Data Collection Off the Tracks: Old NSA Data Center vs. New Hyperscale AI Data Centers

Revealed: The Old NSA Data Center Had the Capacity to Track and Store 100 Years’ Worth of Data on Not Only Americans but Also Every Person Worldwide

For years, brave whistleblowers sounded the alarm about the U.S. government’s growing surveillance capabilities that threaten our privacy and freedom. Edward Snowden exposed the NSA’s broad bulk data collection programs (like PRISM and metadata sweeps on phone records), revealing a “collect it all” mindset that raised major alarms about government surveillance of Americans. Wikipedia – Utah Data Center

William Binney, a former NSA technical director and earlier whistleblower, was far more specific about the old NSA Utah Data Center’s capabilities in Bluffdale. Speaking in 2012-2013 about that facility (long before any new projects like Stratos existed), he estimated it could handle 5 zettabytes of data (with some references to even larger yottabyte-scale potential). In his words:

“They would have plenty of space with five zettabytes to store at least something on the order of 100 years worth of the worldwide communications, phones and emails and stuff like that.”

Binney warned the old NSA facility was built to vacuum up and store content (not just metadata) of communications for potential retroactive searching and analysis on a massive scale. He sounding the alarm that this infrastructure already enables unprecedented tracking and long-term storage of personal data from Americans and people worldwide. Skeptics Stack Exchange – Binney Quote

The old NSA center already possessed jaw-dropping capabilities to store and process vast amounts of personal communications for decades. Yet the new hyperscale AI data centers will dwarf it entirely in power and processing – potentially handling more information than humanity has ever generated or imagined before, thanks to AI-driven analysis at speeds and scales that turn “big data” into something truly unimaginable.

During the course of writing this article, intense lobbying and public pressure significantly altered the proposed Stratos AI data center I had originally planned to use as the primary example. The Stratos Project in Box Elder County, Utah, was dramatically scaled back from its original ambitions. I am therefore now comparing the revised and reduced plan to the old NSA Utah Data Center (the “Bumblehive”) in Bluffdale.

For additional perspective, below is a comparison of the old NSA Bumblehive to five of the largest hyperscale AI data centers proposed or under development in the United States (ranked roughly by announced power capacity and overall campus scale as of mid-2026).

The original plan for the Stratos Project envisioned a sprawling 40,000-acre campus in Box Elder County. That is an area larger than Manhattan and roughly the size of Washington, D.C., with up to 9 gigawatts of on-site natural gas power generation at full build-out and a projected cost of up to $100 billion. It was positioned to become one of the largest data center developments in the world, featuring dozens of massive AI-optimized buildings.

Public Pushback Forces Scaling Back

Fierce pushback from citizens in Box Elder County, environmental groups worried about impacts on the Great Salt Lake, and political leaders at the state level, including a public call from Utah Senate President J. Stuart Adams for a 75 percent reduction in the footprint, prompted the developer, Kevin O’Leary of O’Leary Digital, to scale the project back significantly in June 2026. The project area was reduced by about 20,000 acres, bringing the total project area down to roughly 20,000 acres, with the effective development footprint for buildings and infrastructure now focused on approximately 10,000 acres or less. Substantial portions of the remaining land are reserved for potential solar areas. 1

Even this scaled-back model of Stratos remains immense.

It will operate on a vastly larger scale and with far greater power and computational capacity than the old NSA Utah Data Center in Bluffdale (nicknamed Bumblehive). That facility, already operational by 2014, represented a major leap in surveillance infrastructure for its era. The new generation of AI hyperscale centers, even when trimmed by public pressure, dwarfs that capability in land area, electricity demand, data throughput, and especially in the speed and sophistication of AI-powered analysis they enable.

Old NSA Utah Data Center (Bluffdale / “Bumblehive”) vs. Scaled-Back Stratos Project (Box Elder County)

Category Old NSA Utah Data Center (Bluffdale) Scaled-Back Stratos Project (Box Elder County)

Overall Size About 1 to 1.5 million square feet total (roughly 20-30 football Focused development footprint of approximately 10,000 acres for data center

Category Old NSA Utah Data Center (Bluffdale) Scaled-Back Stratos Project (Box Elder County) fields). Only about 100,000 square feet dedicated to actual servers. Wikipedia – Utah Data Center buildings, power infrastructure, and support facilities within a reduced project area of roughly 20,000 acres after removal of about 20,000 acres of sensitive land. Salt Lake Tribune, June 4, 2026 FOX 13 News

Power Use 65 megawatts (MW) – enough to power roughly 30,000 to 65,000 average homes. Up to 7.5-9 gigawatts (7,500-9,000 MW) at full build-out in the revised plans (approximately 115 times the power of the old NSA center). [Salt Lake Tribune; FOX 13 News, June 2026]

Storage /

Compute

Capacity Estimated 3-12 exabytes, with some references to even larger yottabyte-scale potential in older claims. Focused on bulk storage of intelligence data such as emails and calls for later analysis. Skeptics Stack Exchange – Binney Quote Not publicly detailed in precise exabyte or zettabyte figures. Engineered specifically for hyperscale AI training and real-time inference workloads that require and generate vastly larger and more dynamic datasets than traditional intelligence storage facilities. The AI processing capabilities represent a fundamental escalation in analytical power.

Cost About 1.5 billion dollars to build. Wikipedia Projected in the tens of billions of dollars. (Original ambitious vision approached $100 billion.) [Stratos Project reporting]

Purpose NSA signals intelligence, cybersecurity, bulk data storage, and analysis in support of national security. Large-scale AI data centers and cloud computing infrastructure to support artificial intelligence development, with potential applications in national security and commercial sectors.

Water /

Cooling Approximately 1 million gallons or more per day for cooling systems. Water consumption remains a major local concern given the project’s location near the ecologically stressed Great Salt Lake. The developer has stated commitments to water recycling, closed-loop cooling where feasible, and potential dedication of excess water back

Category Old NSA Utah Data Center (Bluffdale) Scaled-Back Stratos Project (Box Elder County) to the lake system (subject to agreements). Full environmental reviews are pending. (Problems with closed loop water systems explained later in this article)

Status Fully operational since approximately 2014, following some early power-related delays. Revised design announced in June 2026. Key approvals were granted in May 2026 by Box Elder County and the Military Installation Development Authority (MIDA), but major construction has not started. Additional permitting and environmental analysis required. Build-out will occur in multiple phases over many years. MIDA Stratos materials; recent news reporting

Scale

Perspective A landmark government surveillance and data storage facility for the early 2010s – powerful in its time but constrained by the technology and mission of that era. Even after meaningful reductions this project stands as one of the largest planned hyperscale AI infrastructure campuses in the country. It is designed explicitly for the unprecedented data and compute demands of modern artificial intelligence and will possess capabilities that far exceed those of the previous generation of data centers in both commercial AI advancement and potential surveillance applications. Ars Technica, June 2026

Don’t Be Fooled by the Closed Water Loop System

Developers of the Stratos Project claim their closed-loop cooling system will save water by recycling the same fluid inside sealed pipes without constant refilling. While this approach does reduce overall consumption compared to traditional evaporative cooling towers, it still requires significant amounts of water. Even in a closed system, water is lost through evaporation in heat exchangers and cooling processes, especially at the massive scale of a hyperscale AI data center. In addition, the system needs regular maintenance flushes and

top-offs. A top-off means adding fresh water to replace what has been lost through evaporation or small leaks.

More concerning, operators must add chemicals such as biocides to prevent bacteria and algae growth, along with corrosion inhibitors to stop rust and wear on equipment. These toxic additives build up over time, and when the system is drained or blowdown water is released, the polluted mixture can harm local waterways, soil, and the already stressed Great Salt Lake ecosystem. Blowdown is the process of draining some of the concentrated, chemically treated water to prevent mineral and contaminant buildup, then replacing it with fresh water. What looks like a green solution on paper still demands huge volumes of water and creates real pollution risks for the surrounding community.

Comparing Bumblehive NSA Data Center to top five AI Data Centers in the United States

Old NSA Bumblehive (Bluffdale, UT) vs. each of the top five (by announced power capacity and campus scale as of mid-2026):

· Old NSA Bumblehive (Bluffdale, UT): ~65 MW power, ~1–1.5 million sq ft total site (~100k sq ft servers), ~$1.5B cost. Purpose: Bulk surveillance storage/analysis.

Picture of Bumblehive NSA data center below:

· vs. Delta Gigasite (Creekstone Energy, Millard County, Utah): >10 GW planned power (150x+ larger), 20+ million sq ft campus on thousands of acres. Positioned as potentially the world’s largest AI infrastructure site for hyperscale training/inference at unprecedented scale. Dwarfs Bumblehive in every infrastructure metric.

Stunning promotional/aerial image below of this planned >10 GW AI infrastructure campus below shows the enormous scale with integrated power generation and vast solar arrays:

· vs. Meta Hyperion (Richland Parish, Louisiana): Up to 5 GW power (scalable; phases starting at 2+ GW, ~77x larger), ~3,650 acres, 4+ million sq ft. One of the most ambitious single-company AI campuses built for advanced AI model training — vastly exceeds Bumblehive in power, land, and AI analytical capability.

At Amazon’s biggest data center, everything is supersized for AI

Picture below of Meta Hyperion (Richland Parish, Louisiana):

· vs. Major AWS / Stargate-related sites (e.g., New Carlisle, Indiana or Abilene, Texas clusters): Multi-GW scale (~2+ GW power, ~34x larger) with significant AI focus and massive campus footprint. Hyperscale AI/cloud workloads far beyond the old NSA facility’s surveillance focus.

Real aerial photo below of the large Stargate Abilene campus (already live and expanding) built for hyperscale AI workloads including OpenAI’s Stargate project:

· vs. Microsoft large AI campuses (e.g., Wisconsin, Texas, or other hyperscale clusters): Multiple 1+ GW sites (15x+ larger) with advanced GPU/liquid cooling designs. Represents a generational leap in compute power over Bumblehive-era tech. Picture of Microsoft’s Wisconsin campus below. The world’s most

powerful data center 2:

· vs. Google large clusters (e.g., Columbus, Ohio or similar multi-site campuses): Exceeding 1 GW total per key location, heavily AI-optimized with far greater scale, speed, and analytical sophistication than the 2010s-era Bumblehive.

These modern AI facilities dwarf the old Bumblehive in raw size, power demand, and especially in real-time AI processing capabilities.

Even If You Don’t Live Near a Data Center, This Threatens Your Freedom and Everyday Life

The infrastructure for total surveillance is not coming. It is already here and expanding at breakneck speed. These facilities do not just store data. They enable real-time tracking, retroactive searches, and AI-powered analysis of every aspect of our lives. They will collect information from all the cameras at every intersection (such as Flock Safety license plate readers), every Ring doorbell and home security camera, the entire Internet of Things (smart thermostats, watches, appliances, lights), smartphones and apps, social media platforms, financial transactions, cell towers, public WiFi, smart city infrastructure, drones, and countless other connected devices, creating a comprehensive digital picture of our daily lives.

We need to stop these data centers entirely, not just accept compromises on smaller versions or temporary moratoriums.

This is a high-speed train racing toward disaster for privacy and individual freedom. The battle appears already lost at the national level, with President Trump clearly supporting the rapid expansion of these AI data centers. We can only hope to slow this train down by fighting fiercely at the state, county, and city levels. Demand transparency and bans from

local officials, support community opposition efforts like those in Box Elder County, Nashville and all over the country. Speak out boldly.

Even if you do not live near one of these proposed hyperscale facilities, you must join the battle. These data centers will power the infrastructure for total surveillance, enabling real-time tracking, AI analysis of every aspect of your life.

Even if you do not live near one these AI data centers will power widespread digital money (CBDCs/Stablecoins or both) and digital IDs. And in doing so, will it hand governments and corporations unprecedented control over our financial lives and everyday personal freedoms.

Even if you do not live near one your water and power rates will increase dramatically and local watersheds will be drained and polluted.

Question for every American: Will the United States position itself as the central hub for “global data collection” with all its powerful data centers?

Our constitutional rights and way of life may depend on how effectively we resist at the grassroots level. The time to act is now.

Be a warrior, not a worrier.

Source: https://www.sgtreport.com/2026/06/data-collection-off-the-tracks-old-nsa-data-center-vs-new-hyperscale-ai-data-centers/

Fight the Censorship Join the Hopegirl Mailing List and get my free AudioBook: The QEG Chronicles Free Energy Full Disclosure: https://bit.ly/HGQEGBook

HopeGirl Youtube: https://www.youtube.com/@hopegirlblog
HopeGirl Telegram Channel Chat and Groups: https://t.me/Hopegirl587 
Get Our Free EMF Survival Guide : https://bit.ly/FTWEMF
10 Percent Off our EMF Protection Products with this Link: https://www.ftwproject.com/ref/tenoff

The post Data Collection Off the Tracks: Old NSA Data Center vs. New Hyperscale AI Data Centers appeared first on HopeGirl Blog.

Source: https://www.hopegirlblog.com/2026/07/06/data-collection-off-the-tracks-old-nsa-data-center-vs-new-hyperscale-ai-data-centers/