The Infinite Digital Deluge: Is It True That 90% of the World's Data Was Created in the Last Two Years?

The Anatomy of a Viral Statistic: Where Did the 90% Claim Originate?

Let us look back to May 2013. A marketing report by SINTEF, a Norwegian research organization, dropped a bombshell claim that would echo through tech keynotes for the next thirteen years. They asserted that 90% of all data in human history had been generated over the preceding twenty-four months. At the time, the world was rapidly transitioning to smartphones, Instagram was finding its footing, and corporate cloud migration was shifting into overdrive. But here is the thing: what was a reasonably accurate calculation in 2013 has become an outdated zombie stat in 2026. People don't think about this enough, but quoting a ten-year-old metric to describe today's hyper-accelerated internet is like using a map of colonial America to navigate modern Manhattan traffic.

The Sintef Legacy and the Zettabyte Era

When that study was published, the global data volume was measured in single-digit zettabytes. One zettabyte equals one trillion gigabytes—a number so colossally large that standard human brains struggle to conceptualize it without some sort of physical anchor. In 2013, we were collectively dealing with about 4.4 zettabytes of total information. By the time 2025 wrapped up, International Data Corporation (IDC) analysts estimated the global datasphere had ballooned past 175 zettabytes, a dizzying trajectory driven by high-definition video streaming, IoT sensors, and corporate telemetry. Yet, math dictates that you cannot mathematically sustain a literal "90% every two years" growth rate indefinitely without collapsing the energy grid powering these storage facilities, which explains why the actual rolling two-year percentage has naturally stabilized into a lower, though still terrifyingly steep, curve.

Deconstructing the Math Behind Modern Data Proliferation

How do data scientists actually measure everything from a teenager's TikTok draft in Ohio to a banking transaction in Zurich? It is a complex blend of tracking hard drive shipments, data center utilization rates, and network traffic metrics. The issue remains that a massive portion of what we define as "created data" is actually transient rubbish. We are talking about automated system logs, temporary cache files, and duplicate backups that vanish into the ether mere milliseconds after their creation. Why do we count them? Because from a hardware perspective, that throughput still requires processing power and bandwidth.

The Exploding Metric: Storage vs. Throughput

I find it deeply amusing that we conflate what we save with what we create. If a security camera records an empty warehouse in Chicago for forty-eight hours in 4K resolution and then overwrites that footage automatically, does that count toward the grand total of human knowledge? According to industry frameworks, yes. And that changes everything because it means our unstructured data accumulation is heavily artificially inflated by machine-to-machine chatter. Statistically, the gap between installed storage capacity and total data generated is widening vastly. We create hundreds of zettabytes annually, yet our global physical storage capacity—the actual silicon and magnetic tape available in server farms—is estimated to be under 15 zettabytes globally. We are essentially living in a digital house of cards where the vast majority of our creations are fleeting whispers.

The Hyper-Scale Data Center Monopoly

Where does the permanent stuff actually live? The answer lies within massive, football-field-sized compounds operated by a handful of tech behemoths in places like Virginia, Ireland, and Singapore. These hyper-scale facilities handle the heavy lifting of structured data management, housing the photos you forgot you took in 2018 alongside critical banking records. But the sheer volume of infrastructure needed to keep up with even a 30% annual growth rate is staggering. Think about it: every single minute, users upload over 500 hours of video to YouTube alone. That is not just a statistic; it is an infrastructural nightmare requiring millions of gallons of cooling water and megawatts of electricity every single day.

The True Catalysts of the Current Digital Explosion

If the 90% myth is an oversimplification, what is actually driving the real, verified surge in data creation today? Hint: it is no longer just humans typing on keyboards or snapping selfies on vacation. We have officially entered an era where machines are far more talkative than the biological entities that created them. The explosion is automated, silent, and happening right under our noses.

The Rise of Generative AI and Algorithmic Synthesis

Enter the real disruptor of the mid-2020s: generative artificial intelligence. Since the massive boom of large language models around 2023, synthetic data generation has skyrocketed. Think about a single developer running an AI agent that spins up 10,000 lines of code in seconds, or an automated image generator spitting out millions of high-resolution variations for synthetic training sets. This is not just linear growth—it is an autonomous feedback loop where AI models are creating data to train the next generation of AI models. Where it gets tricky is assessing the quality of this output. We are effectively diluting the global information pool with synthetic noise, which creates a bizarre paradox where data is becoming infinitely abundant but increasingly less valuable.

The Internet of Things and Industrial Telemetry

Aside from AI, your smart toaster and the jet engine of a Boeing 777 are contributing to the pile. Modern industrial manufacturing relies on thousands of minute sensors checking temperature, vibration, and pressure every millisecond. This continuous stream telemetry feeds directly into predictive maintenance algorithms. In short, your local municipal water treatment plant probably generates more raw data packages in a single afternoon than the entire library of Alexandria held during its peak glory years.

Challenging the Definition: What Actually Counts as "Data"?

To truly dismantle the "90% of the world's data was created in the last 2 years" narrative, we have to talk about semantics. If you copy a 5-gigabyte movie file from your laptop to an external hard drive, have you created 5 gigabytes of new data? Industry analysts at firms like Gartner and IDC often count this as a creation event because it generates network traffic and consumes media space. Except that from an intellectual perspective, it is a complete illusion. It is redundant mirroring.

The Redundancy Paradox and Dark Data

The reality is that corporate networks are bloated with what experts call dark data. This is information that is collected, processed, and stored during regular business activities but generally ignored or forgotten. It includes outdated employee accounts, unedited corporate videos, and log files from systems that do not even exist anymore. Honestly, it's unclear exactly how much of the world's 175+ zettabytes is just digital garbage rotting in forgotten cloud buckets. Some studies suggest up to 55% of all stored corporate data is completely useless dark data. We are building massive, energy-devouring monuments to house digital waste, yet we treat the overall growth percentage as a badge of human progress. We are far from it.

Common mistakes and misinterpretations of data velocity

The "90% of the world's data was created in the last 2 years" trap

The fatal flaw in accepting that 90% of the world's data was created in the last 2 years lies in treating all bits as equal. We conflate raw capacity with actual informational utility. Every single hour, thousands of hours of redundant high-definition video are uploaded to cloud servers, alongside millions of automated system logs and duplicate backups. Does this represent a sudden explosion of human knowledge? Absolutely not. Most of this volume constitutes digital exhaust, meaning that while the data footprint itself swells exponentially, the unique insights we generate do not match that trajectory.

Confusing storage with creation

Another frequent blunder is misinterpreting storage infrastructure expansion as proof of creative output. A massive percentage of what we label as new data actually consists of machine-to-machine pings and cached files that vanish within seconds. Automated sensors on industrial machinery can generate terabytes of telemetry hourly, yet the vast majority of these data points are completely identical. Data velocity does not automatically equal value creation. When analysts parrot the line that 90% of the world's data was created in the last 2 years, they mistake the echo for the original voice.

The zombie stat phenomenon

Let's be clear: this specific statistic has been circulating since at least 2013, which makes its mathematical permanence literally impossible unless global storage capacity magically resets every twenty-four months. It has mutated into a zombie statistic, a piece of folklore repeated by marketing departments to sell cloud subscriptions. The issue remains that the phrase has outlived its original context. If it were true a decade ago and remains true today, the compounding math would imply we are producing zettabytes at a rate that defies physical silicon constraints.

The dark data crisis: What the experts know

The hidden burden of digital hoarding

Here is the inconvenient truth that enterprise storage architects rarely discuss openly: we are drowning in digital landfills. Estimates suggest that up to 85% of corporate data is redundant, obsolete, or trivial, a category specialists call ROT data. Organizations compulsively hoard everything because the cost of spinning disk storage dropped significantly over the past decade. Except that storing this junk consumes immense electrical power. We are currently burning megawatts of grid power to preserve useless telemetry and forgotten email attachments from 2018. Which explains why looking only at the sheer volume of global bytes paints a dangerously misleading picture of human progress.

The signal-to-noise ratio collapse

As the global information footprint marches toward an estimated 180 zettabytes, our capacity to extract meaning is actually shrinking. The problem is that our analytical tools cannot keep pace with the sheer chaos of uncurated information pools. We are building larger haystacks without adding any needles. True information density is dropping, meaning that finding actionable intelligence requires far more computational effort than it did when the web was smaller.

Frequently Asked Questions

Is the global volume of digital information actually tracked by anyone accurately?

No single entity possesses a definitive yardstick, but market research firms like International Data Corporation track global storage shipments and enterprise consumption to estimate the size of the Global DataSphere. Their models suggested the world generated roughly 120 zettabytes of information globally in recent years, a number projected to climb past 180 zettabytes. However, these figures are aggregated estimates based on hard drive sales, server deployments, and internet traffic metrics rather than an exact file-by-file count. As a result: the metrics should be viewed as directional indicators of infrastructure capacity rather than precise audits of human creative output.

Why does the "90% of the world's data was created in the last 2 years" claim keep resurfacing?

This happens because the phrase serves as an incredibly potent marketing hook for technology vendors selling artificial intelligence, cloud analytics, and database management systems. It triggers a profound fear of missing out among corporate executives who worry their companies are failing to monetize this supposed gold rush of new information. But how can a statistic remain identically true for over a decade? The reality is that the specific percentage is treated as a rhetorical device rather than a dynamic mathematical equation. It persists because it sounds shocking, making it the perfect opening slide for venture capital pitches and tech conference keynotes.

Does the rise of generative AI mean this data creation rate will accelerate even faster?

Synthetic information generation is currently triggering an unprecedented spike in automated content production that will alter the digital landscape forever. Large language models and AI image generators can output millions of words and graphics in seconds, synthetic content that then gets crawled, indexed, and stored across the web. Because this machine-generated content requires no human latency to produce, it threatens to pollute public data pools with self-referential loops. This shift will undeniably expand total storage consumption, yet it simultaneously threatens to degrade the overall quality of our shared digital ecosystems.

Beyond the metric: A call for digital minimalism

We must stop worshiping the false god of exponential digital volume. The relentless fixation on the idea that 90% of the world's data was created in the last 2 years has blinded organizations to the catastrophic environmental and cognitive costs of digital hoarding. More data is not inherently better; it is frequently just heavier, riskier, and more expensive to secure against modern cyber threats. We have reached a historical tipping point where the ability to delete, filter, and ignore useless information is infinitely more valuable than the capability to accumulate it. True technological mastery belongs to those who curate with precision, not those who hoard the digital exhaust of a hyperactive internet.