Back to BlogAnalysis · AI Computing Center Operations

    Project Landman in Odessa: How to Prove a Closed-Loop Cooling Promise

    STN Incorporated told Ector County commissioners that Project Landman, its AI data center campus in south Odessa, Texas, runs a sealed closed-loop cooling system that will not draw local water, and that it will publish impact studies. A commitment like that is easiest to keep, and to believe, when the operator can show the readings behind it.

    November 2026 6 min readDa Li
    Project Landman Odessa: Proving Closed-Loop Cooling Claims: illustration of a cooling fan with airflow
    // 01

    What happened in Ector County

    NewsWest 9 reported this week that STN Incorporated CEO Sabur Mian presented Project Landman to the Ector County Commissioners' Court in front of a packed courtroom. The campus sits on former industrial land south of Interstate 20 between Grandview and Dixie. During public comment, residents raised its proximity to neighborhoods and possible effects on local water supplies, noise levels and light pollution.

    Mian answered that the facility uses a sealed, closed-loop cooling system that will not draw from local water supplies, and said the company is not seeking tax abatements. He also said STN will publish an impact study, an environmental impact study and a job study. After public comment, the commissioners went into executive session to discuss the feedback and a proposed development agreement. Construction continues while the court has yet to take formal action on that agreement.

    Water had come up before. Ahead of an earlier court session, County Judge Dustin Fawcett told the Odessa American that the project is using a factory-made coolant for its closed loop, with no local water for cooling. In an earlier KWES report on the county's authority, Fawcett said more data centers are likely to come to the region and that water use is where he has to engage with them. Residents in that report also raised noise, grid capacity and a lack of public information about the project.

    // 02

    Why it matters for an AI data center operator

    A closed loop moves the water question. The coolant is charged once and circulates between the cold plates or heat exchangers and the coolant distribution units (CDUs), so routine operation should need little or no water for cooling. The heat still has to leave the site through some form of heat rejection, and the reports so far do not describe how Project Landman handles that. Whatever the design, the commitment can be checked, because a sealed loop leaves a clear signature in operating data.

    That signature is what residents and a county can reasonably ask to see, month after month. Most of it falls into four groups of readings:

    • Water: metered site water use reported as WUE, the water used per kilowatt hour of IT energy. A sealed loop shows up as a low, flat line.
    • Loop health: CDU status, supply and return temperature, flow and pressure differential. Steady pressure with steady flow is what a tight loop looks like, and a slow pressure decline is a loss someone has to find.
    • Leaks: leak events from sensors under racks and along pipe runs, each with a location and a recorded response.
    • Load: rack power and site PUE, so the cooling readings can be read against the heat the IT equipment was producing at the time.

    In most liquid-cooled halls the CDUs and facility systems already expose many of these points. The work is collecting them in one place, keeping the history and turning it into a report a commissioner can read. A published study describes the design. Measured trends show how the site runs once its halls fill with GPUs.

    PROMISE AND EVIDENCESAID IN PUBLICSHOWN IN DATANo local waterfor coolingWater use, WUElow, flat trendSealed loopfactory coolantPressure, flowstable readingsno leak eventsPublished studiesimpact, environmentMeasured reportstrends over time
    Each commitment reported for Project Landman, paired with the operating data that would show it over time.
    // 03

    What the operator sees in Sensaka

    Sensaka for AI Data Centers keeps facility, cooling, power and accelerator data in one inventory, so the evidence above comes from a single record. Its liquid cooling monitoring reads CDU operating status, supply and return temperatures, pressure differential, flow indicators and leak events from supported CDU and facility interfaces, organized by site, room, cooling zone and rack. A leak event is placed beside the racks and equipment around it, and the responsible team gets an operational ticket with the response recorded.

    Energy data becomes PUE and WUE for the site and per zone, live and as a trend, next to rack power from PDUs and UPS and circuit data. Historical trends, alarms and operational records stay in the platform, which is what a report to the county draws on. The exact data points depend on the interfaces the cooling equipment and facility systems expose.

    For a quick answer, the AI Cockpit takes the question in plain language: "Is the liquid-cooling loop healthy: temperature difference, flow and leaks?" or "Which racks or zones are near their rated power?" Each question is matched locally to a preset operational question and gets its own board, and one click renders the board as a 30-second data video for a recurring briefing. No external AI model is involved.

    FROM READINGS TO EVIDENCECDUsstatus, temps, flowLeak sensorsevents by locationPDU, UPSrack power, circuitsEnergy datasite and zonesOne record per zonesite, room, zone, rackPUE, WUElive and trendAlarmsand ticketsCockpit board30 s videoMonitoring only: control stays with site procedures
    Cooling, leak, power and energy readings go into one record per zone, and come out as PUE and WUE trends, alarms with tickets, and cockpit boards.

    This is monitoring. Sensaka does not control CDUs, pumps or valves, and any control or maintenance action stays with the site's approved procedures. It works with NVIDIA, AMD and other accelerators.

    // 04

    Takeaway

    The developer of Project Landman has made specific commitments in public: a sealed closed loop, no local water for cooling and published studies. Counties elsewhere are asking new AI factory projects similar questions. The operators who answer them most easily will be the ones who measure water, loop pressure, flow, leaks and power from the first day of operation and can show the trend when asked.

    // FAQ

    Frequently asked questions

    What is Project Landman?

    Project Landman is STN Incorporated's AI data center campus in south Odessa, in Ector County, Texas. NewsWest 9 reported that the site is former industrial land south of Interstate 20 between Grandview and Dixie. Construction is under way while the Ector County Commissioners' Court considers a proposed development agreement.

    Will the Project Landman data center use local water?

    STN CEO Sabur Mian told county commissioners that the facility uses a sealed, closed-loop cooling system that will not draw from local water supplies. County Judge Dustin Fawcett told the Odessa American that the project uses a factory-made coolant for its closed loop, with no local water for cooling. STN has said it will publish an impact study, an environmental impact study and a job study.

    How can a closed-loop cooling claim be checked?

    Through operating data kept over time: metered water use and WUE, CDU supply and return temperature, flow and pressure differential, leak events, and rack power and PUE. A sealed loop shows low, flat water use and steady loop pressure. A change in any of these shows up as an alarm and a record that can be reported.

    See CDU readings, leak events, rack power and PUE and WUE trends for the same zones in one record. Start a free 30-day pilot on your own hardware, inside your own network.

    Start a Pilot →