Texas’ Data Center Freeze and AI Power Demand, Explained by Catalyst with Shayle Kann

Open Circuit from Latitude Media (on the Catalyst with Shayle Kann channel) focuses this episode on a fast-moving story: Texas and New York pausing big new data center projects amid a bipartisan backlash.

Host Steven Lacy, co-host Jigar Shah, and guest Jane Flegel frame Texas Governor Greg Abbott’s directive to regulators as a political response to public anger, not just a technical grid issue. According to Lacy, Abbott has ordered Texas regulators to freeze new data center grid connections until every project in the queue is audited for water use, power draw, and tax breaks.

They note that New York Governor Kathy Hochul has also signed an executive order pausing approvals for the largest data centers, anything above 50 megawatts, for a year, with carve-outs for sectors like hospitals and universities. Shah argues the federal “data center pledge” from the White House has failed to reassure voters, leaving governors scrambling to look tough on tech ahead of elections.

Flegel contends that opposition is now being driven less by questions about clean versus dirty power and more by distrust of whether communities will see concrete benefits. Throughout the discussion, the hosts emphasize that this is framed as a political story, with energy, water, and noise becoming proxies for deeper anxieties about AI, corporate power, and fairness.

Shah’s main argument is that the current wave of moratoriums and pauses is being shaped by political incentives and mistrust, rather than by a rational evaluation of grid or climate impacts. He claims that “the facts don’t matter here” because voters no longer believe the White House, governors, utilities, or data center developers have their best interests at heart.

According to Lacy, Abbott is in a statistical dead heat with challenger Gina Hinojosa and previously promoted Texas as an AI hub, illustrating how fast sentiment has flipped. Flegel agrees this is “almost entirely a political story,” adding that she sees two overlapping drivers:

• Broad skepticism that AI as a product will deliver public benefits
• Local fears about physical impacts and whether communities actually gain from hosting data centers

Flegel suggests that for most residents, benefits will not be judged by whether a facility is solar-powered, but by tangible gains like jobs, local investments, and noise mitigation. She criticizes the industry for doing “an absolutely terrible job” on community engagement and benefit-sharing.

The trio also discuss how environmental and climate groups are fragmented about what they actually want from data center policy. Flegel argues there is no shared, coherent package of “good grid citizenship” demands yet, which means pauses could simply kick the can down the road without resolving either the politics or the infrastructure challenges.

To ground the discussion, Lacy cites specific scale figures from the episode: he says there are 1,800 data center projects in the Texas grid queue, representing more than five times the state’s peak electricity demand. In New York, he notes that developers are pursuing enough power to cover roughly a third of statewide annual consumption. Both governors have responded with pauses framed around assessing water and grid impacts.

Flegel points to emerging policy experiments on how to treat large loads. She mentions Senator Martin Heinrich’s recently introduced “grid savings act” and says states including New Jersey, Pennsylvania, Virginia, and Oregon are exploring their own versions of big-load rules. According to her, these efforts wrestle with questions like:

• How to define the size and type of loads that trigger special treatment
• How to handle “cost causation” and whether to use take-or-pay obligations so ratepayers are not left on the hook
• What flexibility standards and expectations for clean, firm power should apply to large users

Shah highlights Virginia as an example of regulators pushing data centers to shoulder more costs. He recounts a recent State Corporation Commission decision that, in his telling, allocated most transmission upgrade costs to a data center tariff rather than to residential customers, reducing an expected residential increase from an extra $3 per month to about $0.94. He presents this as a “huge breakthrough” in aligning costs with large users.

Shah also references various tools and technologies—demand response management, virtual power plants, and grid-enhancing technologies—that he believes could help bend the cost curve, if incentives and regulatory frameworks are aligned.

Flegel and Lacy both warn that current policy responses carry significant risks. Flegel’s “big anxiety” is that elected officials are calling for pauses without a concrete sense of what “doing it right” means, leading to vague gestures around community benefits and “paying your fair share” but little practical change.

She also worries that extreme approaches to data centers could create problematic precedents for other energy-intensive sectors that many policymakers want to reindustrialize, such as aluminum smelting. In her view, taxing electricity use—even if aimed only at data centers—sends the wrong signal when the broader goal is to encourage electrification and economic growth.

On cost allocation, Flegel questions how precisely regulators can determine which network upgrades are “caused” by particular loads in a shared transmission system. She argues that calls to make data centers pay exactly what they cause on the grid are conceptually appealing but technically imprecise, especially for transmission.

At a deeper level, Lacy stresses that distrust of tech companies is about more than bills and wires. He cites patterns like non-disclosure agreements between developers and local officials, projects advancing before community input, and fears that the AI boom could collapse and leave stranded facilities. He also mentions worries that AI will create an economic underclass, making community benefit frameworks necessary but insufficient to resolve underlying anger.

Looking ahead, Flegel and Shah both argue that the data center pause period could be either wasted or transformative, depending on how advocates and companies respond. Flegel sees a “white space” where environmental groups, clean energy developers, and policymakers still lack consensus on the highest-leverage interventions to limit emissions from data center growth.

She describes ongoing efforts to define “good grid citizenship,” including her own paper on an American grid infrastructure fund and a menu of policy design variables: load thresholds, flexibility requirements, cost-causation rules, and obligations around clean firm power. Her concern is that, without a clear package ready, states will emerge from moratoriums in roughly the same position.

The conversation later turns to tech giants’ climate strategies. Lacy notes that Microsoft, Meta, Google, and Amazon have all reported emissions trending “up and to the right,” diverging from prior net-zero pledges. He references analyst Ketan Joshi’s work tracking these trends and asks whether net-zero targets remain useful metrics.

Flegel suggests evaluating large technology companies not just on whether their own emissions hit net zero by a given date, but on whether their capital and contracting power expand clean capacity faster than would otherwise occur. She highlights:

• Their use of advanced market commitments and long-term offtake agreements for new technologies (such as geothermal, nuclear, and clean materials)
• Their lobbying priorities and the policy issues they choose to advance or defend

Shah links this to industrial decarbonization, arguing that corporate offtake agreements are essential to moving loan-backed projects in areas like clean steel, cement, and sustainable aviation fuel from application to real deployment. Flegel frames the current surge in AI power demand as exposing the system-level barriers—interconnection delays, weak interregional transmission, permitting, and utility incentives—that will also matter for broader electrification.