2 Day Congress Live & Online: 9 & 10 February 2027
Real Lessons In Balancing Water, Community & Power Trade-Offs
9 & 10 February 2027, Dallas Tx.
Real Lessons In Balancing Water, Community & Power Trade-Offs
9 & 10 February 2027, Dallas Tx.
One of the most striking observations from chairing this conference was just how much the industry had changed in the seven months since our previous AI Data Centre Power and Cooling Conference in March 2026.
Back then, water was already becoming an important consideration. By October, the conversation had moved considerably further.
Advances in low-water cooling technologies have continued at pace. At the same time, questions about freshwater availability, drought resilience, increasing industrial demand and the reliability of water-consumption forecasts have become much harder to ignore.
We have also seen greater political scrutiny, proposed moratoriums, development pauses and increasingly vocal opposition in communities concerned about the consequences of large-scale data centre expansion for their local water resources.
A recurring question throughout the conference, therefore, was not simply how much water data centres will consume, but how reliably we can forecast that demand. Where will the water come from — potable supplies, reclaimed water or other alternatives — and how will evaporative versus non-evaporative cooling technologies influence consumption?
Just as importantly, how will peak demand affect local water supplies during prolonged drought, and what level of resilience and, perhaps more importantly, robustness will be needed to ensure supplies remain dependable under the most challenging conditions?
Multiple speakers flagged up that a facility's average annual water consumption may look perfectly manageable on paper. But the picture can change considerably during an extended drought, particularly when several developments, existing businesses, agriculture, and residential customers draw from the same resources.
Likewise, a region may have substantial water resources but insufficient infrastructure to collect, treat and deliver the volumes required. Conversely, an impressive new treatment plant or pipeline does not, by itself, create additional water.
Another difficulty is that many of the assumptions being used to plan future infrastructure remain uncertain. Cooling technologies are changing, proposed developments may not all proceed, and the water requirements of future facilities could look quite different from those being discussed today.
That creates a genuine problem for utilities. They are being asked to make expensive, long-term investment decisions against demand forecasts that may change substantially before the infrastructure is completed.
What the speakers, attendees and I found particularly revealing was how often the discussions returned to the same fundamental uncertainties. Developers, utilities, regulators, researchers and technology providers approached the challenges from different perspectives, yet many of the underlying questions were closely connected. Encouragingly, as the following summaries demonstrate, the conference made real progress in bringing greater clarity to several of these issues.
The Q&A discussions were especially valuable. This was where we could challenge some of the assumptions, dig into the numbers and distinguish what is already happening from what remains an ambition or an interesting technical possibility.
Overall, the conference provided a much clearer picture of what the industry can demonstrate today, what depends heavily on local circumstances and where important questions remain unanswered.
The Breakfast Briefing — Powering AI in a Thirsty World
The conference began promptly at 8:15 a.m. with a breakfast briefing, Powering AI in a Thirsty World, delivered by Professor Shaolei Ren of the University of California, Riverside.
Professor Ren provided an excellent foundation for the discussions that followed, examining why the water footprint of AI infrastructure varies so considerably between regions, cooling technologies and operating conditions.
One of the important points emerging from his analysis was the danger of making simplistic comparisons between data centres without understanding where they are located, how they are operated and how their water consumption is calculated.
Shaolei's presentation also provided some useful context for the wider debate about direct and indirect water consumption, which several later speakers explored in considerably more detail.
The breakfast briefing transcript is available through the conference website, and I would particularly recommend reading it. There is some excellent foundational material here that helps make sense of the more detailed discussions that followed.
About These Report Highlights
What follows is a summary of my assessment of the principal findings, important data, strategic insights and wider industry implications emerging from the presentations.
The intention is to give readers a clear picture of the most consequential issues raised over the two days, without requiring them to work through every individual session.
The full conference media package goes considerably further, including video recordings of the presentations, the comprehensive conference intelligence report, full transcripts of the Q&A discussions and additional technical, commercial and strategic analysis.
For anyone wanting to examine the evidence in greater detail, that is where the complete material can be found.
The Central Findings — What Really Matters?
Before turning to the individual presentations, particularly the legislative update delivered by Andrew Parks from the Texas Senate Committee on Water, Agriculture, and Rural Affairs, it is worth drawing together the main conclusions.
For those short on time, I would argue that much of the conference came down to three connected problems.
The first is the scale and reliability of future water-demand forecasts.
Texas and other fast-growing regions are being presented with substantial development pipelines, but it remains difficult to establish how much of the proposed capacity will be built, which cooling technologies will be used and what the resulting water demand will actually be.
The second is the difference between water availability and dependable water supply.
Having water somewhere within a region does not necessarily mean a utility can deliver it to a particular development. Treatment capacity, storage, conveyance infrastructure, drought conditions and competing demands can all become limiting factors.
Just as importantly, a water system that can satisfy average requirements may still struggle when several customers experience high demand simultaneously.
The third is the question of who should finance the infrastructure needed to accommodate that uncertainty.
Should existing customers pay for additional supply and treatment capacity intended primarily for new data centres? Should developers make binding long-term commitments before utilities invest? And what happens if the facilities use considerably less water than originally forecast?
These are not theoretical questions. They are already influencing infrastructure planning and development decisions.
A further theme concerned the distinction between water resilience and water robustness, introduced particularly effectively by Dr Robert E. Mace of Texas State University.
Resilience is the ability to recover after a failure. Robustness is the ability to avoid that failure in the first place.
The distinction is important for data center facilities that cannot afford a prolonged interruption to their water supply.
We also spent considerable time examining the difference between direct water consumption at a data centre and water consumption associated with the infrastructure supporting it.
Reducing onsite freshwater requirements is clearly valuable. But the wider water implications of a particular cooling system still need to be understood. The discussion becomes more complicated when different technologies alter resource requirements elsewhere.
Finally, there was the question of public confidence.
People understandably want to know where the water will come from, whether their own supplies could be affected, who will pay for additional infrastructure and whether the assurances being given today will remain valid in ten or twenty years.
None of these questions can be answered adequately by a generic statement about sustainability.
They require credible figures, realistic infrastructure plans and a willingness to explain both the benefits and the limitations of a proposed development.
With that broader context in mind, Andrew Parks provided an excellent starting point.
The full conference media and intelligence package includes over 14 hours of video content, covering the presentations and all the in-depth Q&A discussions.
One of the distinguishing features of the conference was its format — approximately 50% presentations and 50% extended Q&A. Some of the most revealing insights emerged during these discussions, as speakers were challenged on their findings, assumptions, practical experiences and the implications for the wider industry.
The package also includes the comprehensive conference report, which goes well beyond the headline takeaways to provide detailed analysis of individual presentations, technical findings, commercial implications, regulatory developments and the many important questions raised throughout the two days.
If you were unable to attend, this is an excellent opportunity to catch up on the discussions and gain a much deeper understanding of the commercial, regulatory and technical challenges — and emerging opportunities — surrounding AI data centres and water management.
Special Offer — Book for February 2027 and Receive the Full Conference Media Package Free
As an additional benefit for attendees, anyone booking a place at our February 2027 AI Data Center Water Infrastructure Congress in Dallas by 30 November 2026 will receive the complete October conference report and video package free of charge.
This provides an opportunity to catch up on the findings from October before joining us in February, when we'll examine how the industry's water challenges, technologies and regulatory landscape have evolved.
After 30 November, the media package will be available as a separate purchase.
To take advantage of this offer, simply book your place at the February congress before the deadline.
Texas Begins Asking Harder Questions About Data Centre Water Use
Mr Andrew Parks — Committee Director, Texas Senate Committee on Water, Agriculture, and Rural Affairs
Representing the office of Senator Charles Perry, Mr Andrew Parks delivered one of the conference's most consequential presentations, offering an unusually candid preview of how Texas legislators are examining the rapidly growing demand for water from AI data centres.
His central message was clear.
Texas remains enthusiastic about attracting investment, but legislators are becoming increasingly concerned about the industry's consumption of naturally occurring freshwater, the quality of the information available to policymakers and the potential consequences for existing water users.
Mr Parks began by explaining how the Texas legislature operates, including its biennial sessions and the interim committee process.
This was important context because the evidence gathered during 2026 could materially influence legislative proposals considered during the 2027 session.
Andrew Parks outlined four relevant interim charges on the Senate side and two on the House side, demonstrating how data centre development had moved into the legislative agenda.
What made Andrew's presentation particularly interesting, however, was the evidence emerging from the Texas Water Development Board and Public Utility Commission surveys.
The Data Gap Is Considerable
According to the figures presented, only 30% of operational data centres approached by the Texas Water Development Board responded to its water-use survey, while just 41% responded on time to the separate Public Utility Commission exercise.
Those numbers deserve attention.
Texas is being asked to plan for extraordinary industrial growth without a sufficiently complete picture of how much water existing facilities consume, which sources they use or what future demand might look like.
Andrew also pointed out that the state still lacked a comprehensive register of existing and proposed data centres.
That is a fairly remarkable starting point for an industry whose development pipeline could have major consequences for regional infrastructure planning.
The distinction here is important. A low response rate does not, by itself, establish that the non-responding facilities consume more water or are operating irresponsibly.
But it does mean that policymakers are being asked to draw conclusions from incomplete information.
That is hardly a satisfactory position when considering long-term resource commitments.
Where the Water Comes From
Andrew's survey figures were particularly revealing when the results were placed side by side.
Operational facilities
Potable water
83.5%
Non-potable water
5.2%
Planned facilities
Potable water
78.6%
Non-potable water
7.1%
Among responding operational facilities, 83.5% reported potable water as their source.
That figure does not tell us the proportion of total gallons supplied by potable systems, because facilities differ greatly in scale and consumption. Nevertheless, it illustrates just how dependent many existing developments remain on water supplied through conventional drinking-water infrastructure.
Cooling Technology Is Becoming a Legislative Issue
Cooling architecture was another major consideration.
Andrew reported survey evidence suggesting that evaporative cooling consumed approximately 3.8 times more water per unit of energy than non-evaporative alternatives.
The comparison is relevant, although it must be interpreted carefully because actual consumption depends on the technologies, climate and operating conditions being compared.
Senator Perry had articulated two broad principles that appeared likely to influence the committee's recommendations.
The first was opposition to evaporative cooling practices dependent on naturally occurring freshwater.
The second was greater transparency around industrial consumption of a public resource.
Andrew expected legislative proposals reflecting both concerns, but he was careful to explain that the committee's final recommendations had not yet been published.
That distinction should not be lost. These were emerging legislative priorities, not enacted requirements.
The Implication for Developers
What came through most strongly was that Texas's approach to data centre water use was becoming more demanding.
Developers could no longer assume that obtaining a water-service commitment would necessarily settle the wider questions.
Increasingly, they may need to explain the source of their water, why a particular cooling configuration has been selected, what alternatives have been examined and what the implications are for local resources.
Andrew's contribution was especially useful because it brought the regulatory debate back to actual numbers.
The difficulty for Texas is not simply the potential volume of future data centre water demand. It is that the state is trying to establish the scale of that demand while still lacking reliable information about a significant proportion of the facilities already operating.
Until that improves, both infrastructure planning and public confidence will remain more difficult than they need to be.
.JPG/:/cr=t:0%25,l:0%25,w:100%25,h:100%25/rs=w:400,cg:true/fx-gs)
Fort Worth Tries to Establish the Facts Before Making the Decisions
Dr Mike Slattery — Chair, Data Center and Infrastructure
Commission, City of Fort Worth
If Andrew Parks explained why the Texas legislature was beginning to take a closer interest in data centre water consumption, Dr Mike Slattery provided a useful account of how those same concerns were being examined at local government level.
Dr Slattery chaired Fort Worth's Data Center and Infrastructure Commission, an independent investigation commissioned by City Council on 11 August 2026.
The commission faced a remarkably compressed timetable. Its findings were due by 10 November, leaving approximately three months to examine a complicated and increasingly contentious development issue.
Mike approached the assignment from an academic and evidence-based perspective.
His argument was that it would be a mistake to assess a major data centre solely through one technical or environmental consideration.
A facility might appear acceptable when water, land use, infrastructure, economic contribution and community impacts are examined independently, yet still present a poorly understood overall development proposition.
Water Availability Is Only Part of the Assessment
For the water component, the commission was examining more than an annual consumption estimate.
It wanted to understand where the water would come from, the cooling technology proposed, whether existing infrastructure could provide sufficient capacity, what would happen during drought and how several developments might affect the same local resources.
These are important distinctions.
A proposal might appear manageable when considered on its own, while becoming considerably more difficult when combined with several other developments pursuing the same water infrastructure.
The commission also had to understand the quality of the available evidence.
This led to one of Mike's most important concerns: confidentiality.
The Problem With Withholding Information
Dr Slattery questioned whether confidentiality arrangements between developers and public authorities were preventing communities from obtaining information necessary to understand the implications of major developments.
Some commercial information plainly requires protection.
But a public authority also needs enough evidence to assess a proposal properly, and residents need to understand the reasons behind important decisions.
Noise modelling provided a useful example. Meaningful acoustic analysis requires dependable information about the equipment proposed and how it will operate.
If the relevant technical details are withheld, the ability to assess impacts independently becomes limited.
The same principle applies to water.
A statement that a facility will use efficient cooling tells a planning authority relatively little without a credible description of its expected demand, source and operating assumptions.
Listen, Act and Explain
Mike brought his presentation back to three particularly useful principles.
Listen. Understand the concerns before decisions become fixed.
Act. Change plans where the evidence and legitimate concerns justify doing so.
Explain. Make the decision, its reasoning and its limitations understandable.
I particularly liked this distinction because holding a consultation is not the same as allowing consultation to influence the result.
A developer may organise several public meetings and still leave residents feeling that the decision has already been made.
By contrast, changing a proposed setback, equipment arrangement, water strategy or construction plan in response to evidence demonstrates that the engagement has achieved something practical.
Mike's central point was that local authorities need an independent, publicly defensible basis for their decisions.
That will become increasingly important as more communities are asked to accommodate developments involving resource requirements and technologies with which they have comparatively little experience.
The wider lesson is that credible local decision-making depends not only on sound engineering, but on being able to show how the evidence was obtained, how concerns were considered and why the final decision was reached.
1. FREE CHAIR’S REPORT — INFOGRAPHIC HIGHLIGHTS
Available to download today, this condensed, visually presented report captures the salient points from every presentation at the October 2026 AI Data Center Water Infrastructure Congress.
It provides an accessible overview of the principal findings, important statistics, speaker insights and key takeaways from the two days.
2. FULL POST-CONFERENCE INTELLIGENCE PACKAGE
For those wanting to explore the findings in considerably greater depth, the comprehensive package includes the infographic report, all speaker presentation materials, over 14 hours of conference video, and the Chair’s independent post-conference analysis of the presentations.
Importantly, it also includes full Q&A transcripts, together with detailed analysis and evaluation of the questions, answers and wider discussions.
One of the distinguishing features of the congress was its format: approximately 50% presentations and 50% extended, in-depth Q&A.
Some of the most revealing insights emerged during these discussions, as speakers explored the practical realities, challenged assumptions and examined the commercial, regulatory and technical implications of different approaches to water management.
The result is a genuine treasure trove of industry intelligence, extending well beyond the formal presentations and offering valuable lessons for developers, operators, utilities, regulators and technology providers.
