162 forecasters
ClosesAug 9, 2026Aug 9, 2026
162 forecasters
3 more...
Will any U.S. state or EU member state restrict large-scale data center grid connections before September 2026?
PredictFollow
share
embed
35%chance
10% this week
TimelineHistogram
1d1w2mall
0%18.8%37.5%56.3%75%0%18.8%37.5%56.3%75%May 26May 27May 28May 29May 30May 31Jun 01Jun 02Jun 03Jun 04Jun 05Jun 06Jun 07Jun 08Jun 09Jun 10Jun 11Jun 12Jun 13Jun 14Jun 1535%
Top Key Factors
View all (2)
New York state legislature passes one-year data center moratorium techxplore.com • Jun 5, 2026techxplore.com•Jun 5, 2026
Increases Likelihood
10
Hochul's potential veto could delay moratorium
Decreases Likelihood
10
Comments (39)Key Factors (2)Question Info
Increases Likelihood
10
Hochul's potential veto could delay moratorium
Decreases Likelihood
10
Add key factor
39 comments
recent
Log in
WasteTimeContinuum· last weekJun 5, 2026·(edited last week(Jun 6, 2026))
Some interesting news broke yesterday regarding this question. As of Thursday night, June 4th, The New York State Legislature just passed the one-year data center moratorium. Introduced by Sen. Kristen Gonzalez as S.10642, which folded S9144A into it along with several other bills and changed the moratorium from three years down to one year. It's called the " Responsible Data Center Development Act," it pauses new permits for data centers above 20 megawatts for a year, and it's now sitting on Hochul's desk waiting on her signature or veto. If she signs it, New York becomes the first state in the country to do this. So this isn't a "still in committee" situation anymore. The legislature's part is done and the whole thing now hinges on Governor Hochul.
"This is a historic moment for New York. For too long Big Tech has benefited from under-regulation, writing their own rules on large scale data center development with little accountability or transparency on local impacts. - Senator Kristin Gonzales
I think an easy mistake here is to treat the legislature passing the bill as most of the work being done. After digging into it, I came away thinking it’s much less certain. It comes down to one person and what she thinks her supporters want.
Hochul hasn't actually backed a statewide moratorium. Her line has been that this kind of decision should be left to municipalities, and as recently as last week she was refusing to comment on the bill while saying she's "l ooking closely at the issue." So this isn't a governor who's eager to sign. She also doesn't have to decide quickly. From what I read she has until December to act on bills sent to her, which matters a lot for this question. There's a very real version of this where she's broadly sympathetic, signs eventually, but doesn't do it in time. "She signs in October" still resolves this No.
What I’m wondering about now is who she's trying not to upset. There are basically three groups pulling on her. Regular voters who are angry about their electricity bills, and the timing here is interesting to me. The lawmakers leaned into affordability right before the November election. Then the building trades unions, who want the construction jobs and are actively lobbying her against it. And then the tech and AI money, which has become a real donor base for her this cycle. Even if a lot of those donors don't personally care about one data center in upstate New York, what matters is whether Hochul and her fundraising people perceive them as a base worth protecting. Politicians worry about the next check, and I don't think it's crazy for her to be cautious about kicking a sector that's been good to her. I’d worry about that.
The catch, and the reason I don't think the donor angle is decisive, is that appeasing the voters can be done without giving them the actual hard moratorium. She can sign a watered-down version, demand a carve-out for a specific project the way Maine's governor did, or just let it sit and study it to death, and still tell voters she "acted on data centers." That's the outcome I think could potentially be underweighted by this news. Not a dramatic veto, but a soft dodge that doesn't meet the bar this question needs.
Something that swings my reasoning the other way and made me less confident in my own skepticism is that Hochul already signed a moratorium on crypto mining back in 2022, over basically the same union objections we're seeing now. So she's done this exact kind of thing before when the politics lined up, and that’s a solid precedent to factor into this.
Key Factors
Hochul's potential veto could delay moratorium
Decreases Likelihood
10
New York state legislature passes one-year data center moratorium techxplore.com • Jun 5, 2026techxplore.com•Jun 5, 2026
Increases Likelihood
10
4
Reply(0)Changed my mind (0)Mind (0)(0)
hide 8 replies
WasteTimeContinuum· last weekJun 5, 2026
@admins Afrer rereading the resolution criteria I have a question about how the New York bill would resolve this, if Hochul signs it. The bill that passed Thursday night is a moratorium on permits for the siting and construction of new data centers, run through the Department of Environmental Conservation. It's not written as a grid connection or interconnection rule in the technical sense. But the resolution criteria specifically says "applying to new grid connection requests," so I wanted to ask how literally that phrase is meant to be read.
Tanken strictly, a construction moratorium should count, and here's my reasoning. A data center can't connect to the grid if it isn't allowed to be built in the first place. The grid connection is one step in a chain, you site it, you build it, you connect it. If the state freezes the very first step, every step after it is frozen too, including the connection. So in practice a construction moratorium is a stricter version of a connection restriction, not necessarily a weaker one. A connection rule stops you at the last gate. A construction rule stops you at the first gate, which means you never even reach the connection stage. The end result for a new large-scale data center trying to get onto the grid is the same either way, it can't. The spirit of this question is in the same spirit of the NY memorandum ultimately. To prevent/curtail data center power consumption.
I'd also point out that the criteria uses three different terms, "moratorium, suspension, or binding restrictive rule," which to me reads like it's meant to capture the category of state actions that block new large-scale data centers from getting connected, rather than only rules that use the specific word "interconnection." If the intent were strictly limited to technical interconnection rules, a lot of what's been discussed in this thread, including the bills in the base-rate table, probably wouldn't qualify either.
So my question is really, does a binding construction or siting permit moratorium that prevents new large-scale data centers from being built (and therefore from connecting) satisfy the "applying to new grid connection requests" language? Or is the question looking for something that targets the interconnection process specifically? I think this matters a lot for how the NY bill resolves, and I'd guess for some of the other state bills too, so I think it's worth clarifying.
3
Reply(0)Changed my mind (0)Mind (0)(0)
hide 7 replies
WasteTimeContinuum· last weekJun 8, 2026
@DaiLothbrook What are your thoughts regarding my question to the @admins above? Your question is a good one, but it’s resolution is highly complex. I’m curious if you consider S.10642 within the scope of resolving this question if Hochul passes it. Thanks!
1
Reply(0)Changed my mind (0)Mind (0)(0)
hide 2 replies
DaiLothbrook· 5 days agoJun 9, 2026
@WasteTimeContinuum Dear collegue, thank you for the comments and investigation. First, yes, I agree this is a more complex resolution than many forecasters may be used to. But I think that is partly because the underlying phenomenon is complex itself. This is not like forecasting CPI, VIX, temperature, or another single numerical metric where we can simply go to one source and read off one number. This question is trying to forecast an institutional response to a real-world infrastructure problem, high energy demand risks, key variable to measure IA scaling capabilities . which might push us out the comfort zone than a simpler resolutions and questios types
That makes the resolution harder, of course, but also more interesting. In questions like this, there will inevitably be edge cases, because laws, moratoria, permits, siting rules, interconnection procedures, environmental reviews, utility regulation, and cost-allocation mechanisms do not always use the same legal language. That means the fine print has to do more work than usual. I wish I could have put my last added resolutions and fine print clauses comments in the formal field to make the clarifications explicit .
Now, concerning if NY case should or not be a possible object of Yes reolutions.
In terms, The object of the question is not simply “are data centers stressing the grid?” but whether that stress becomes institutionalized into formal restrictions. In shorthand, I see the question as: AI/data-center demand + grid stress + local/state regulation + institutional restriction. So, on your specific point, I think the key distinction is between the phenomenon the question is trying to map and the exact wording of the current resolution criteria. At the phenomenon level, I agree with your reasoning: a binding construction, siting, or permitting moratorium is clearly part of the same family of institutional pushback against large-scale data centers. In fact, it is arguably a stronger restriction than a grid-connection rule, because it blocks the project earlier in the chain. If a data center cannot be permitted, sited, or built, it also cannot proceed to grid connection. So, it is a case that does capture the essence
But , and it is a big BUT, under the current wording, I would be more cautious. The current question asks about
“a new moratorium, suspension, or binding restrictive rule will impede a grid connection requests from large-scale data centers.”
Read strictly, that seems to target grid-connection or interconnection restrictions, not construction or siting moratoria. So my current view is: the New York bill maps the same underlying phenomenon the question was made to capture : institutional response and resistance to AI Infrastructure burdens and risks. but it should not resolve this exact question as Yes under the current wording unless the resolution criteria are explicitly broadened. For the New York-style case to clearly count, the question would need to be written more broadly, for example:
“Will any U.S. state or EU member state announce rules restricting or preventing new grid connection requests from large-scale data centers, or preventing new large-scale data centers from being permitted, sited, or constructed, before September 1, 2026?”
The resolution criteria would also need to say that qualifying restrictions include binding rules, moratoria, or suspensions that prevent new qualifying data centers from being permitted, sited, constructed, operated. So both questions and resolution would have to change to accept this case. But adding this would also make the question broader than it already is. It would no longer only monitor grid-connection restrictions; but also monitor both grid-connection impediments and construction/permitting impediments at the same time. Given that the question is already geographically broad, covering U.S. states and EU member states, adding construction and siting restrictions may be too much for one question. It might be better to keep this question narrower and create a separate question about permitting/construction moratoria for large-scale data centers.
Addtionally, on the New York bill specifically, my understanding is that it is not law yet. The official New York Senate page describes S.10642 as placing a one-year moratorium on the issuance of data center permits and requiring utilities to establish an independent classification of service for large data centers: https://www.nysenate.gov/legislation/bills/2025/S10642. The Verge reported that the legislature passed the one-year moratorium, but that it still depends on Governor Hochul’s signature or veto: https://www.theverge.com/policy/944041/new-york-data-center-moratorium. So legislative passage alone should not be enough for any Yes resolution.
I also think your political read is useful. Hochul does not look like an automatic signer. E&E/Politico reported that she has been skeptical of a statewide moratorium and has framed data-center siting as a local land-use issue: https://www.eenews.net/articles/new-yorks-governor-doesnt-support-statewide-data-center-moratorium/. Supporters seems to be pushing her to sign because of electricity bills, environmental impacts, and community concerns; Earthjustice, for example, explicitly urged her to sign the bill: https://earthjustice.org/press/2026/nys-legislature-passes-data-center-moratorium. Industry groups,have demonstrated they are against the restrictions and limitatios: https://www.theguardian.com/us-news/2026/jun/06/new-york-datacenter-temporary-ban.So, in terms of political economy, it is a very complicated issue and I think it will hard to be decided this year.
So my bottom line is: S.10642 is highly relevant evidence for the broader phenomenon this question is trying to monitor. But under the current grid-connection-focused wording, I would not treat it as clearly resolving this question Yes. It would clearly count only if the question and resolution criteria were broadened to include permitting and construction moratoria. which might be a future question but probably with a longer time horizon, feel free to create it. I dont knwo if we can co-create questions here anyway,
2
Reply(0)Changed my mind (0)Mind (0)(0)
hide 1 replies
WasteTimeContinuum· 5 days agoJun 10, 2026
@DaiLothbrook Thank you for this, I appreciate you taking the time for this thoughtful response. You clearly thought hard about it.
I think we agree on more than we disagree. We both see the New York bill as covering the same underlying phenomenon this question is about, institutional pushback against large-scale data centers, and you even put it better than I did when you described it as blocking the project earlier in the chain. So the disagreement isn't really about what the bill does or what the question is trying to capture. It's narrower than that. It's just about whether the current written wording, "applying to new grid connection requests," is broad enough to count a permitting and siting moratorium, or whether it has to be read as interconnection-specific.
1
Reply(0)Changed my mind (0)Mind (0)(0)
WasteTimeContinuum· 5 days agoJun 10, 2026
@admins @skmmcj Could we get an official ruling on my interpretation question above, because I think it's outcome-determinative and @DaiLothbrook and I have different reads on the wording. This determines whether a signed New York bill resolves the question, and I think it affects how several other state bills should be read too. It would be very helpful if this were settled while we're predicting this question, rather than leaving it to interpretation at resolution time. Thanks!
0
Reply(0)Changed my mind (0)Mind (0)(0)
hide 3 replies
skmmcj· 4 days agoJun 10, 2026·(edited 4 days ago(Jun 10, 2026))
@WasteTimeContinuum I agree with DaiLothbrook that a moratorium on building datacenters shouldn't count. Under your interpretation we'd have to count anything that blocks datacenters being built, which is a bit too broad imo. And the criteria do not really point to anything like this.
2
Reply(0)Changed my mind (0)Mind (0)(0)
hide 2 replies
WasteTimeContinuum· 4 days agoJun 10, 2026
@skmmcj Fair enough. Thanks!
0
Reply(0)Changed my mind (0)Mind (0)(0)
skmmcj· 4 days agoJun 10, 2026
@predictors please take note
3
Reply(0)Changed my mind (0)Mind (0)(0)
DaiLothbrook· 2 weeks agoMay 28, 2026·(edited 2 weeks ago(May 28, 2026))
Here I will share my first base rate search for this questions. First I recognise I should have made differen questions without mixing geographical location. One only for US and Another only for EU.
Below, I amde a base rate exercise cutting from January 2025 to May 25, 2026. So t his table covers only bills that explicitly name data centers as the direct target of the restriction, not general energy or large-load legislation. ad inly US states.
this Table was made by my agent researcher following my request:
| State | Bill | Status | Type of action | | Georgia | HB 1012 | Introduced in legislature, in committee as of April 2026, not enacted | Bars counties and cities from issuing permits for construction or development of new data centers until March 1, 2027. Exception for approvals already issued before July 1, 2026. | | Maryland | HB 120 | Introduced in legislature, in committee as of April 2026, not enacted | Prohibits construction of new data centers unless the General Assembly enacts co-location legislation (natural gas, nuclear, or SMRs). | | Michigan | House Resolution 240 | Introduced in legislature, not enacted | Temporary statewide pause on state incentives for data centers. Not a construction ban. | | New Hampshire | HB 1265 | Introduced in legislature, not enacted | Prohibits construction of any new data center statewide for one year from effective date. Forms a committee to study environmental impacts. | | Minnesota | SF 4298 | Introduced in legislature, not enacted | Bars state and local governments from issuing data center permits until one year after the PUC submits an impact report. | | New York | S9144 | Introduced in legislature, not enacted | Statewide moratorium on permits for new data centers. Directs PSC to develop a plan to minimize impact on electricity and gas rates. | | Oklahoma | SB 1488 | Introduced in legislature, not enacted | Moratorium on building new data centers until November 1, 2029. Tasks Oklahoma Corporation Commission with studying water, rates, property values, and site impacts. | | South Carolina | H 5286 | Introduced in legislature, not enacted | Bars state and local governments from granting permits, approvals, or incentives for new data centers until January 1, 2028. | | South Dakota | SB 232 / HB 1301 | Introduced in legislature, not enacted | One-year moratorium on construction or expansion of hyperscale data centers. HB 1301 also addresses cost and risk allocation. | | Vermont | S 205 | Introduced in Senate, not enacted | Creates a temporary moratorium through July 2030 while regulators study impacts and recommend a regulatory framework. Different bill from H.727, which passed (see Part 3). | | Virginia | HB 1515 | Introduced in legislature, not enacted | Bars localities from granting final approvals for rezonings or special permits for new data centers until specific grid-interconnection conditions are met. | | Wisconsin | LRB-6377/1 | Introduced in legislature, not enacted | Prohibits operation of a data center unless the legislature first enacts statewide rules, including planning authority and explicit bans on cost-shifting to residential customers. | | Maine | no bill number in sources | Passed House and Senate. Awaiting signature from Democratic Governor Janet Mills as of late April 2026. Not yet enacted. | Freezes approvals for data centers requiring more than 20 MW of power until October 2027. Requires a state-appointed council to analyze impacts on grid, electricity bills, air, and water. |
All thirteen bills were introduced in 2025 or 2026. As of May 28, 2026, none of them have become enacted law. The most advanced case is Maine, which passed both legislative chambers and is awaiting the governor's signature. The remaining twelve are at various stages of committee review, floor debate, or dormancy, and none have cleared their full legislative process. All thirteen bills explicitly target data centers by name.
They were written to address data center impacts specifically, not as general energy regulation. This is the critical distinction from Texas SB-6 and California Title 24, which regulate large loads broadly and affect data centers only as a secondary consequence. THIS, I believe is crucial to understand. Moreover, the resolution criteria specifies 50 MW as the threshold for "large-scale." Most bills in the table use a 20 MW threshold (Vermont H.727 and Maine) or target hyperscale broadly without a specific MW floor (South Dakota, Georgia, Oklahoma).
None of the bills reviewed in this cycle were written specifically as grid connection restrictions in the technical utility sense. No earlier precedent exists in U.S. state law for this exact class of restriction (explicit data center grid connection moratorium with force of law targeting 50 MW+ facilities). So, the US base rate from this historical window: 0 out of 13 introduced bills became enacted within the observation period. Maine is the one case that could change this, but it has not been signed. I would say the base rate is between 1-5%.So it seems that is a phenomena starting to happening now, very important to monitor. Concerning Vermont, I will make a research about it by saturday. @symbiosis But for What I already read, it will not count
. Some sources: More Than a Dozen U.S. States Consider Temporary Data Center Bans
a very good one: Data Center Regulation by State 2026: Cost, Water, Disclosure & Noise Laws
I even found a Moratorium tracker, which can be used for future resolutiions: Data Center Moratorium by State 2026: Every Bill, Status & Threshold
4
Reply(0)Changed my mind (0)Mind (0)(0)
hide 3 replies
symbiosis· 2 weeks agoMay 28, 2026
@DaiLothbrook Thanks for the helpful update. FWIW Vermont H.727 was vetoed today, which simplifies that issue!
2
Reply(0)Changed my mind (0)Mind (0)(0)
hide 2 replies
symbiosis· 2 weeks agoMay 29, 2026
@symbiosis Although it passed with a large enough margin that it could still be overriden
1
Reply(0)Changed my mind (0)Mind (0)(0)
hide 1 replies
DaiLothbrook· last weekJun 5, 2026
@symbiosis Yep maybe . But not necessarily it will be a case to reolve as Yes. I have found an outlet saying it is a huge law for Data center but it seems and exaggeration I will accompany it though. Next queation about it mught be if one of these laws will pass, or how many others laws like that ( in the table ) will pass in each state.
0
Reply(0)Changed my mind (0)Mind (0)(0)
baybar· 3 weeks agoMay 26, 2026
I am pretty confused about what qualifies as a "binding restrictive rule". Would for example, something like Texas S.B. 6 count under this definition, if it happened after May 20th? This legislation required new connections of 75 MW or more to pay an interconnection fee of $50,000 per MW, among other things.
1
Reply(1)Changed my mind (1)Mind (1)(1)
symbiosis· 3 weeks agoMay 25, 2026
@admins Would rules requiring large data centers to get public utility commission approval, obtain a finding that they won't harm grid reliability, reach special interconnection agreements, etc. count as "restrictive rules" for the purposes of this question?
0
Reply(0)Changed my mind (0)Mind (0)(0)
hide 7 replies
skmmcj· 3 weeks agoMay 26, 2026
@symbiosis @darkives @DaiLothbrook Do you have any thoughts regarding the above and this comment by @baybar?
0
Reply(0)Changed my mind (0)Mind (0)(0)
hide 3 replies
darkives· 3 weeks agoMay 26, 2026
@skmmcj I understand where @DaiLothbrook is coming from here; analytically it makes sense to me, though it may be difficult to operationalize. As to SB 6 in particular, I think we can avoid the issue because SB 6 has a 75 MW threshold, whereas this question has a 50 MW threshold. I think it necessarily cannot cause a Yes resolution for that reason alone. @baybar
The question from @symbiosis is more abstract and thus more difficult. I would think that if the rules were truly new and really had teeth—i.e., they were designed such that some data centers would genuinely fail to receive approval, rather than as merely a box-checking or hoop-jumping exercise—that would count. I recognize that this sort of "you know it when you see it" type of answer is not super helpful, though.
2
Reply(0)Changed my mind (0)Mind (0)(0)
hide 2 replies
symbiosis· 3 weeks agoMay 26, 2026
@darkives Thanks. My concern in particular is Vermont H.727 which seems plausibly like regulatory cost-allocation management but may also in practice be a fairly stringent requirement. Whether or not that would count could potentially have a big impact on the probability. @DaiLothbrook @skmmcj @baybar
1
Reply(0)Changed my mind (0)Mind (0)(0)
hide 1 replies
darkives· 3 weeks agoMay 27, 2026
@symbiosis I think this is a very fair question. Having looked at the bill and some commentary around it (for example, from Vermont PIRG), my inclination is that this probably would count as a restrictive rule, but I agree with you that the answer is not obvious.
2
Reply(0)Changed my mind (0)Mind (0)(0)
DaiLothbrook· 3 weeks agoMay 26, 2026·(edited 3 weeks ago(May 26, 2026))
@symbiosis Thanks everyone for the thoughtful questions and engagement here. I think perhaps one important distinction that may help clarify some of the confusion is the difference between forecasting the raw phenomenon itself versus forecasting institutional responses to that phenomenon. Which is something I am actively working on now.
In other words, this question is not directly asking whether data centers will create stress on energy grids.
I created it to ask: Whether that stress will become institutionalized into formal restrictions, moratoria, suspensions, or legally binding limitations. That distinction matters a lot because many underlying physical or economic problems never actually translate into formal policy action. Perhaps one of the main points of confusion here I see is exactly where ordinary governance ends and where a genuine restrictive policy begins. Very important consideration. So, It is important to consider that every energy-grid connection process already involves some level of regulation, review, queue management, interconnection agreements, reliability studies, and administrative approval, Ok? If any procedural requirement automatically counted as a “restriction,” then almost any regulatory update could potentially resolve this question as Yes, which would make the question far too broad and noisy to be forecasted.
So regarding the questions raised by @symbiosis and @baybar about the meaning of “binding restrictive rule,” my current interpretation would be that ordinary procedural governance alone should not count. The Texas S.B. 6 example raised by @baybar is actually a very useful edge case. The key question there is: is this simply a normal regulatory cost-allocation mechanism, or is it effectively functioning as a restrictive barrier intentionally designed to slow or discourage large-scale data center expansion? Which is the real point of the question
Economically, it may indeed behave as restrictive, ok. But juridically and administratively, it may still be interpreted primarily as cost allocation or grid-management governance rather than an actual restrictive policy( what we are looking for). That is exactly why these edge cases matter indeed so much for clarity.
I also think @WasteTimeContinuum raises an important point regarding the word “new.” Renewals, extensions, or reaffirmations of existing restrictions should not count as new restrictions. We must consider that governements frequently extend pre-existing moratoria, reapply older limits, or maintain already existing constraints. But that is not necessarily a regime shift., right? my intent here is specifically to capture the emergence of genuinel y new restrictions that did not previously exist in substantially similar form.Pehaps, given the nature of the problem i could have made the fine print, background more complete and explicit about it.
So I have worked on these new defintiton fine print all afternoon on the hope it clarifies enough
Ordinary interconnection procedures, reliability reviews, queue management, or standard cost-allocation mechanisms shall not count unless they explicitly create NEW Material barriers, delays, caps, suspensions, or moratoria specifically targeting large-scale data center connections.
Rules imposing unusually burdensome financial or procedural requirements specifically targeting large-scale data centers may count if t heir practical effect is to materially restrict or delay new grid connections.
Restrictions motivated by grid reliability, transmission congestion, water stress, energy security, or infrastructure-capacity concerns all qualify
=So the motivations matter if the consequence is restrict or delay new grid connections. So a question that can help you can make: Is this an ordinary procedure, rule or this is Unusual?\\ might help to distinguis h Another one might help: T he pratical effect of this varialbe X is to restrict such connections? I hope these clarifications preserve the spirit of the question while reducing ambiguity around edge cases like Texas S.B. 6.
2
Reply(0)Changed my mind (0)Mind (0)(0)
hide 2 replies
symbiosis· 3 weeks agoMay 26, 2026
@DaiLothbrook Thanks for the response. I'm still not sure I understand how this question will be resolved. It seems to me that Texas SB6 would impose a binding, restrictive rule that isn't ordinary in the sense that it doesn't apply to smaller data centers. So I would have thought on the basis of the resolution criteria that it would count as a restriction for the purposes of this question. But in your comment above you say "it may still be interpreted primarily as cost allocation or grid-management governance rather than an actual restrictive policy." Am I right in understanding that you think in its current form Texas SB6 would be a normal regulatory cost mechanism rather than materially restrictive in practice?
0
Reply(0)Changed my mind (0)Mind (0)(0)
hide 1 replies
DaiLothbrook· 2 weeks agoMay 28, 2026
@symbiosis and collegues, After doing research to understand Texas SB 6, I think there are two separate reasons why it definetely should not resolve this question as Yes.
First, timing alone seems sufficient to exclude it. Texas SB 6 was signed by Governor Greg Abbott in June 2025 and took effect immediately. This question is trying to forecast what happens after the question opens, from May 2026 to September 2026. So it is a pre-existing policy, not a new restriction announced during this question’s forecasting window. We cannot forecast the past.Second, substantively, my reading is that SB 6 is not best understood as a moratorium, suspension, or explicit cap on data center grid connections. The bill itself is described as relating to
“the planning for, interconnection and operation of, and costs related to providing service for certain electrical loads.”
In other words, it creates rules for large-load customers in ERCOT, with an initial 75 MW threshold.So SB 6 it is a new large-load interconnection governance, grid-reliability, cost-allocation, and risk-management regime for the Texas energy system. It is meant to protect the grid from the risks and costs created by very large electricity loads, which happens to including data centers, but also other large-load customers rather than to prohibit or suspend data center development.Of course, such a regime may create administrative costs, additional procedural requirements, higher financial obligations, or even possible delays for some large-load projects. But those effects appear to be consequences of grid reliability planning and risk management, not the direct object of a moratorium or restrictive ban. THIS IS CRUCIAL TO DISTINGUISH.
For clarity, I would define the key terms this way:
A moratorium would mean a formal temporary pause or ban on accepting, processing, or approving new large-scale data center grid connection requests.
A suspension would mean a formal temporary halt in the operation of the normal connection process, approvals, or authorizations for such requests.
A binding restrictive rule would mean a formal written rule with legal force that directly restricts, caps, delays, suspends, denies new grid connection requests from large-scale data centers.
Under that reading, ordinary interconnection governance, reliability review, queue management, study fees, cost allocation, and project-readiness requirements , specially those with the objective to protect the energy secutirty of the state and its risk management and costs should not count unless they explicitly create a new , delay, cap, suspension, moratorium, or denial mechanism specifically targeting large-scale data center grid connections.
So 1- It must target specifically large-scale data center grid connections.
2- The Restriction is an objective, clearly stated. Not a consequence of energy security acts.
So my current interpretation is: Texas SB 6 is highly relevant background evidence for why this question matters, but it must clearly not itself resolve the question as Yes.
For who wnat to read the law: 89(R) SB 6 - Enrolled version
1
Reply(0)Changed my mind (0)Mind (0)(0)
Russell-viper· 3 weeks agoMay 25, 2026·(edited 3 weeks ago(May 25, 2026))
Such a great question, though it's very complex as there are lot of moving parts to it.
First of all i would say that there is an iron law in economics: " extremely good and extremely bad circumstances rarely stay that way for long because supply and demand adapt in hard-to-predict ways" and that's something we have to keep in mind.
The largest Al data centres cover more than 17 football fields of land, use as much electricity as 100,000 households and up to 5 million gallons of water every day to cool the servers. That's the daily water use of a city of 50,000 people.
Issues caused by Data centers 1 - They create no jobs other than the only labours required to build it and few number of engineers. They generate relatively few permanent jobs compared to the land and incentives they receive.
2 - Water prices are going up as they consume a huge amount. For a reference a typical nuclear power plant consumes roughly 2 to 3.5 times more water annually than one the world's largest and resource intensive AI data centers. A nuclear power plant relies on "once-through" cooling system which pulls billions of gallons from the a nearby source usually a sea or a river and uses it pass through condensers to absorb heat and dump it right back into source slightly warmer. While the AI data centers relies on "evaporative cooling", which uses phase change of water (transitions between solid (ice), liquid, and gas (vapor/steam) states) to absorb heat and evaporates it entirely into atmosphere. They are often built near municipal hubs or in tech corridors (Arizona or parts of American West) and frequently rely on local potable tap water. though they are efficient but in comparison to nuclear power plant which means they still consume a lot.
3 - The zoning of AI data centers is horrible, they are growing like topsy with huge amount of investments. They are occupying enormous tracts of zoned land and consuming vast amount of resources of the community (which causes competition among local residents for limited resources and eventually a price increase) and that too without contributing nothing to the local job growth, and this causes civil unrest, we already have witnessed the incident of Molotov cocktail at OpenAI CEO Sam Altman’s San Francisco home in April 2026.
The rise of populism in the world If we look at the major waves of populism they were strongly associated with men without college degrees, workers in manufacturing, mining and industrial sectors or communities who got negatively affected by globalization and trade shocks. They experienced wage stagnation or declining real incomes but looking at the current situation we can see the up coming wave of populism is hugely or primarily going to be driven by the AI or data centers with its impacts on rising unemployment (in the short term) and it seems likely that the upcoming wave is going to be led by women with advanced degrees urban and suburban, who are worried about their jobs and worried about their kids.
we already are witnessing the rise of populism around, even in the US, Trump got elected on the back of men who had good jobs and were making good money but they lacked advanced degrees, and they felt that the world is moving away from them, they saw robotics and automation on the assembly line, they saw free trades and jobs going to much poor and less expensive labour around the world e.g China, India and Vietnam, they saw immigrants coming in and taking their place that's why they voted for Trump not once but twice despite of everything he is, this is akin to a Nigel Farage moment in the UK, which gave rise to the Reform UK (right-wing populist party) led by the Nigel Farage since June 2024, formerly the brexit party.
I think it is fair to label the AI moment as a fuel for populism and hence this will likely to put pressure on top AI moguls or hinder the progress of "Data centers", I mean it's not even surprising that the we are already hearing the idea of Universal Basic Income "UBI" that is thrown around to calm the angry people. It indicates the existence of an upcoming chaos that is waiting to explode. This will likely to play out in long term and less likely to in next 4 months.
2
Reply(0)Changed my mind (0)Mind (0)(0)
hide 7 replies
exmateriae· 3 weeks agoMay 25, 2026
@Russell-viper upvoted but some caveats, not specifically against you but to point some stuff:
The largest Al data centres cover more than 17 football fields of land, use as much electricity as 100,000 households and up to 5 million gallons of water every day to cool the servers. That's the daily water use of a city of 50,000 people.
These numbers seem huge and that's why people use them, but they're actually not that much? The land does not matter at all, especially in the US which is 90% desert. 17 000 football fields is 0.0008% of the US. (with 3 additional OOMs!!)
Households' electricity usage is only an issue because it is focused at specific times of the day and rarely because of the size of demand. Enterprises are the true users of energy. I don't think we care it uses the equivalent of 100k households. We might if we build no additional energy in the next 5 years though but we shouldn't worry: coal, oil and gas will be there if no one wants to help make energy for the data centers.
Water is the same. Enterprises use insane amounts of water, the only reason this is an argument is because people "dislike" the end product. If they want to solve this problem, they should stop using chatbots and social media, that will help much more. (I am serious)
Issues caused by Data centers
1 - They create no jobs other than the only labours required to build it and few number of engineers. They generate relatively few permanent jobs compared to the land and incentives they receive.
Direct jobs
By the same reasoning, "normal" data centers don't create jobs either? I doubt we think that's the case considering the part of GDPs which are purely online nowadays.
A nuclear power plant relies on
"once-through"
cooling system which pulls billions of gallons from the a nearby source usually a sea or a river and uses it pass through condensers to absorb heat and dump it right back into source slightly warmer. While the AI data centers relies on
"evaporative cooling",
which uses phase change of water (transitions between solid (ice), liquid, and gas (vapor/steam) states) to absorb heat and evaporates it entirely into atmosphere. They are often built near municipal hubs or in tech corridors (Arizona or parts of American West) and frequently rely on local potable tap water. though they are efficient but in comparison to nuclear power plant which means they still consume a lot.
I'm not convinced this is true, many nuclear reactors in France have towers, if not the majority. Beyond that, sure it consumes water, electricity etc, but that's just the way industry works. Like above, I see no one complain about the massive energy costs of making aluminum.
3 - The zoning of AI data centers is horrible, they are growing like topsy with huge amount of investments. They are occupying enormous tracts of zoned land and consuming vast amount of resources of the community (which causes competition among local residents for limited resources and eventually a price increase) and that too without contributing nothing to the local job growth, and this causes civil unrest, we already have witnessed the incident of Molotov cocktail at OpenAI CEO Sam Altman’s San Francisco home in April 2026.
If the argument is that you can't put a data center in the middle of the nevada desert because no one will want to live there, sure. But the towns in question are pretty small ones, I really think this is an overblown issue.
The rise of populism in the world
If we look at the major waves of populism they were strongly associated with men without college degrees, workers in manufacturing, mining and industrial sectors or communities who got negatively affected by globalization and trade shocks. They experienced wage stagnation or declining real incomes but looking at the current situation we can see the up coming wave of populism is hugely or primarily going to be driven by the AI or data centers with its impacts on rising unemployment (in the short term) and it seems likely that the upcoming wave is going to be led by women with advanced degrees urban and suburban, who are worried about their jobs and worried about their kids.
I believe it already exists. The two reasons why we don't talk about it: it is socially fine, even if you were to argue for discriminatory measures against men, and their parties mostly lose so that has less political impact. Women have radicalized more than men but when media talk about radicalization, they do not care about that part, on the contrary they encourage it. (South Korea really looks like a completely insane country the more you learn about it)
I think it is fair to label the AI moment as a fuel for populism and hence this will likely to put pressure on top AI moguls or hinder the progress of "Data centers", I mean it's not even surprising that the we are already hearing the idea of Universal Basic Income "UBI" that is thrown around to calm the angry people. It indicates the existence of an upcoming chaos that is waiting to explode.
This will likely to play out in long term and less likely to in next 4 months.
Yes. Although, there might not be enough time for anyone to revolt, if things were to go incredibly well or incredibly wrong.
3
Reply(0)Changed my mind (0)Mind (0)(0)
hide 6 replies
Russell-viper· 3 weeks agoMay 26, 2026·(edited 3 weeks ago(May 28, 2026))
Households' electricity usage is only an issue because it is focused at specific times of the day and rarely because of the size of demand. Enterprises are the true users of energy. I don't think we care it uses the equivalent of 100k households. We might if we build no additional energy in the next 5 years though but we shouldn't worry: coal, oil and gas will be there if no one wants to help make energy for the data centers. Water is the same. Enterprises use insane amounts of water, the only reason this is an argument is because people "dislike" the end product. If they want to solve this problem, they should stop using chatbots and social media, that will help much more. (I am serious)
Yes, again the same point, I talked about this in detail in a comment in the Labour automation tournament. People don't like new technology just because it introduces uncertainty into the future. Take Industrial Revolution, the revolt against the printing presses and later on Luddite movement and so on and so forth, all were aimed at contemporary innovations but things always go to the path of least resistance.
Secondly, the US has one big advantage over the China, it's energy independent but the US has shortage of electricity, the demand is much greater than the supply because of the obvious reasons, the growth in Crypto MIning and Ai, both consume massive amount of electricity. Henry Paulson (Former US Secretary of the Treasury) admitted himself on the bloomberg "We have a shortage of electricity to power data centers".
You said
We might if we build no additional energy in the next 5 years though but we shouldn't worry: coal, oil and gas will be there if no one wants to help make energy for the data centers
you are right, but the supply of electricity is inelastic in the short run, meaning that sudden price spikes or surges in demand cannot be quickly met with increased production. As i said, the US is energy independent but they are behind in comparison to China, just to share some stats; Increase in TW/h of new generation.
China since 2021, in just 5 years, it has added more power capacity across all energy technologies than the US has built in its entire history and that gap is only going to grow, look the graph below;
Source:Bloomberg
Over the next 5 years China plans to add more than 3.4 TW of energy generation capacity which is nearly 6 times as much as the US.
According to Bloomberg, In the US power demand from the data centers alone is expected to triple by 2035. In that sense the US could possibly face a backlash from public, you can't jack the prices and expect consumers to pay with a straight face. This is the biggest limiting factor going forward for here, off-course supply and demand adapt in hard to predict ways but the caveat here is that the data is arguing against it.
Direct
jobs By the same reasoning, "normal" data centers don't create jobs either? I doubt we think that's the case considering the part of GDPs which are purely online nowadays.
I'm not not sure of the offsite jobs (online), but yeah they do create jobs, though they are limited in numbers whether onsite or offsite, a quick search reveals 20 to 50 permanent staff per shift and if we compare that same 17-football-field plot of land with a manufacturing plant then I'm sure it would create thousands of onsite jobs. Since the inception of AI, it has been synonymous with job losses. The core idea behind AI is that it causes job losses in the short term while creating new ones in the long term, exactly what we have seen with every major technological advancement in history.
I'm not convinced this is true, many nuclear reactors in France have towers, if not the majority. Beyond that, sure it consumes water, electricity etc, but that's just the way industry works. Like above, I see no one complain about the massive energy costs of making aluminum.
Many nuclear power plants don't use towers at all because Once-Through Cooling (No Towers) pulls water directly from a massive source and passes it through the condenser, and returns it. The point is if there is a lake or sea nearby you don't need a tower to cool down as it can afford to discharge warmer water back to a source as because the source is big. whereas, the cooling towers are used next to smaller rivers or lakes that don't have enough water supply and where environmental laws strict dumping warmer water back to the source. In short, cooling towers are massive evaporation engines. Instead of dumping hot water into a source, the plant uses the tower to dump that heat directly into the atmosphere as water vapors.
You made a valid point about the cost of aluminum, i looked at the numbers both are almost neck and neck, aluminum consumes 3% of the entire world’s electricity supply while the data centers between 2% to 3%, but the point here is data centers are only going to consume more and more energy going forward (Jevons paradox) as they haven't reached its peak, whereas aluminum is not likely to exceed this number because the world has already an established process or network. Aluminum also has a low S2F ratio of 5 (Gold has 62) meaning supply is highly elastic.
I believe it already exists. The two reasons why we don't talk about it: it is socially fine, even if you were to argue for discriminatory measures against men, and their parties mostly lose so that has less political impact. Women have radicalized more than men but when media talk about radicalization, they do not care about that part, on the contrary they encourage it. (South Korea really looks like a completely insane country the more you learn about it)
Yes, as per data it exists and the gap is widening by a huge margin according to FT. The same point I'm making, I think the recent widening of gap is happening everywhere in the world or at-least in the big countries. As women are getting advanced degrees they are equipping themselves with a necessary knowledge and a confidence to speak up in relevant matters.
1
Reply(0)Changed my mind (0)Mind (0)(0)
hide 3 replies
exmateriae· 2 weeks agoMay 30, 2026
@Russell-viper pretty much agree with everything! And to the risk of repeating myself, I was in agreement with the original post anyway, it was mostly nitpicking. Sorry for the late response
I feel like btc mining is overestimated and dwarved by OOMs by AIs but don't remember the actual values so may be wrong.
Completely agree on China, there are a few forecasting questions here exactly on that! While I was with EpochAI we actually thought about that and I think there is a lot more elasticity on energy than people believe but you'll have to pay for it. (like double or triple the normal price but electricity is a small portion of AI spending so they don't care) Looking at AI spending, that seems doable, the harder sell is local permitting and regulation.
To be clear my point about data centers is that you should not look at the people working in the data centers but at all of the businesses which could not exist without these data centers (a lot) and then apply that to AI: a lot of value is going to be created by these data centers which will not bring direct jobs but will create some indirect ones.
I think most nuclear points in France try to use both methods but I may be wrong as I've rarely seen one without towers or away from a water source. True about aluminum!
I think advanced degrees are overrated in their usefulness (I have a PhD) honestly, I think there have been more important factors at play that shifted women position in society but anyway that's besides the point
0
Reply(0)Changed my mind (0)Mind (0)(0)
hide 2 replies
Russell-viper· 2 weeks agoJun 1, 2026
@exmateriae Some analysts share the opinion that bitcoin will go bust as it doesn't create anything valuable in the economy and in exchange eats up electricity. The point that bitcoin is valueless comes from the idea that it has no practical purpose. Price isn't the right indicator of any sort of value, people buy bitcoin mere for the very reason that someone in future will buy it at higher price. Though no one can predict its collapse. Anyway it's not the topic.
completely agree with the point you make in data centers contributions to the jobs or economy, and i think we both share the same opinion that in the short run the pain is inevitable, the point is things will go berserk before they go the right path.
Yes, i agree that advanced degrees aren't the sole criteria, it's one variable among others.
1
Reply(0)Changed my mind (0)Mind (0)(0)
show 1 replies
darkives· 3 weeks agoMay 26, 2026
the US which is 90% desert
This is a minor quibble in the grander scheme of your comment, but I just wanted to flag that this is not at all right! We do have some deserts, but they are substantially less than half of the country's area.
0
Reply(0)Changed my mind (0)Mind (0)(0)
hide 1 replies
exmateriae· 3 weeks agoMay 27, 2026
@darkives my whole comment was minor quibbles!
I meant population deserts and not strictly deserts in the Arizona sense. That was not clear enough yeah. 90% is still too high, even with what I had in mind, this is likely between 50-80% depending on how strict you are on the criteria
1
Reply(0)Changed my mind (0)Mind (0)(0)
WasteTimeContinuum· 3 weeks agoMay 22, 2026·(edited 4 days ago(Jun 10, 2026))
This is a good question, because it is highly uncertain. I haven't researched this yet, but my initial impression is that I don't think this will not happen before September. It takes a lot of time to move the needle when it comes to policy change. And pay close attention to the word "new" in the resolution criteria. I don't think reactivating paused restrictions count as a "Yes."
0
Reply(0)Changed my mind (0)Mind (0)(0)
jordan1casady· 3 weeks agoMay 22, 2026
I’m forecasting 38% Yes / 62% No. My pattern read is Grid Saturation → Political Gatekeeping. The underlying pressure is real: AI and large-scale data centers are pushing grid systems into capacity stress, and governments are beginnin...
DaiLothbrook· on May 12May 12, 2026
@darkives do you beleive it is still possible to add this question in the cup summer this week?
0
Reply(0)Changed my mind (0)Mind (0)(0)
hide 6 replies
darkives· last monthMay 17, 2026
@DaiLothbrook Thanks for the question. For future reference, to submit a question for the Cup or any other tournament, you can click on the "+ Question" button on the tournament page, and it will be considered. (You can also select the tournament from the "Advanced Options" within the question-creation page.) I've now done this for this question, so it will be considered for the Cup (though perhaps in a future week). I also made a number of edits, including adding some links to the background and simplifying the resolution criteria.
0
Reply(0)Changed my mind (0)Mind (0)(0)
hide 5 replies
DaiLothbrook· last monthMay 20, 2026
@darkives Thanks for the answer. Would like to point out somethings: 1- I did press the button inside the Cup page. 2- I tried also to choose in the advanced options but it didnt show up the option for the kind of tournament we wanted to submit, only the tags options to say if it was related to geopolitics, etc.. I will pay attention on it next times, if I dont see it I'll let you know. 3- Can you change the dates for 'Before September" ? I believe the question will benefit to have more time to be monitored given the nature of the event. I Chose August because I didnt know until when the cup would last. so here it changes for:" A restriction that is formally announced and then officially reversed or rescinded before August 31, 2026, at 23:59 UTC, will not cause this question to resolve as Yes." Thank you
0
Reply(0)Changed my mind (0)Mind (0)(0)
hide 4 replies
darkives· last monthMay 20, 2026
@DaiLothbrook The advanced options are hidden in a dropdown box that appears if you click below the geopolitics, etc. tags field.
I've updated the question for the new September deadline.
1
Reply(0)Changed my mind (0)Mind (0)(0)
hide 1 replies
skmmcj· 3 weeks agoMay 22, 2026
@darkives @DaiLothbrook Made some small edits to the title and criteria/fine print.
1
Reply(0)Changed my mind (0)Mind (0)(0)
skmmcj· last monthMay 21, 2026
1- I did press the button inside the Cup page.
Each tournament has its own button to add questions, separate from the +Create button at the top:
1
Reply(0)Changed my mind (0)Mind (0)(0)
hide 1 replies
DaiLothbrook· 3 weeks agoMay 23, 2026
@skmmcj Yep, I have tried it , I believe have found it . Thank you
0
Reply(0)Changed my mind (0)Mind (0)(0)
5 comments20 forecasters\ US 2030 Electricity Below 11 cents per kWh? \ \ 16%chance
4 comments22 forecasters\ ai lab aluminum 2030 \ \ 5%chance
We use cookies 🍪 to understand how you use Metaculus and to improve your experience.
Learn more about how we use cookies in our Privacy Policy
Necessary onlyCustomize
Accept and close