From User Behavior to Energy Grids and the Economics of Online Carbon

Across continents and income levels, daily life now unfolds inside systems that appear immaterial yet depend on industrial-scale infrastructure. Roughly 5.35 billion people – 66 percent of the global population – are connected to the internet, spending an average of 6 hours and 38 minutes online each day. Over a year, that equates to more than 100 days spent inside digital environments. Among individuals aged 16–24, usage frequently exceeds seven hours daily. Social media alone absorbs approximately 2 hours and 20 minutes per day, while video streaming accounts for more than 60 percent of total global internet traffic. Connectivity has moved from accessory to default.

What makes this transformation difficult to perceive is its invisibility. A streamed briefing before work, a map recalculating traffic conditions, shared documents updating in real time, an AI tool drafting a summary, groceries ordered through an app, and an evening of high-definition streaming all activate distant data centers linked through fiber networks and cooled by industrial systems. Every search query and uploaded file draws electricity from grids whose carbon intensity varies dramatically by region. And with each passing year, usage expands as businesses digitize operations, households adopt additional connected devices, and emerging markets integrate into global digital ecosystems.

Global Digital Usage and Associated Carbon Indicators

Source: DataReportal Digital 2025; Anthropocene Magazine (2024); Sandvine Global Internet Phenomena Report.

Measured in energy terms, the scale is significant. Global data center electricity consumption already exceeds 400 terawatt-hours annually and is projected to approach 945 terawatt-hours by 2030. An individual spending roughly 3,230 hours online each year is associated with approximately 229 kilograms of CO₂ emissions when device manufacturing, network transmission, and server operations are included. At the sector level, digital technologies account for an estimated 2–4 percent of global greenhouse gas emissions. The marginal energy draw of a single streamed episode may be small, yet repetition across billions of users converts incremental behavior into persistent demand.

The economic architecture of digital platforms helps explain this trajectory. Revenue models built on advertising, subscriptions, and marketplace commissions reward longer engagement and higher interaction frequency. Features such as infinite scroll, autoplay video, personalized feeds, and push notifications extend sessions by design. Once broadband access is secured, additional usage carries little visible cost to the user, lowering the behavioral barrier to extended consumption. Convenience becomes default, and default becomes norm.

Layered onto this behavioral logic is device proliferation. Global smartphone subscriptions exceed 6.8 billion, and tens of billions of connected devices are now active worldwide. In many high-income economies, individuals move seamlessly between smartphones, laptops, tablets, and smart televisions within a single day. Manufacturing emissions can represent more than half of a smartphone’s total lifetime carbon footprint, and replacement cycles typically span just two to three years. Digital participation is therefore inseparable from recurring material production.

Geography ultimately mediates the carbon outcome. In coal-intensive systems, electricity generation can exceed 800 grams of CO₂ per kilowatt-hour; in renewable-heavy grids, it may fall below 100 grams. The same hour of streaming carries different implications depending on location. Online carbon is thus the cumulative expression of demographic growth, engagement-driven design, device density, enterprise digitization, and the energy mix beneath the grid.

Scale Indicators Underpinning ICT Carbon Trajectories (Units), 2015–2030

Source: ITU; ITU-T L.1470 (01/2020), Table IV.1–IV.2 activity assumptions and traffic ranges for 2015/2020/2025/2030.

From Routine to Regional Load

As digital behavior scales, it becomes visible in places far removed from the screen – in energy forecasts, corporate capital expenditure, and grid interconnection queues. The more than 400 terawatt-hours already consumed annually by data centers, and the projected 945 terawatt-hours by 2030, are not abstract figures. They represent the cumulative result of normalized habits repeated billions of times.

A single day illustrates the shift. Morning news streams to a phone while navigation software recalculates a commute. Video meetings replace conference rooms. Shared cloud documents update continuously. AI tools generate summaries. Payments clear digitally. By evening, multiple screens stream high-definition content in parallel. None of these actions resembles heavy industry, yet together they require uninterrupted server capacity and cooling systems running around the clock.

Video traffic provides the clearest example of how preference becomes load. With more than 60 percent of global internet traffic attributed to video, the transition from standard definition to high-definition and 4K formats has increased data intensity per viewing hour. Households now stream across multiple devices simultaneously, creating demand peaks unlike the centralized broadcast model of previous decades. Gaming platforms maintain persistent server connections, and social media feeds refresh automatically. Engagement design extends session length; extended sessions increase throughput.

Every element of digital life carries a measurable carbon imprint, and in aggregate, those increments accumulate.

Structural Drivers Expanding Online Carbon

Source: DataReportal (2025); IEA (2025); U.S. DOE (2024); IoT Analytics (2024); Global retail projections (2025).

Work patterns reinforce this baseline. A 2025 global study estimates that college-educated workers perform roughly 25 percent of workdays from home on average, embedding cloud collaboration and video conferencing into the structure of employment. Microsoft reports more than 320 million monthly active Teams users, underscoring the normalization of real-time digital coordination. Hybrid work is no longer transitional; it is infrastructural.

Commerce follows a similar trajectory. Netflix reported approximately 270 million subscribers in early 2024, reflecting the scale of streaming demand. Global e-commerce is projected to exceed 20 percent of total retail sales in 2025, integrating recommendation engines, predictive analytics, and cloud-based logistics into daily transactions. Retailers such as Amazon rely on real-time data flows to coordinate supply chains. Financial institutions process billions of digital transactions annually through secure cloud infrastructure. Expectations of immediacy and personalization expand computing requirements.

Artificial intelligence adds computational density to this system. Tasks that once required simple database retrieval now rely on model inference running on specialized processors. AI-generated responses, automated fraud detection, logistics optimization, and personalization algorithms increase compute per interaction. The user interface appears simplified; the infrastructure becomes more energy intensive.

Even when screens are idle, digital demand persists. IoT device counts reached approximately 18.5 billion in 2024 and are projected to exceed 21 billion in 2025. Smart thermostats adjust temperatures, vehicles receive over-the-air updates, wearable devices transmit health metrics, and industrial sensors report performance continuously. Background connectivity sustains data traffic independent of conscious user engagement.

Through this sequence – habit, enterprise architecture, infrastructure expansion – everyday behavior translates into regional electricity demand. Consumer expectation influences corporate strategy; corporate strategy influences grid planning. The footprint of online activity is no longer confined to data centers. It is embedded within national energy systems.

Governing Growth in a Compute-Driven Economy

As digital infrastructure approaches a projected 945 terawatt-hours of annual electricity demand by 2030, online activity has become a measurable force in national energy systems. In the United States alone, data centers consumed approximately 4.4 percent of total electricity in 2023 – rising from 58 terawatt-hours in 2014 to 176 terawatt-hours in 2023 – and could reach between 6.7 percent and 12 percent by 2028. At that scale, digital expansion shapes transmission planning, generation investment, and regional load forecasts.

Regulatory oversight has begun to formalize this shift. Within the European Union, large data centers are required to report standardized sustainability metrics under the Energy Efficiency Directive, including energy performance and operational transparency. Reporting enables benchmarking, and benchmarking introduces accountability. Non-compliance can result in administrative penalties, reputational risk, and increased regulatory scrutiny. Compliance itself carries cost: expanded reporting systems, efficiency upgrades, and performance monitoring increase capital expenditure and may raise barriers to entry for smaller operators.

Carbon accounting frameworks further influence corporate behavior. Under the GHG Protocol’s Scope 2 guidance, firms distinguish between location-based emissions and market-based emissions tied to renewable procurement contracts. This distinction has encouraged major technology firms to secure multi-gigawatt renewable power purchase agreements, accelerating wind and solar deployment in several regions. Firms without comparable capital flexibility remain exposed to grid-average emissions and potential carbon pricing risk.

Utilities and grid operators are responding in parallel. In regions hosting hyperscale clusters, digital load has become central to interconnection queues and transmission upgrades. Some jurisdictions have debated specialized tariff structures or demand charges for large facilities to reflect infrastructure strain. Where carbon pricing regimes apply, electricity-intensive operations may face indirect cost exposure tied to emissions allowances.

Artificial intelligence intensifies the planning horizon. AI workloads require high-density computing and stable power quality, encouraging co-location near renewable generation or negotiated supply agreements. Efficiency improvements in server architecture and cooling systems partially offset rising compute intensity, yet clustering can strain local grids and trigger permitting delays.

Governments have also begun addressing behavioral drivers. In parts of Europe, digital services regulation has expanded scrutiny of recommender systems and engagement-based design. While framed primarily around consumer protection and competition, such measures acknowledge that engagement is engineered. Ecodesign standards and right-to-repair policies aim to extend device lifespans, reducing embodied emissions associated with frequent upgrades. Public digital literacy initiatives increasingly include environmental awareness components, linking consumer behavior to energy impact.

With IoT device counts projected to exceed 21 billion and indirect emissions among major technology firms rising significantly between 2020 and 2023 as data center demand expanded, the underlying inputs remain aligned: engagement intensity, device proliferation, AI integration, and demographic momentum. Governance responses – reporting mandates, renewable procurement, efficiency standards, pricing mechanisms, and behavioral regulation – form a layered framework rather than a single solution.

Whether online carbon stabilizes will depend on the relative pace of grid decarbonization and digital demand growth. Digital culture and terrestrial energy infrastructure now operate within the same physical limits. The economics of engagement and the physics of electricity converge at scale.

Key Takeaways

• The average internet user spends over 3,200 hours online annually, associated with approximately 229 kilograms of CO₂ emissions per year.
• Global data center electricity consumption exceeds 400 terawatt-hours and could approach 945 terawatt-hours by 2030.
• Digital technologies account for an estimated 2–4 percent of global greenhouse gas emissions.
• Video traffic represents more than 60 percent of global internet traffic, significantly increasing data intensity per user.
• Remote and hybrid work models embed high-bandwidth digital services into labor markets, with college-educated workers averaging roughly 25 percent of workdays from home.
• IoT device counts reached approximately 18.5 billion in 2024 and are projected to exceed 21 billion in 2025, sustaining background connectivity.
• In the United States, data centers consumed about 4.4 percent of national electricity in 2023 and could reach up to 12 percent by 2028.
• Governance responses include mandatory sustainability reporting, renewable procurement contracts, carbon accounting frameworks, and ecodesign standards.
• Behavioral economics – engagement-driven design and attention-based revenue models – plays a central role in driving compute demand.
• The future trajectory of online carbon depends on whether grid decarbonization keeps pace with digital demand growth.

Sources

•  International Telecommunication Union (ITU); Recommendation ITU-T L.1470 – Greenhouse gas emissions trajectories for the ICT sector compatible with the UNFCCC Paris Agreement; – Link
• International Energy Agency (IEA); Energy and AI; – Link
• U.S. Department of Energy; DOE Releases New Report Evaluating Increase in Electricity Demand from Data Centers; – Link
• Lawrence Berkeley National Laboratory; United States Data Center Energy Usage Report; – Link
• DataReportal; Digital 2025 Global Overview Report; – Link
• Sandvine; Global Internet Phenomena Report; – Link
• IoT Analytics; State of IoT 2024 – Number of Connected IoT Devices Growing 13% to 18.8 Billion; – Link
• Greenhouse Gas Protocol; Scope 2 Guidance – An amendment to the GHG Protocol Corporate Standard; – Link
• European Commission; Energy Efficiency Directive (Recast) – Data Centre Sustainability Reporting; – Link
• Anthropocene Magazine; The average internet user spends 3,230 hours online every year. Here’s the carbon footprint of that; – Link
• Netflix; Q1 2024 Earnings Report; – Link
• Microsoft; Microsoft FY24 Earnings and Product Usage Disclosures (including Microsoft Teams); – Link