What are the measured energy and carbon footprints of popular messaging apps (WhatsApp, Signal, Telegram)?

1. What the question really asks — and what reporting can answer

The user seeks quantified, comparable footprints for three apps; the available reporting instead provides scattered per‑message or per‑use estimates, aggregate calculations for “online activity,” and model-based numbers from NGOs and consultants — not company-published life‑cycle footprints — so any direct comparison must be framed as approximate and caveat‑laden ^{[1] [2] [4]}.

2. WhatsApp: the most‑reported but still uncertain footprint

WhatsApp appears most frequently in public estimates: Zero Waste Scotland modelled a WhatsApp group chat at about 2.35kg CO2 per week in one analysis cited by The Guardian, a figure that reflects data transfer and server-side storage assumptions rather than a company disclosure ^[1]. Other outlets and commentators treat WhatsApp messaging as “only slightly less carbon intensive than sending an email,” an assessment attributed to researcher Freitag and reported by the BBC, again reflecting modelling choices rather than direct measurement by WhatsApp or its owner Meta ^[3].

3. Signal and Telegram: near‑silence from measurement and reliance on proxies

Independent footprint studies of social apps have generally excluded one‑to‑one messaging apps or treated them as too heterogeneous to rank; Greenspector’s 2023 social‑app study explicitly excluded messaging apps such as WhatsApp, Messenger, iMessage, Skype, Discord and Telegram, leaving Signal and Telegram effectively unmeasured in that programme ^[2]. Reporting and advocacy pieces make claims about Telegram’s cloud architecture or Signal’s minimalist design implying different energy intensities, but these are not backed by published, comparable CO2e figures in the sources provided ^{[5] [6]}.

4. Tiny per‑message numbers, big aggregation effects — and contradictory micro‑estimates

Multiple micro‑estimates show how small single actions look: SMS messages have been variously estimated at fractions of a gram of CO2e per message (figures in reporting include ~0.014g, ~0.14g and even ~0.00215g depending on byte assumptions and methodology), while a tweet is commonly cited near 0.2g CO2e — highlighting how different unit definitions (SMS length, metadata, routing, device energy) produce order‑of‑magnitude changes ^{[3] [7] [8]}. The methodological origin of these per‑message numbers can often be traced to consultants such as Mike Berners‑Lee’s early work on digital footprints, underscoring that numbers are model outputs rather than direct emissions readings ^[4].

5. Why numbers diverge: datacentre power mix, feature set, measurement framing

Discrepancies stem from several clear drivers: whether calculations account only for network transit versus full device energy use, whether datacentre PUE and electricity carbon intensity are modelled or measured, and whether an app’s cloud architecture (end‑to‑end encrypted peer transfers vs. persistent cloud copies) alters server load; Greenspector warns that apps with newsfeeds exchange far more data than messaging apps, complicating apples‑to‑apples comparisons ^{[2] [1]}. Moreover, public studies often rely on “educated guesses” because companies rarely publish app‑level energy or emissions data ^[1].

6. The responsible bottom line for comparison and action

There is no robust, published CO2e per‑year or per‑user metric for Signal or Telegram in the sources provided, and estimates for WhatsApp vary widely depending on modelling choices, so any definitive ranking would be premature; the best-supported conclusions are that single messages have very small emissions but that aggregated usage (group chats, multimedia, high‑frequency use) can add up — and that lack of company transparency remains the central obstacle to firm measurement ^{[3] [1] [2]}. Journalistic and NGO calls therefore converge on recommending system‑level scrutiny (datacentre energy mix, app design for data minimisation) rather than relying on individual micro‑behaviour alone ^{[1] [9]}.