary will remove 19,430 acres from the plan. In practical terms, that means the project’s size would be cut by about half, reducing the overall footprint to something closer to 20,570 acres rather than the original 40,000-acre vision.
For residents and activists who have been pushing back against the project, the change is significant not only because it reduces land use, but because it reflects a willingness to negotiate with state-level pressure rather than treating the proposal as a fixed blueprint. For Utah lawmakers, it also provides a clearer path to evaluating the project under evolving expectations—especially around water use, wildlife habitat, and the broader environmental footprint of data centers.
What makes this moment particularly notable is the timing. The downsizing agreement arrives just days after Adams publicly called for an even more dramatic reduction. In that earlier push, Adams asked O’Leary to slash the project’s size by 75 percent, which would have brought the footprint down to about 10,000 acres. Adams also urged the company to adopt technology intended to minimize water consumption—an issue that has become increasingly central to data center debates across the western United States, where drought conditions and water scarcity are not abstract concerns but daily realities.
In other words, O’Leary’s latest concession appears to land between the original plan and the most aggressive reduction Adams had requested. It’s not the full 75 percent cut Adams asked for, but it is a substantial adjustment that suggests the political pressure is working—at least enough to force a major redesign of the project’s land footprint.
The location matters. The project is planned in and around the Locomotive Springs Waterfowl Management Area, a setting that raises immediate questions about impacts to wildlife and habitat. Waterfowl management areas are typically established to support migratory birds and other species that depend on specific ecological conditions. When a data center proposal is placed near such a site, the conversation inevitably shifts from “how much power will it use?” to “what happens to the land and water systems that sustain the surrounding ecosystem?”
That shift is part of what has made the controversy around Project Stratos so persistent. Data centers are often discussed in terms of energy demand and grid capacity, but the physical reality of building them—clearing land, constructing roads and utilities, managing stormwater runoff, and sourcing cooling water—can create a different kind of environmental stress. In arid regions, cooling is especially sensitive. Even when facilities aim to use less water through advanced cooling methods, the question becomes whether those technologies are feasible at scale, whether they reduce water use enough to satisfy regulators and communities, and whether they introduce other tradeoffs.
O’Leary’s letter to Adams indicates that the acreage removal is a direct response to the concerns raised by the state leadership. While the public reporting around the letter focuses on the land reduction itself, the broader context suggests that the downsizing is tied to a larger set of expectations. Adams’ earlier request included not just shrinking the footprint, but implementing technology designed to minimize water consumption. That combination—less land, less water—reflects a common pattern in how communities and governments evaluate data center proposals: they want both immediate mitigation (reduce the area) and operational safeguards (reduce resource use during operation).
There’s also a political dimension to the story that goes beyond the numbers. Utah’s legislative leadership has been positioning itself as a gatekeeper for major development projects, particularly those that could strain local resources or alter sensitive landscapes. When the Senate President calls for a 75 percent reduction, it’s not merely a technical suggestion; it’s a signal that the state expects meaningful changes before it will treat the project as acceptable.
O’Leary’s decision to remove 19,430 acres can be read as an attempt to keep the project moving while acknowledging that the original plan is no longer politically viable in its current form. Large projects often face a choice: either dig in and risk delays, legal challenges, or reputational damage—or adjust early enough to preserve momentum. In this case, the adjustment appears to be aimed at preventing the project from becoming a prolonged standoff.
Still, the downsizing does not automatically resolve all concerns. Cutting the footprint in half may reduce the intensity of land disturbance, but it doesn’t eliminate the fundamental questions about water use, wildlife impacts, and the long-term sustainability of hosting large-scale computing infrastructure in a region where environmental constraints are real. Even a smaller footprint can require significant construction activity, new utility connections, and ongoing operational demands.
This is where the unique angle of the story emerges: the project is evolving in real time, and the evolution is being shaped by community pressure and state-level demands. That dynamic is increasingly common in the era of AI infrastructure, where the pace of technological adoption often outstrips the pace of local planning and environmental assessment. Communities frequently find themselves reacting to proposals that arrive with enormous scale and urgency, leaving limited time for thorough public engagement.
But in this case, the pressure seems to have produced a tangible outcome quickly enough to matter. The fact that O’Leary agreed to remove nearly 20,000 acres suggests that the project’s planners were able—or willing—to reconfigure the site layout rather than treating the initial acreage as untouchable. That willingness is important because it implies the project is not simply a fixed industrial plan; it’s a proposal that can be redesigned based on feedback, at least to a point.
Another layer to consider is how this story fits into the broader national debate about data centers and AI. Across the country, data centers are being built to support cloud services, streaming, and increasingly AI workloads. The demand is real, and the economic incentives are obvious: jobs, investment, and tax revenue. But the externalities—water use, land impacts, energy consumption, and the strain on local infrastructure—are also real. As a result, data center proposals are increasingly treated like environmental and political issues, not just business developments.
Utah is not alone in facing these tensions. Western states have been grappling with water scarcity for years, and many communities are wary of new industrial loads that could worsen drought conditions. Meanwhile, environmental groups often argue that data centers should be held to higher standards because they are expanding rapidly and because their growth is tied to energy-intensive computing. Even when companies claim efficiency improvements, critics argue that efficiency gains can be overwhelmed by the sheer scale of expansion.
O’Leary’s situation illustrates how those debates play out when a high-profile figure is involved. Public attention tends to intensify scrutiny, and when a project is associated with a celebrity investor, it can become a symbol—either of innovation or of disregard, depending on who tells the story. In this case, residents and activists have raised concerns, and the state leadership has responded with direct calls for reductions.
That’s why the letter to Adams matters beyond the acreage itself. It represents a formal acknowledgment that the project must adapt to local governance and public pressure. It also suggests that the state is not waiting passively for environmental review processes to unfold; instead, it is actively shaping expectations upfront.
At the same time, it’s worth noting that the downsizing request from Adams was more aggressive than what O’Leary ultimately agreed to. That gap raises a question that will likely persist: how much further will the project need to change before it satisfies the state and the community? If Adams’ earlier target was around 10,000 acres, then the revised plan—still roughly double that—may still be viewed by some as too large, especially given the proximity to a waterfowl management area.
In other words, this may be a step forward, but it may not be the final word. Major projects often undergo multiple rounds of negotiation, redesign, and compliance adjustments. The first concession can be followed by additional requirements once regulators and stakeholders examine the revised plans in detail—particularly if the project triggers further questions about water sourcing, cooling methods, habitat impacts, traffic, and long-term operational practices.
The mention of water consumption technology is also a clue that the conversation is shifting from land size alone to operational sustainability. Data centers can use different cooling approaches, including air cooling, evaporative cooling, and hybrid systems. Each approach has different implications for water use and energy efficiency. The state’s emphasis on minimizing water consumption suggests that Utah wants assurances that the project’s operational design will align with regional water realities, not just that the land footprint will be reduced.
If the project’s planners can demonstrate credible water-saving measures—through technology choices, water recycling, or other strategies—then the downsized footprint may become easier to justify. But if water use remains a concern, the project could face continued resistance even after the acreage reduction.
There’s also a broader lesson here about how AI infrastructure is being governed. The story shows that local and state authorities can exert influence even when the project is backed by significant capital and high-profile investors. It also shows that public pressure can translate into concrete policy outcomes, at least in the form of negotiated changes.
That doesn’t mean the process is smooth or guaranteed. It means the political system is responding to the scale of the challenge. As AI demand grows, more communities will likely face similar proposals, and more lawmakers will likely adopt similar strategies: demanding reductions, requiring water-minimizing technologies, and insisting on transparency about environmental impacts.
For residents near Locomotive Springs, the immediate takeaway is that their concerns have produced a measurable change. For O’Leary and his team, the takeaway is that the project’s original scope is no longer acceptable without modification. And for Utah