We are excited to share more about our work to maximize its reach and beneficial impact. This includes posts that highlight our thinking, resources that explain what we are building, and sharing novel datasets that we think can benefit the work of others. Please stay tuned for updates or reach out to learn more.
Data center locations today show a moderate skew toward grid regions with lower carbon intensity—but how much does this really influence site selection?
At CERAWeek 2025, data centers were very heavily discussed, including topics such as geopolitics, data sovereignty, monetization opportunities for previously-stranded energy supplies, and the overall uncertainty data centers have introduced into energy demand forecasting.
Climate change factored into conversations on the role of natural gas, nuclear, and geothermal, but global warming was generally just one of the many complexities described by the developers, funders, and policymakers working on the future of compute.
Data center locations today show a moderate skew toward grid regions with lower carbon intensity—but how much does this really influence site selection?
Speed to market, economics, data sovereignty, and end-user latency are some of the most critical factors for developers, operators, and financiers of compute / AI infrastructure.
Focus on carbon intensity of compute will continue to grow however, particularly as policymakers and the public become ever more aware of the rising energy demand and climate impact of data centers.
But as the annual solar PV capacity factor data above shows, today’s data centers show little sun-seeking behavior. Hourly capacity factor data shows a similar pattern.
This makes sense given the challenges created by solar intermittency and limited incentive to experiment with new operating strategies to date.
But in the right locations, this will likely shift going forward due to:
⬡ Falling delivered costs of solar power
⬡ Increasing market and behind-the-meter supply options
⬡ Corporate GHG goals and public attention shaping demand
⬡ Proving out reliable data center operating strategies that take advantage of intermittent power and/or load shifting.
Annual solar PV capacity factor data above shows, today’s data centers show little sun-seeking behavior. Hourly capacity factor data shows a similar pattern.
Today, water basin stress appears to play little role in data center site selection. The distribution of locations with data centers and all global locations are well aligned, supporting the notion that developers are far more concerned with other factors.
However, regional water stress may become a more significant consideration over time as water stress in various regions increases 🥵 and the environmental impact of data centers faces greater public scrutiny 🧐.
If so, this could influence site selection strategies or drive shifts in the cooling technologies—for example, accelerating moves from water-intensive chillers, evaporative and adiabatic cooling toward air-cooled and direct-to-chip liquid cooling solutions in some locations, especially as rack densities continue to rise.
Stay tuned for future posts where we’ll look at the connection to factors as physical access to water and water tariffs.
Today, water basin stress appears to play little role in data center site selection. The distribution of locations with data centers and all global locations are well aligned, supporting the notion that developers are far more concerned with other factors.
Proximity to the grid is a make-or-break factor for data centers, driving key factors such as capital costs and time to market. So this tight clustering is no surprise—but how does it compare to other site criteria?
The Othersphere platform has the answer—bringing together dozens of spatial and time series data layers, to pre-model billions of projects, at millions of locations worldwide.
For those working on developing, funding, or utilizing data centers, this search engine for sustainable infrastructure saves time, reduces risk, and enables better decisions.
We’ll shortly be launching data center coverage within the Othersphere Explorer tool, and in advance will share dozens of similar posts that highlight the many criteria that enable high performance compute infrastructure.
Prior to launch we’ll release an ebook containing this insight series, offering a unique, data-driven guide to the ideal characteristics of world-class data centers.
Our data center module is currently in closed beta, but if you would like to be notified when this ebook is released—or request a demo today—we’d love to hear from you.
Proximity to the grid is a make-or-break factor for data centers, driving key factors such as capital costs and time to market. So this tight clustering is no surprise—but how does it compare to other site criteria?
At Othersphere we are constantly working to make it easier for our users to site, design, and vet infrastructure assets. For example, we recently integrated our Global Land Cost Model 🌍 , which estimates land values globally. This 5-tier ensemble model is trained on terabytes of data, considering factors such as infrastructure, land use, topography, population density, and GDP.
Included within our Explorer tool, this new data helps users more quickly assess one of the many (many) considerations that go into infrastructure developments. In this case, land cost can have a significant impact on project economics, especially for assets with large physical footprints and more modest hardware or operating costs.
Curious about the key factors influencing the results? The example SHAP (SHapley Additive exPlanations) graph provides a breakdown of the six most impactful drivers for a rural landscape, highlighting GDP, population density, and water stress as important drivers of land value.
We recently integrated our Global Land Cost Model, which estimates land values globally and adds this new level of depth into results in Othersphere Explorer.
Equipment sizing and operating strategy LCOH optimizations for hydrogen electrolysis plants are cumbersome and time-consuming. It can take days to weeks to get the data, with each optimization run often taking hours. This limits analytical efforts to a handful of potential sites, leaving many opportunities unexplored.
The result — sub-optimal capital allocation and higher risks, with countless missed opportunities.The difference between initial cost calculations versus hourly optimizations can be dramatic, and highlights location as a critical element of project success.
Othersphere has revolutionized this process by combining the power of our underlying spatial economics platform with neural networks for dimensionality reduction and surrogate modeling.
Our global search engine now recommends pre-optimized projects in seconds, having already run an hourly optimization considering:
All of this is delivered through the Othersphere Explorer tool, which provides a novel, bottom-up view of potential hydrogen projects, evaluating key factors such as production costs, emissions, available offtakers, and fit with local surroundings.
Introducing the world’s first global optimizer for hydrogen projects
Please reach out if you would like to learn more about Othersphere, our products, and opportunities to partner in accelerating global industrial decarbonization.
