# New York suspends new data centers above 50 MW

Governor Kathy Hochul signed an order pausing new 50 MW+ data centers for up to a year while the state runs an environmental study. It targets AI's pressure on the grid, water use and cost-shifting to the public — and signals a new regulatory risk for anyone designing cloud and AI workloads.

- URL: https://fernando.moretes.com/blog/nova-york-suspende-data-centers-acima-de-50-mw

- Markdown: https://fernando.moretes.com/blog/nova-york-suspende-data-centers-acima-de-50-mw/article.md?lang=en

- Published: 2026-07-20T12:00:00.000Z

- Category: AWS & Cloud

- Tags: data-centers, sustentabilidade, energia, inteligencia-artificial, nova-york, regulacao, finops, well-architected

- Reading time: 7 min

- Source: [First Statewide Moratorium on New Hyperscale Data Centers](https://www.governor.ny.gov/news/first-statewide-moratorium-new-hyperscale-data-centers-launched-governor-kathy-hochul)

---

New York became the first U.S. state to impose a statewide moratorium on new hyperscale data centers — doing so via an executive order that pauses projects of 50 MW or more while the state assesses real impacts on the power grid, water supply, and air quality. The AI infrastructure boom just hit its first regulatory wall in the United States. If you design cloud workloads or AI platforms, this is not local news: it is a signal that region selection has become a risk decision.

## Key takeaways

- Governor Kathy Hochul's executive order (~July 14, 2026) pauses new or expanding data centers that would consume 50 MW or more of power in New York State.
- The pause lasts up to one year or until a Generic Environmental Impact Statement (GEIS) is completed, whichever comes first; DPS leads the study, examining energy demand, water use, and air quality.
- The stated rationale: AI-driven growth strains the power grid, raises electricity rates for homes and businesses, consumes large volumes of water, and shifts costs from cancelled projects onto the public.
- Excluded from the moratorium: facilities primarily for manufacturing, research, education, and medical care; permit applications already deemed complete before the order are also exempt.
- For architects: region selection now carries regulatory and energy risk; the Well-Architected Sustainability pillar is no longer optional; efficiency (Graviton, serverless, right-sizing, optimized inference) is the real lever — and this policy pattern will spread.

## What New York did

On approximately July 14, 2026, Governor Kathy Hochul signed an executive order directing the Department of Environmental Conservation (DEC) to pause discretionary state permit applications for **new data centers or expansions consuming 50 MW or more**. The pause applies to applications not yet deemed complete before the order was signed.

The maximum duration is **one year**, but it may end sooner if the state completes a **Generic Environmental Impact Statement (GEIS)**. The Department of Public Service (DPS) leads that study, focusing on energy demand, water use and quality, and air quality. Empire State Development (ESD) is also involved in the process.

Within **60 days**, the state must publish a **Community Investment Framework (CIF)**, creating a mechanism for host communities to negotiate local investment with operators. Additionally, DPS is directed to consider a 'NY Grid Acceleration Fund' funded by operators themselves, and the Governor is pursuing legislation to **repeal sales-tax exemptions** currently granted to large data centers.

## The numbers that matter

- **50 MW** — Moratorium threshold. Power draw that triggers the permit pause
- **≤ 1 ano** — Length of the pause. Or until the environmental study (GEIS) is done
- **426 TWh** — US data centers by 2030. +133% vs 2024 — up to 9% of the grid
- **20→40%** — AI's share of the energy. Of data-center energy by 2030

## Why now: AI and the pressure on the grid

The numbers justify the urgency. U.S. data centers consumed approximately **183 TWh in 2024**, representing more than 4% of all American electricity. Projections point to **426 TWh by 2030** — a **+133% increase** in six years. EPRI estimates data centers could reach **up to 9% of U.S. electricity generation** by the end of the decade.

Artificial intelligence is the primary driver of this acceleration. Today, AI workloads account for roughly **20% of data center energy consumption**; that share is expected to nearly double to approximately **40% by 2030**. Training large language models and serving real-time inference demands power density that traditional infrastructure was simply not designed to handle.

New York is not acting in a vacuum. The state has ambitious climate targets and a power grid already facing capacity pressures. Approving hyperscale projects without a systemic assessment would, in practice, mortgage the state's energy transition to subsidize private infrastructure expansion. The moratorium is, above all, an attempt to reclaim control over the pace of that transformation.

![US data center electricity is set to more than double by 2030 — from ~183 TWh (4% of the grid) to ~426 TWh (up to 9%).](/blog/ny-data-centers/energy-growth.svg)

_Pew Research / U.S. EIA / EPRI_

## Water, rates, and who pays the bill

Electricity is the most visible cost, but it is not the only resource under pressure. Hyperscale data centers depend on **large volumes of water for cooling** — evaporative cooling towers can consume millions of liters per day in a single facility. The GEIS led by DPS will examine exactly this impact on water use and quality, a point frequently overlooked in debates about IT infrastructure sustainability.

On the economic side, the Governor's argument is straightforward: when data center demand strains the grid, the costs of expanding and modernizing electrical infrastructure are often socialized — passed on through **higher energy rates for homes and small businesses**. Similarly, when large-scale projects are cancelled or scaled back, infrastructure costs already committed can fall on the public.

A June 2026 Siena poll found that **46% of New Yorkers** considered a one-year moratorium a good idea, versus only **21% opposed**. That political data point matters: it signals that energy regulation of data centers has popular support, making similar policies more likely in other states.

![Among New Yorkers, support for the moratorium (46%) beats opposition (21%) by more than 2 to 1.](/blog/ny-data-centers/siena-poll.svg)

_Siena College Research Institute, jun/2026_

## What the moratorium excludes

The executive order was drafted with deliberate exclusions that reveal the state's priorities. **Facilities whose primary use is manufacturing, research, education, or medical care** are out of scope — even if they consume 50 MW or more. This protects hospitals, universities, research laboratories, and industrial plants from being inadvertently stalled by a rule aimed at cloud and AI operators.

Equally important: **permit applications already deemed complete** by DEC before the executive order was signed continue to proceed normally. This avoids retroactivity and protects investments that had already moved through the regulatory process.

These exclusions matter to architects because they define where regulatory pressure will **not** fall — at least for now. Academic research workloads or hospital systems hosted on correctly classified dedicated infrastructure may be in a different position than a general-purpose AI inference platform. Scope matters as much as the rule itself.

## What the moratorium covers vs excludes
| Criterion | Subject to the pause | Out of scope |
| --- | --- | --- |
| Project type | Hyperscale data centers ≥ 50 MW (new or expansion) | Manufacturing, research, education and healthcare |
| Permitting | State applications not yet deemed complete | Applications already complete before the order |
| Energy trigger | ≥ 50 MW power draw | Below 50 MW |

## What this means for cloud and AI architects

I will be direct: **region selection is no longer just a latency and compliance decision**. It now carries regulatory and energy risk. Before anchoring an architecture in us-east-1 or any region with known capacity pressure, you need to understand that region's energy profile, the state or country's climate targets, and the likelihood of future constraints.

The **AWS Well-Architected Sustainability pillar** is no longer a checklist item for ESG reports. It is an operational lever. Workload efficiency — **aggressive right-sizing, Graviton adoption, serverless architectures, optimized inference with smaller or quantized models** — directly reduces the energy footprint. This is not altruism; it is architectural resilience against a tightening regulatory environment.

FinOps needs to evolve into what I am calling **'PowerOps'**: energy as a first-class constraint, not just a derived cost. This means tracking consumption per workload, optimizing scheduling for lower carbon-intensity windows, and using tools like the **AWS Customer Carbon Footprint Tool** and **carbon-aware placement** — available today via APIs such as the Carbon Aware SDK.

Most importantly: **design for the pattern to spread**. New York was the first state, but it will not be the last. Virginia, Texas, Oregon — all major data center hubs — face similar grid and water pressures. Architectures that depend on unlimited capacity expansion in a single region are fragile architectures. **Multi-region design, workload portability, and efficiency are not nice-to-haves; they are regulatory hedges.**

> **The signal behind the news:** Energy is becoming the binding constraint on compute — not bandwidth, not latency, not instance cost. When a state imposes a moratorium on physical infrastructure, it is essentially saying that the cloud abstraction layer no longer insulates architects from the consequences of the physical world. **Workload efficiency is the new competitive edge**: those who can deliver the same outcome with fewer watts will have access to capacity, lower costs, and reduced regulatory exposure. Ignore this and you will be designing on a foundation that regulators in multiple jurisdictions are actively questioning.

## What comes next

The GEIS has up to one year to be completed, but DPS may finish sooner — and the outcome will define the permanent regulatory framework for data centers in New York. Watch the **GEIS public hearings** and the **Community Investment Framework** due within 60 days: those documents will reveal where the state intends to land.

In the broader landscape, **other states are watching**. Virginia — home to the world's largest data center cluster — is already debating grid capacity constraints. Texas and Oregon face water and generation pressures. The European Union already operates with energy efficiency regulation for data centers under the **European Green Deal**. The regulatory pattern is converging globally.

For architects, the next 12 months are a window to **audit regional capacity dependencies, revisit workload placement strategies, and build the business case for energy efficiency** — before the regulatory environment forces that conversation in a far less comfortable way.

## Verdict

New York did not stop AI growth — it paused irresponsible infrastructure expansion without impact assessment. The distinction matters. What this order signals to the industry is that **the era of capacity growth without regulatory friction is over**. Architects who treat energy efficiency as a design differentiator — rather than last-minute compliance — will build systems that are more resilient, cheaper to operate, and more defensible before regulators. The question is no longer 'which region do I deploy to?'; it is 'will this region still have available capacity in two years, and at what political and energy cost?'

## Sources

- [Governor Hochul — First Statewide Moratorium on New Hyperscale Data Centers](https://www.governor.ny.gov/news/first-statewide-moratorium-new-hyperscale-data-centers-launched-governor-kathy-hochul)
- [CNBC — New York becomes first U.S. state to impose AI data center ban](https://www.cnbc.com/2026/07/14/new-york-ai-data-center-ban.html)
- [Data Center Knowledge — New York Data Center Moratorium](https://www.datacenterknowledge.com/data-center-construction/new-york-data-center-moratorium-state-pauses-projects-over-50-mw)
- [Pew Research — Energy use at US data centers amid the AI boom](https://www.pewresearch.org/short-reads/2025/10/24/what-we-know-about-energy-use-at-us-data-centers-amid-the-ai-boom/)
- [EPRI — Data Centers Could Consume up to 9% of US Electricity](https://www.epri.com/about/media-resources/press-release/q5vu86fr8tkxatfx8ihf1u48vw4r1dzf)
- [IEA — Energy demand from AI](https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai)