Share
The United States is home to 51% of the 1,000+ hyperscale data centers in the world. Most of these (60%) hyperscale data centers are owned and used by tech giants like Amazon, Microsoft, and Google for their cloud services.
Technically, data centers only have to exceed 10,000 sq ft in size to qualify as “hyperscale.” But the ones owned by cloud service providers are more than just bigger versions of standard data centers – they’re titans of the tech industry.
If you want to learn more about the best hyperscale data centers, Scout Cities has all the resources you need.
What is a Hyperscale Data Center?
A hyperscale data center is a massive, specialized facility designed to support the computing needs of large-scale cloud service providers and internet giants. True to the same, these centers can span hundreds of thousands of square feet. They house hundreds of thousands or even millions of servers.
To qualify as hyperscale, a data center typically consumes more than 50 megawatts of power, enough to power a small town of 50,000.
They provide the infrastructure required to deliver cloud services like IaaS, SaaS, and PaaS. These services rely on the massive computing power and storage capacity of hyperscale facilities.
Hyperscale data centers also support data-intensive tasks that involve artificial intelligence (AI), machine learning, and data analytics.
The Titans of Tech: Top Hyperscale Data Centers
How does a data center qualify as “hyperscale”? It’s not just about being big..
Size and Scale
A hyperscale data center typically exceeds 10,000 sq ft in size. It must accommodate more than 5,000 servers for massive data processing and storage capabilities.
For instance, the Inner Mongolian Information Hub, owned by China Telecom, has a total area of 10.7 million square feet. The Switch SuperNAP is another good example, covering 3.3 million sq ft in Los Angeles.
Power Capacity
Hyperscale data centers generally provide at least 40 megawatts (MW) of power capacity.
Technology and Infrastructure
Hyperscale data centers often use standardized designs for servers, cooling systems, and power supplies to improve efficiency and reduce costs. They also support network connections at 40 Gbps or faster.
Many hyperscale data centers aim for low Power Usage Effectiveness (PUE) metrics, which indicate a strong focus on energy efficiency. For instance, Google's facilities report PUE values as low as 1.1.
Key Players in the Hyperscale Data Center Arena
Let’s take a deep dive into some of the top hyperscale data centers in the world.
Google Data Center, The Dalles, Oregon
The Google data center opened in 2006 and was Google's first company-owned data center.
It’s located in The Dalles, a small city with a population of about 16,000. Due to Google's presence in the city, The Dalles has seen major economic revitalization.
The exact size of the facility isn’t publicly available, but it’s designed to accommodate thousands of servers. The same goes for the facility’s exact power consumption. However, Google’s data centers globally consume around 258 million watts of power, equivalent to the energy needs of about 188,000 homes.
In 2022, the data center consumed 274.5 million gallons of water. Local residents were concerned about the sustainability of water resources in the region, especially during drought conditions.
As a result, Google decided to transfer groundwater rights to the City of The Dalles permanently. Now, the city receives over 100 million additional gallons each year.
The facility mainly uses all that water for evaporative cooling. That’s why Google’s cooling overhead is only 6%, compared to the industry average of 57%.
In terms of job creation, Google employs around 150 people in The Dalles. The company also uses renewable hydropower from the nearby The Dalles Dam, which is much cheaper than traditional energy sources.
There have been certain debates regarding Google's transparency, particularly concerning its water usage. The city of The Dalles initially tried to keep this information confidential. Eventually, the court ruled that the water usage data should be made public.
Microsoft Data Center, Quincy, Washington
The Quincy data center was opened in 2007, with property purchased in 2006. At one point, it was considered the largest data center in the world and remains one of Microsoft's major facilities.
The Quincy campus covers around 800,000 sq ft across multiple buildings on a 270-acre site. It includes facilities built to Microsoft's second-, third-, and fourth-generation designs, with plans for a fifth-generation facility.
According to 2012 data, the data center consumes between 30 to 50 megawatts of power, but current figures are likely higher due to increased operational demands. This is enough to support over 1 million servers, handle more than 30 trillion stored data objects, and process over 1.5 million network requests per second.
Other than that, the Quincy data center uses adiabatic cooling and traditional hot aisle/cold aisle containment strategies. This allows them to maintain a Power Usage Effectiveness (PUE) of below 1.4. Plus, sourcing their hydroelectric power from the nearby Grand Coulee Dam allows Microsoft to get electricity at rates as low as 1.9 to 2.5 cents per kilowatt-hour.
The Quincy data center employs around 100 permanent staff, with around 30 of those in IT-related roles.
Microsoft has faced scrutiny over its water consumption in Quincy, especially given the region's agricultural needs. The brand is currently working to manage its environmental footprint.
Amazon Web Services (AWS) Data Center, Northern Virginia
AWS has invested $35 billion in data centers located in Northern Virginia over the last decade for various data centers in:
- Ashburn: 8 data centers
- Sterling: 10 data centers
- Manassas: 7 data centers
- Chantilly: 5 data centers
- Dulles: 3 data centers
- Herndon: 1 data center
- Haymarket: 3 data centers
Today, Amazon has more than 50 data centers in Northern Virginia, the largest single concentration of corporate data center infrastructure in the world. AWS is also planning to build a hyperscale data center campus just south of the Dulles Airport, which could add 2.5 million square feet of cloud capacity.
In 2020 alone, Amazon’s investments in the construction and operation of this data center contributed $1.3 billion to the GDP and supported over 13,500 jobs in Virginia's economy. It’s also among the biggest private-sector employers in Virginia, with over 8,800 full-time, well-paying jobs in data centers.
AWS has also paid over $220 million in business personal roperty taxes in 2020 for its data centers located in Fairfax, Loudoun, and Prince William counties. This amounts to 20% of personal property tax revenues received by these counties in the 2020 fiscal year.
By 2023, AWS plans to build 15 new utility-scale solar farms across Virginia, with a total capacity of 1,430 megawatts (MW) — enough renewable electricity to power 225,000 households, more than in Prince William County.
Facebook Data Center, Prineville, Oregon
The Facebook Data Center in Prineville, Oregon, covers around 3.2 million sq ft. Facebook is adding two more data centers to the area, increasing the total area to 4.6 million sq ft. Each of the new data halls will be 450,000 sq ft.
The Prineville facility was Facebook's first data center, opened in 2009. Over the years, the company has invested more than $2 billion in this campus, making it the company's largest data center complex.
Thanks to its high standards in energy efficiency and environmental design, the facility has even achieved a LEED Gold certification. Its PUE of 1.07 makes it one of the most energy-efficient data centers in the world. This means it operates 38% more efficiently than other hyperscale data centers.
The data center has an evaporative cooling system and an airside economizer, which brings in 100% ambient air. This system allows the facility to run without mechanical cooling, using dry desert air and misters to maintain optimal temperatures.
Other than that, it's powered by 100% renewable energy, including solar panel farms financed by Facebook in Oregon.
In terms of job creation, the data center employs around 350 people. Facebook's presence has also majorly contributed to the local economy, providing tax revenues and supporting many community projects. The company has saved nearly $130 million in local tax breaks from 2012 through 2020.
Facebook provides grants to local educational programs and supports multiple initiatives in Oregon. For instance, they recently announced a $60,000 grant for the Crook County School District's K-12 robotics program.
Apple Data Center, Maiden, North Carolina
The Apple Data Center in Maiden, North Carolina, covers over 500,000 sq ft –184,000 sq ft of it is dedicated to raised floor space for colocation.
Apple has invested in three solar farms near the data center, generating a total of 40 megawatts (MW) of solar power. The facility also includes a 10 MW fuel cell installation that converts biogas from nearby landfills into electricity.
This combination of solar and fuel cell energy sources provides about 60% of the power required by the data center. Plus, Apple's renewable energy generation at the Maiden site produces enough energy to power approximately 14,000 homes!
This data center has also achieved LEED Platinum certification – the highest level attainable under the LEED rating system.
The facility's construction was fairly sustainable, using 14% recycled materials and diverting 93% of construction waste from landfills. Around 41% of purchased materials were sourced within 500 miles of the Maiden site to reduce transportation emissions.
Apple has invested over $2 billion in the Maiden data center since its establishment.
The Maiden data center uses high-efficiency solar cells with tracking systems to maximize energy efficiency and minimize environmental impact. By integrating on-site renewable energy generation, Apple has also made the data center resilient against power outages and grid instability.
Architecture and Design of Hyperscale Data Centers
Hyperscale data centers are pushing the boundaries of design and sustainability. Now, they’re able to meet the growing demands of cloud service providers and reduce their environmental impact.
Here are some key architectural designs they use:
- Modular and Scalable Design: Modern hyperscale data centers have modular components that can be easily added or replaced as capacity needs change.
- Standardized Hardware: Hyperscale operators use stripped-down hardware components that can be mixed and matched for better modularity.
- Innovative Cooling Systems: Hyperscale data centers use ultra-efficient cooling systems to reduce energy consumption. For instance, AirTrunk uses refrigerant-free and liquid cooling technologies.
- Sustainable Site Selection: Hyperscale data centers are strategically located to make the most of renewable energy sources. For example, The Citadel Campus in Nevada is powered by 100% renewable energy and guarantees low latency to major markets. To get the best data on the best sites, companies are submitting RFIs through Scout Cities.
Technologies Powering Hyperscale Data Centers
Hyperscale data centers run on highly advanced technologies to enhance energy efficiency, cooling, and data storage solutions. Here’s a breakdown.
- Liquid Cooling: Liquid cooling technology is more efficient than traditional air cooling, allowing for higher server densities and lower energy consumption. Facebook’s Prineville data center uses both evaporative cooling and liquid cooling to maintain optimal temperatures.
- Renewable Energy Integration: Hyperscale data center owners are aggressively investing in renewable energy to power their facilities with net zero emissions. For instance, Google, Microsoft, and Amazon are hoping to match 100% of their electricity consumption with clean energy.
- AI-Powered Cooling: AI algorithms are used to monitor and adjust cooling systems in real time. Microsoft uses AI to optimize cooling based on server load and environmental conditions.
- Operational Management: AI and machine learning can analyze data from sensors to identify potential failures before they occur. Facebook uses AI to monitor equipment health and predict maintenance needs.
- Automated Workflows: AI-powered automation allows for faster service deployment in hyperscale data centers. Google Cloud also uses AI to automate routine tasks in its data centers so its engineers can focus on higher-level strategic initiatives.
Conclusion
Hyperscale data centers are more than just places for tech giants to store their digital information. They’ve had huge impacts on the tech industry and the economies of the cities they’re located in.
For instance, Amazon Web Services (AWS) has invested over $108 billion in cloud computing infrastructure just for their data centers. Plus, it’s also encouraged tech giants to invest in more sustainable energy use practices and create tens of thousands of new jobs.
The future of hyperscale data centers looks bright, too, with the advent of direct-to-chip cooling and immersion cooling. The investments will also be huge – Saudi Arabia is spending $500 million on a hyperscale data center to support the development of the smart city Neom.
If you are considering locating a data center or need more information about data centers, you can sign up for Scout, and we will provide off-market real estate, local data, and incentive information.
