How NvironX Built a Data Center Siting Intelligence Platform in 26 Hours

This past weekend, Impulse AI was proud to sponsor DataHacks 2026, hosted by UCSD's Data Science Student Society. After 26 hours of building, we walked away genuinely impressed, not just by the polish of the work, but by the ambition of the problems teams chose to solve.

First place in the Impulse AI challenge went to NvironX, a team that took on one of the quietest but most consequential questions in modern infrastructure: where should we actually build data centers?

Here's their story.

The Problem: Billions at Risk, No Way to Price It

Every time you stream a video or prompt an AI model, a data center somewhere is doing the work. Siting these facilities is taken seriously. Teams weigh community sentiment, environmental constraints, regulatory exposure, and utility capacity carefully before committing.

The issue isn't a lack of diligence. It's a lack of data.

There's no unified public record of data center projects that have been approved, killed, or stalled, and why. That absence shapes the entire planning process. Teams invest months of work and significant capital into approvals, land options, and utility negotiations, then watch projects collapse for reasons that a better signal could have flagged early. The goal isn't to avoid analysis. It's to put a probability on whether a site will actually survive the process.

NvironX set out to build that signal.

The Solution: A Unified Intelligence Layer for Siting

NvironX is a geospatial intelligence platform that predicts whether a proposed U.S. data center will succeed or fail. The system ingests public records, utility grid data, environmental indices, and public sentiment signals, then returns:

• An overall risk score plus sub-scores across Community, Regulatory, Environmental, Utility, and Governance

• The top contributing factors driving the score

• Mitigation recommendations and alternative site suggestions

• An interactive map-first UI, exportable risk reports, and an API

In the demo, the team scored an Oregon address against a dataset of more than 3,700 renewable energy sites, surfacing both positive signals (water availability, renewable proximity) and risk signals on a single interactive map, along with a correlation view linking data center success to renewable energy proximity.

How They Built It, With Impulse at the Core

NvironX used Impulse as their core computational engine, processing high-velocity geospatial parameters to calculate "System Absorption" scores, the central predictive signal behind the platform. They generated SHAP values on top of the model to explain why a site scored the way it did, turning raw predictions into something a planner could actually defend in a meeting.

The notable thing here: the team shipped a working predictive platform in a little over a day without a dedicated ML engineer and without building model infrastructure from scratch. Impulse took them from multi-source, messy geospatial data to deployable predictions, which freed them up to spend their hackathon hours on the product itself, the UI, and the actual domain problem.

If you have data and a problem worth solving, Impulse handles the data prep, model selection, training, and deployment, so you can focus on what your tool actually does.

The Outcome: Retroactive Validation and First Place

The team validated the model retroactively against real data center projects that had been killed or stalled, and NvironX flagged them. Combined with a polished map-first UI, a credible multi-jurisdiction dataset, and a clear story for who would actually use it, it was enough to take first place in the Impulse AI challenge at DataHacks 2026.

Congratulations to the Team

To Romy Bornstein, Sarah Grover, and Roberto Echeverria Sosa, congratulations from all of us at Impulse. You took on a real problem worth hundreds of billions of dollars and came up with a genuinely innovative solution in a weekend. We can't wait to see where you take this next.