Data Created Every
Single Minute in 2025
500 hours of video uploaded to YouTube. 695,000 Instagram stories shared. 41 million WhatsApp messages sent. 5.9 million Google searches. Every 60 seconds — simultaneously — while you read this sentence.
All figures are per-minute averages based on 2025 platform data, annual reports, and third-party analytics. Sources: Domo “Data Never Sleeps 13.0” 2025, DataReportal, Statista.
Data volume estimated from active users × average data per action × activity rate. Indexed — not all platforms generate comparable data types. Source: Domo 2025, Statista, platform reports.
Zettabytes (ZB) of data created, captured, copied and consumed globally. One zettabyte = 1 trillion gigabytes. Source: IDC Global DataSphere 2025, Statista.
Share of global data by type. Most data is never seen by a human — it is machines communicating with machines. Source: IDC DataSphere 2025, Seagate, Statista.
Most Data Is Never Seen by a Human
The data explosion is primarily a machine-to-machine phenomenon — not a social media one.
The popular image of the data economy — tweets, photos, videos — represents a small fraction of total data created. 42% of all data is machine-generated IoT signals from industrial sensors, smart infrastructure, and autonomous systems that never touch a human interface.
Data volume is an industrial story as much as a consumer one.
AI Is Eating the DataSphere
AI both consumes and generates data at a rate that is reshaping the entire infrastructure layer.
- Training: GPT-4 required ~45TB of training data. Future models will require orders of magnitude more.
- Inference: Every AI query generates logs, context windows, embeddings, and feedback loops — creating new data continuously.
- Synthetic data: AI models are increasingly trained on AI-generated data — a recursive loop that could contribute 10–15% of the DataSphere by 2030.
Whoever builds the infrastructure to store, process, and act on this data controls the economy of the next decade.
The Storage Paradox
120 zettabytes created — but only 2% is ever stored and used.
IDC estimates that less than 2% of data created is retained and analyzed. The rest is transient — generated, transmitted, and discarded. The challenge is not creating data. It is identifying which 2% is worth keeping and building the infrastructure to act on it in real time.
The scarcest resource in the data economy is not storage — it is the ability to extract signal from noise at scale.
The DataSphere Will Hit 175 ZB by 2028
IDC projects continued ~25% annual growth driven by AI, IoT, and edge computing.
At current growth rates, the global DataSphere reaches 175 ZB by 2028 and 290 ZB by 2032. The primary accelerants are autonomous vehicles (each generating 1TB per hour), AI inference workloads, and smart city infrastructure rollouts across Asia.
Data Sovereignty Will Become a National Security Issue
Who controls data infrastructure controls economic intelligence.
The EU’s GDPR, China’s Data Security Law, and the US Cloud Act all reflect the same recognition: data location is a geopolitical variable. The next decade will see accelerating fragmentation of the global internet into regional data blocs — each with its own storage, processing, and AI infrastructure.
Energy Is the Hidden Constraint
Data centers already consume 2% of global electricity — AI is pushing that toward 10%.
The IEA projects data center energy consumption will double by 2026 driven by AI workloads. The constraint on the data economy is not bandwidth or storage — it is the physical energy required to process 120 zettabytes of annual data creation.
