According to a 2024 forecast, global electricity consumption of data centers was projected to grow from 330 terawatt-hours in 2022 to over one petawatt-hour in 2030. This would represent around 3.7 percent of the total electricity consumption worldwide by the end of the period under consideration. Artificial intelligence accounted for around 4.5 percent of the data centers' electricity consumption in 2023. This figure is projected to grow over the next five years.
Traditional data centers globally have decreased their energy demand, from around 97.6 terawatt hours in 2015, to some 50 terawatt hours in 2019, and a forecast indicated that this figure will reach nearly 33 terawatt hours by 2021. On the other hand, hyperscale data centers have doubled their energy demand in the same period of time.
Data centers accounted for more than one fifth of the total electricity consumed in Ireland in 2023. This represented a four-fold increase on 2015, with Irish data center construction driven by an attractive tax environment and strong digital infrastructure.
Google’s energy consumption has increased over the last few years, reaching 25.9 terawatt hours in 2023, up from 12.8 terawatt hours in 2019. The company has made efforts to make its data centers more efficient through customized high-performance servers, using smart temperature and lighting, advanced cooling techniques, and machine learning. Datacenters and energy Through its operations, Google pursues a more sustainable impact on the environment by creating efficient data centers that use less energy than the average, transitioning towards renewable energy, creating sustainable workplaces, and providing its users with the technological means towards a cleaner future for the future generations. Through its efficient data centers, Google has also managed to divert waste from its operations away from landfills. Reducing Google’s carbon footprint Google’s clean energy efforts is also related to their efforts to reduce their carbon footprint. Since their commitment to using 100 percent renewable energy, the company has met their targets largely through solar and wind energy power purchase agreements and buying renewable power from utilities. Google is one of the largest corporate purchasers of renewable energy in the world.
Meta’s electricity use has increased in recent years, as newer data centers have come online. In 2022, the company's electricity usage surpassed **** terawatt-hours, a 22-percent year-over-year increase. Before 2021 the company was known as Facebook.
Meta’s renewable energy commitments
Over the past few years, Facebook/Meta has been moving towards using more renewable energy, with the goal of achieving 100 percent renewable energy to support facilities reached in 2020. This establishment of renewables also allows other nearby businesses to have the option to use renewable energy. Meta is one of the largest corporate purchasers of renewable energy worldwide.
Emissions and water consumption
Meta has set goals to reduce its carbon footprint by ** percent in 2030. In recent years, the company was able to separate growth in the business from increased emissions, annually reducing their operational greenhouse gas emissions. The company is also aware of its water consumption and has committed to a circular system that allows for the reuse of water consumed.
As of the second half of 2024, Greater Tokyo's data center operational capacity reached approximately ***** megawatts (MW), with an addition of about *** MW under construction, and ***** MW planned for development. In comparison, Ho Chi Minh City had around ** MW of data center capacity in operation, and about ** MW in the development pipeline. Although data center capacities and expansion paces vary among metro markets in the Asia-Pacific (APAC) region, the region as a whole is experiencing significant growth, positioning it as a key player in the global data center landscape.
Rapid growth in emerging APAC data center markets
While APAC’s mature metro markets like Tokyo, Singapore, and Sydney remain important for capacity development, emerging data center markets in Southeast Asia, including Johor, Jakarta, Manila, and Bangkok, are growing at a faster pace. For example, driven by growth in Johor, Malaysia’s data center capacity was forecasted to grow at *** percent between 2023 and 2026, compared to around ** percent for Japan. These rising metro areas are attractive to investors seeking secondary regional hubs, offering lower land and power costs, improved connectivity, and supportive government policies.
APAC markets compared to other key global data center markets
Leading APAC data center markets like Tokyo and Singapore each boast over one gigawatt (GW) of live capacity, though they still trail behind Northern Virginia - the largest data center market in the U.S. and globally, with more than *** GW of capacity. In Europe, London leads in data center inventory with just over *** GW, followed by Frankfurt at around *** MW.
https://www.statsndata.org/how-to-orderhttps://www.statsndata.org/how-to-order
The Combined Heat and Power (CHP) System for Data Centers market is emerging as a vital component in the quest for energy efficiency and sustainability in the digital age. By simultaneously generating electricity and useful heat from a single energy source, CHP systems offer data centers a significant solution to re
https://www.marketreportanalytics.com/privacy-policyhttps://www.marketreportanalytics.com/privacy-policy
The Oslo data center market, while lacking precise figures in the provided data, exhibits significant growth potential, mirroring global trends. The 9.33% CAGR indicates a robust expansion trajectory, driven by increasing digitalization across sectors like cloud computing, BFSI, and e-commerce within Norway and the broader Nordics region. The market segmentation reveals a diverse landscape, with data centers varying in size (from small to mega-scale facilities) and serving different tiers of cities and towns. Demand is fueled by the need for low-latency connectivity and high availability, particularly benefiting Tier 1 and 2 cities in Norway. The utilization rates, segmented into colocation types (retail, wholesale, hyperscale) and end-users, highlight the key drivers. Hyperscale deployments are likely to contribute significantly to growth, alongside the expanding needs of cloud providers and other major technology companies. While specific restraints aren't detailed, factors like energy costs and sustainability concerns, common in data center operations, may influence market dynamics. Competitive analysis points to the presence of both established international players and local Norwegian providers, creating a dynamic market with potential for consolidation and further investment. The geographical spread across Norway, likely concentrated in Oslo and surrounding areas, warrants further investigation to understand regional nuances. Considering the provided 9.33% CAGR and a lack of specific market size (XX), a reasonable estimation for the 2025 Oslo data center market size can be made based on comparable regional markets. Assuming a comparable market size to other Nordic capitals, and adjusting for Norway's specific economic conditions, a plausible starting point would be in the range of several hundred million USD. Given this estimate and the CAGR, future year predictions can be derived, although the precision is limited by the absence of baseline data. Market share analysis, while mentioned, is not provided, meaning precise individual company performance remains unknown. However, the list of companies suggests a mix of both established global players and smaller regional players. The absence of detailed regional breakdowns beyond "Nordics" requires further research to assess precise market segmentation and penetration within specific Norwegian regions beyond Oslo. Further analysis integrating local economic data and energy sector information would significantly enhance the insights. Recent developments include: September 2022: Bulk announced several expansion initiatives at its Norwegian data center locations, with investments focused on ensuring long-term power and land availability. Highly connected and scalable sites powered by 100% renewable energy are provided. With many European locations battling with power restrictions and increasing demand for data center capacity, Bulk completed the installation of the N01 onsite substation, which provides 125 MVA of dual connections to the adjacent Kristiansand substation., September 2022: Stack Infrastructure successfully connected an Oslo data center to the local district heating system four years after DigiPlex began the project. The OSL01 data center has been linked to Hafslund Oslo Celsio's district heating system in the Norwegian capital. Following a one-year ramp-up period, waste heat from the facility is now expected to provide heat and hot water for up to 5,000 houses.. Notable trends are: Tier 3 is Expected to Hold Significant Share of the Market.
https://www.statsndata.org/how-to-orderhttps://www.statsndata.org/how-to-order
The Single Phase Lithium-ion UPS (Uninterruptible Power Supply) market has been gaining significant traction due to the increasing demand for reliable backup power solutions across various sectors, including healthcare, data centers, telecommunications, and retail. The adoption of lithium-ion technology in UPS syste
As of March 2025, there were a reported 5,426 data centers in the United States, the most of any country worldwide. A further 529 were located in Germany, while 523 were located in the United Kingdom. What is a data center? A data center is a network of computing and storage resources that enables the delivery of shared software applications and data. These facilities can house large amounts of critical and important data, and therefore are vital to the daily functions of companies and consumers alike. As a result, whether it is a cloud, colocation, or managed service, data center real estate will have increasing importance worldwide. Hyperscale data centers In the past, data centers were highly controlled physical infrastructures, but the cloud has since changed that model. A cloud data service is a remote version of a data center – located somewhere away from a company's physical premises. Cloud IT infrastructure spending has grown and is forecast to rise further in the coming years. The evolution of technology, along with the rapid growth in demand for data across the globe, is largely driven by the leading hyperscale data center providers.
https://www.statsndata.org/how-to-orderhttps://www.statsndata.org/how-to-order
The Single-Phase Immersion Cooling System market is experiencing rapid expansion, driven by its pivotal role in addressing the increasing demands for efficient and reliable cooling solutions in data centers, power electronics, and various industrial applications. These systems immerse electronic components in a non-
CC0 1.0 Universal Public Domain Dedicationhttps://creativecommons.org/publicdomain/zero/1.0/
License information was derived automatically
..Table Name.Manufacturing Summary Series: General Summary: Detailed Statistics by Subsectors and Industries: 2012....ReleaseSchedule.Data are scheduled for release in August 2015....Universe.The universe includes all manufacturing establishments classified in sectors 31-33 with one or more paid employee at any time during the year.....GeographyCoverage.Data are shown at the U.S. level.....IndustryCoverage.Data are shown at the two-, three-, and six-digit North American Industry Classification System (NAICS) levels.....Data ItemsandOtherIdentifyingRecords.This file contains data on:...Number of companies.Number of establishments.Establishments with 0 to 19 employees.Establishments with 20 to 99 employees.Establishments with 100 employees or more.Number of employees.Annual payroll ($1,000).Total fringe benefits ($1,000).Employer's cost for health insurance ($1,000).Employer's cost for defined benefit pension plans ($1,000).Employer's cost for defined contribution plans ($1,000).Employer's cost for other fringe benefits ($1,000).Production workers average for year.Production workers for pay period including March 12.Production workers for pay period including June 12.Production workers for pay period including September 12.Production workers for pay period including December 12.Production workers annual hours (1,000).Production workers annual wages ($1,000).Total cost of materials ($1,000).Cost of materials, parts, containers, packaging, etc. used ($1,000).Cost of resales ($1,000).Cost of purchased fuels consumed ($1,000).Cost of purchased electricity ($1,000).Cost of contract work ($1,000).Quantity of electricity purchased for heat and power (1,000 kWh).Quantity of generated electricity (1,000 kWh).Quantity of electricity sold or transferred (1,000 kWh).Total value of shipments and receipts for services ($1,000).Value of resales ($1,000).Value added ($1,000).Total inventories, beginning of year ($1,000).Finished goods inventories, beginning of year ($1,000).Work-in-process inventories, beginning of year ($1,000).Materials and supplies inventories, beginning of year ($1,000).Total inventories, end of year ($1,000).Finished goods inventories, end of year ($1,000).Work-in-process inventories, end of year ($1,000).Materials and supplies inventories, end of year ($1,000).Gross value of depreciable assets (acquisition costs), beginning of year ($1,000).Total capital expenditures ($1,000).Capital expenditures for buildings and other structures ($1,000).Capital expenditures for machinery and equipment ($1,000).Capital expenditures for automobiles, trucks, etc. for highway use ($1,000).Capital expenditures for computers and peripheral data processing equipment ($1,000).Capital expenditures for all other machinery and equipment ($1,000).Total retirements ($1,000).Gross value of depreciable assets (acquisition costs), end of year ($1,000).Total depreciation ($1,000).Total rental payments or lease payments ($1,000).Rental payments or lease payments for buildings and other structures ($1,000).Rental payments or lease payments for machinery and equipment ($1,000).Total other operating expenses ($1,000).Temporary staff and leased employee expenses ($1,000).Expensed computer hardware and other equipment ($1,000).Expensed purchases of software ($1,000).Data processing and other purchased computer services ($1,000).Communication services ($1,000).Repair and maintenance services of buildings and/or machinery ($1,000).Refuse removal (including hazardous waste) services ($1,000).Advertising and promotional services ($1,000).Purchased professional and technical services ($1,000).Taxes and license fees ($1,000).All other operating expenses ($1,000)....Sort Order.Data are presented in ascending NAICS code sequence.....Related Data Files.Data supersede those released in the Industry Series and Geographic Area Series files...FTP Download.Download the entire table at https://www2.census.gov/econ2012/EC/sector31/EC1231SG1.zip....ContactInformation. U.S. Census Bureau, Economy Wide Statistics Division. Data User Outreach and Education Staff. Washington, DC 20233-6900. Tel: (800) 242-2184. Tel: (301) 763-5154. ewd.outreach@census.gov. . .For information on economic census geographies, including changes for 2012, see the economic census Help Center..Data based on the 2012 Economic Census. For information on confidentiality protection, sampling error, nonsampling error, and definitions, see Methodology. Data in this file represent those available when this file was created; data may not be available for all NAICS industries or geographies. Data in this table may be subject to employment- and/or sales-size minimums that vary by industry..Symbols:D - Withheld to avoid disclosing data for individual companies; data are included in higher level totalsN - Not available or not comparableFor a complete list of all economic programs symbols, see the Symbols Glossary.Source: U.S. Census Bureau, 2012 Economic Census.Note: The data in this file ar...
CC0 1.0 Universal Public Domain Dedicationhttps://creativecommons.org/publicdomain/zero/1.0/
License information was derived automatically
..Table Name.Mining: Summary Series: General Summary: Detailed Statistics: 2012....ReleaseSchedule.Data are scheduled to be released in September 2016.....Universe.The universe includes all mining establishments classified in sector 21 with one or more paid employee at any time during the year.....GeographyCoverage.Data are shown at the U.S. level.....IndustryCoverage.Data are shown at the two- through six-digit North American Industry Classification System (NAICS) levels.....Data ItemsandOtherIdentifyingRecords.This file contains data on:...Number of companies. Number of establishments.Establishments with 0 to 19 employees. Establishments with 20 to 99 employees.Establishments with 100 employees or more. Number of employees.Annual payroll ($1,000). Total fringe benefits ($1,000).Employer's cost for health insurance ($1,000). Employer's cost for defined benefit pension plans ($1,000).Employer's cost for defined contribution plans ($1,000).Employer's cost for other fringe benefits ($1,000).Production, development, and exploration workers for pay period including March 12. Production, development, and exploration workers annual hours (1,000).Production, development, and exploration workers annual wages ($1,000). Total cost of supplies ($1,000).Cost of supplies used, minerals received, and purchased machinery installed ($1,000). Cost of resales ($1,000).Cost of purchased fuels consumed ($1,000). Cost of purchased electricity ($1,000).Cost of contract work ($1,000). Quantity of electricity purchased for heat and power (1,000 kWh). Quantity of generated electricity (1,000 kWh).Quantity of electricity sold or transferred(1,000 kWh).Total value of shipments and receipts for services ($1,000).Value of resales ($1,000).Value added ($1,000).Total inventories, beginning of year ($1,000).Minerals products, crude petroleum, and natural gas liquids inventories, beginning of year ($1,000).Supplies, parts, fuels, etc. inventories, beginning of year ($1,000).Total inventories, end of year ($1,000).Minerals products, crude petroleum, and natural gas liquids inventories, end of year ($1,000).Supplies, parts, fuels, etc. inventories, end of year ($1,000).Capital expenditures (except land and mineral rights) ($1,000).Capital expenditures for buildings, structures, machinery, and equipment ($1,000).Capital expenditures for mineral land and rights ($1,000).Total rental payments or lease payments ($1,000).Rental payments or lease payments for buildings and other structures ($1,000).Rental payments or lease payments for machinery and equipment ($1,000).Total other operating expenses ($1,000).Temporary staff and leased employee expenses ($1,000).Expensed computer hardware and other equipment ($1,000).Expensed purchases of software ($1,000).Data processing and other purchased computer services ($1,000).Communication services ($1,000).Repair and maintenance services of buildings and/or machinery ($1,000).Refuse removal (including hazardous waste) services ($1,000).Advertising and promotional services ($1,000).Purchased professional and technical services ($1,000).Taxes and license fees ($1,000).All other operating expenses ($1,000)...Sort Order.Data are presented in ascending in NAICS code sequence.....Related Data Files.Data supersede those released in the Industry Series and Geographic Area Series files...FTP Download.Download the entire table at https://www2.census.gov/econ2012/EC/sector21/EC1221SG1.zip....ContactInformation. U.S. Census Bureau, Economy Wide Statistics Division. Data User Outreach and Education Staff. Washington, DC 20233-6900. Tel: (800) 242-2184. Tel: (301) 763-5154. ewd.outreach@census.gov. . .For information on economic census geographies, including changes for 2012, see the economic census Help Center..Data based on the 2012 Economic Census. For information on confidentiality protection, sampling error, nonsampling error, and definitions, see Methodology. Data in this file represent those available when this file was created; data may not be available for all NAICS industries or geographies. Data in this table may be subject to employment- and/or sales-size minimums that vary by industry..Symbols:D - Withheld to avoid disclosing data for individual companies; data are included in higher level totalsN - Not available or not comparableFor a complete list of all economic programs symbols, see the Symbols Glossary.Source: U.S. Census Bureau, 2012 Economic Census.Note: The data in this file are based on the 2012 Economic Census. To maintain confidentiality, the U.S. Census Bureau suppresses data to protect the identity of any business or individual. The census results in this file contain nonsampling error. Data users who create their own estimates using data from this file should cite the U.S. Census Bureau as the source of the original data only. For the full technical documentation, see Methodology link in above headnote.
https://www.statsndata.org/how-to-orderhttps://www.statsndata.org/how-to-order
The Single-Mode Optical Fiber Cables market has emerged as a cornerstone in modern communication networks, enabling high-speed data transmission over long distances with low signal loss. Utilized primarily in telecommunications, data centers, and broadband infrastructures, single-mode fiber cables offer a unique adv
CC0 1.0 Universal Public Domain Dedicationhttps://creativecommons.org/publicdomain/zero/1.0/
License information was derived automatically
(CLOB) ..Table Name.Mining: Geographic Area Series: Detailed Statistics for the State or Offshore Areas: 2012....ReleaseSchedule.Data are scheduled for release in September 2015.......Universe.The universe includes all establishments classified in mining sector 21 with one or more paid employee at any time during the year......GeographyCoverage.Data are shown at the state and offshore area levels.....IndustryCoverage.Data are shown at the two-digit North American Industry Classification System (NAICS) level.....Data ItemsandOtherIdentifyingRecords.This file contains data on:..Number of companies.Number of establishments.Establishments with 0 to 19 employees.Establishments with 20 to 99 employees.Establishments with 100 employees or more.Number of employees.Annual payroll ($1,000).Total fringe benefits ($1,000).Employer's cost for health insurance ($1,000).Employer's cost for defined benefit pension plans ($1,000).Employer's cost for defined contribution plans ($1,000).Employer's cost for other fringe benefits ($1,000).Production, development, and exploration workers for pay period including March 12.Production, development, and exploration workers annual hours (1,000).Production, development, and exploration workers annual wages ($1,000).Total cost of supplies ($1,000).Cost of supplies used, minerals received, and purchased machinery installed ($1,000).Cost of resales ($1,000).Cost of purchased fuels consumed ($1,000).Cost of purchased electricity ($1,000).Cost of contract work ($1,000).Quantity of electricity purchased for heat and power (1,000 kWh).Quantity of generated electricity (1,000 kWh).Quantity of electricity sold or transferred (1,000 kWh).Total value of shipments and receipts for services ($1,000).Value added ($1,000).Total inventories beginning of year ($1,000).Minerals products, crude petroleum, and natural gas liquids inventories, beginning of year ($1,000).Supplies, parts, fuels, etc. inventories beginning of year ($1,000).Total inventories, end of year ($1,000).Minerals products, crude petroleum, and natural gas liquids inventories, end of year ($1,000).Supplies, parts, fuels, etc. inventories, end of year ($1,000).Capital expenditures (except land and mineral rights) ($1,000) .Capital expenditures for buildings, structures, machinery, and equipment (new and used) ($1,000) .Capital expenditures for mineral exploration and development ($1,000).Capital expenditures for mineral land and rights.Total rental payments or lease payments ($1,000).Rental payments or lease payments for buildings and other structures ($1,000).Rental payments or lease payments for machinery and equipment ($1,000).Total other operating expenses ($1,000).Temporary staff and leased employee expenses ($1,000).Expensed computer hardware and other equipment ($1,000).Expensed purchases of software ($1,000).Data processing and other purchased computer services ($1,000).Communication services ($1,000).Repair and maintenance services of buildings and/or machinery ($1,000).Refuse removal (including hazardous waste) ($1,000).Advertising and promotional services ($1,000).Purchased professional and technical services ($1,000).Taxes and license fees ($1,000).All other operating expenses ($1,000).........Sort Order.Data are presented in state or offshore area by ascending NAICS code sequence.....FTP Download.Download the entire table at https://www2.census.gov/econ2012/EC/sector21/EC1221A2.zip....ContactInformation. U.S. Census Bureau, Economy Wide Statistics Division. Data User Outreach and Education Staff . Washington, DC 20233-6900. Tel: (800) 242-2184 . Tel: (301) 763-5154. ewd.outreach@census.gov. ..For information on economic census geographies, including changes for 2012, see the economic census Help Center..Data based on the 2012 Economic Census. For information on confidentiality protection, sampling error, nonsampling error, and definitions, see Methodology. Data in this table represent those available when this file was created; data may not be available for all NAICS industries or geographies. Data in this table may be subject to employment- and/or sales-size minimums that vary by industry..Symbols:D - Withheld to avoid disclosing data for individual companies; data are included in higher level totalsN - Not available or not comparableFor a complete list of all economic programs symbols, see the Symbols Glossary.Source: U.S. Census Bureau, 2012 Economic Census.Note: The data in this file are based on the 2012 Economic Census. To maintain confidentiality, the U.S. Census Bureau suppresses data to protect the identity of any business or individual. The census results in this file contain nonsampling error. Data users who create their own estimates using data from this file should cite the U.S. Census Bureau as the source of the original data only. For the full technical documentation, see Methodology link in above headnote.
London remained Europe's largest data center market in 2024, with total inventory in the British capital surpassing ***** megawatts as of the fourth quarter of the year. Frankfurt was the second-largest market, followed by Amsterdam. Major markets feel the strain Collectively referred to as the FLAPD markets, Frankfurt, London, Amsterdam, Paris, and Dublin are the data center hubs powering Europe's digital transformation. However, with data center demand surging across the world, operators are finding it increasingly difficult to secure power and land in these established markets. Local officials in Dublin, for example, have enforced a de facto moratorium on new facilities in the city, with data centers already accounting for more than a fifth of the country's total energy consumption. Striking a balance Rising demand poses a challenge for European authorities. On the one hand, the expansion of digital infrastructure is considered crucial if the region is to fully realize the economic potential of AI and next-generation technologies. On the other hand, the European Union faces increasing pressure to rein in the power and water intensive data center industry. Newly introduced regulations in the EU aim to address this by requiring major operators in the bloc to report key efficiency metrics, thereby helping to quantify the market’s drain on resources.
IPUMS-International is an effort to inventory, preserve, harmonize, and disseminate census microdata from around the world. The project has collected the world's largest archive of publicly available census samples. The data are coded and documented consistently across countries and over time to facillitate comparative research. IPUMS-International makes these data available to qualified researchers free of charge through a web dissemination system.
The IPUMS project is a collaboration of the Minnesota Population Center, National Statistical Offices, and international data archives. Major funding is provided by the U.S. National Science Foundation and the Demographic and Behavioral Sciences Branch of the National Institute of Child Health and Human Development. Additional support is provided by the University of Minnesota Office of the Vice President for Research, the Minnesota Population Center, and Sun Microsystems.
National coverage
Visiting points within Enumeration Areas (EAs)
UNITS IDENTIFIED: - Dwellings: No - Vacant units: No - Households: Yes - Individuals: Yes - Group quarters: Yes
UNIT DESCRIPTIONS: - Dwellings: An occupied dwelling was a premises (visiting point or physical address) that was inhabited by one or more households on census night. An occupied dwelling may have been a house, room, flat or apartment, shack, hut, tent, caravan, houseboat, shop, school, etc. - Group quarters: A special dwelling is one which is not privately occupied by a household. It is usually an institution such as a prison, hotel, hostel, home for the aged, etc. Also hostels: a collective form of accommodation specifically built during the apartheid era for mine, factory, power station, municipal or other employees.
Every person present in South Africa on Census Night, October 9-10, 1996, should have been enumerated.
Census/enumeration data [cen]
MICRODATA SOURCE: Statistics South Africa
SAMPLE UNIT: Households and individuals
SAMPLE FRACTION: 10%
SAMPLE SIZE (person records): 3,621,164
Face-to-face [f2f]
There were five different questionnaires that were used: 1) A household questionnaire which was completed in each household in the country. This questionnaire included information on each individual in the household, for example age and gender, as well as on the household as a whole, for example access to electricity and tap water. 2) An individual questionnaire, which was completed by individuals living on their own, for example those living in hostels or compounds. 3) A summary questionnaire for hostels. 4) A questionnaire for institutions, for example prisons, tourist hotels or homes for the aged. 5) A questionnaire for the homeless.
UNDERCOUNT: 10.7%
In the first quarter of the year 2021, the capacity of Singapore's data centers increased to *** megawatts. Singapore is one of the leading data center hubs in Asia. During the COVID-19 pandemic, the demand for data storage and capacity had increased.
Energy consumption of artificial intelligence (AI) models in training is considerable, with both GPT-3, the original release of the current iteration of OpenAI's popular ChatGPT, and Gopher consuming well over **********-megawatt hours of energy simply for training. As this is only for the training model it is likely that the energy consumption for the entire usage and lifetime of GPT-3 and other large language models (LLMs) is significantly higher. The largest consumer of energy, GPT-3, consumed roughly the equivalent of *** Germans in 2022. While not a staggering amount, it is a considerable use of energy. Energy savings through AI While it is undoubtedly true that training LLMs takes a considerable amount of energy, the energy savings are also likely to be substantial. Any AI model that improves processes by minute numbers might save hours on shipment, liters of fuel, or dozens of computations. Each one of these uses energy as well and the sum of energy saved through a LLM might vastly outperform its energy cost. A good example is mobile phone operators, of which a ***** expect that AI might reduce power consumption by *** to ******* percent. Considering that much of the world uses mobile phones this would be a considerable energy saver. Emissions are considerable The amount of CO2 emissions from training LLMs is also considerable, with GPT-3 producing nearly *** tonnes of CO2. This again could be radically changed based on the types of energy production creating the emissions. Most data center operators for instance would prefer to have nuclear energy play a key role, a significantly low-emission energy producer.
Renewable energy accounted for 64 percent of Digital Realty's total energy consumption in 2023, up from 56 percent the previous year. Digital Realty is one of the world's largest providers of colocation data center services.
According to a 2024 forecast, global electricity consumption of data centers was projected to grow from 330 terawatt-hours in 2022 to over one petawatt-hour in 2030. This would represent around 3.7 percent of the total electricity consumption worldwide by the end of the period under consideration. Artificial intelligence accounted for around 4.5 percent of the data centers' electricity consumption in 2023. This figure is projected to grow over the next five years.