Moore’s Law, AI, and
the Impact on eDiscovery

Anyone who has been around technology long enough has bumped into “Moore’s law” at one point or another. It’s one of those aphorisms and bits of wisdom people in the tech industry bring up from time to time, usually coincidental to a big, giant leap forward in advanced technology.

For starters, it’s not actually a “law,” it’s an observation made by former Intel CEO Gordon Moore way back in 1965 that the number of components per integrated circuit has been doubling every year. He later amended that to every two years, but in practice the average of that doubling had approximated about 18 months, until about 2010 when it slowed and has now effectively been closer to every 3 years. In 2022, NVIDIA CEO Jensen Huang declared Moore’s Law to be dead, yet the pace has continued (ironically though NVIDIA innovations that have increased capacity). The main point though is that the quality and capacity of individual microprocessors are still increasing exponentially in very little time.

Now, to most people, this little observation may mean next to nothing; why is this important at all? Because the greater the advancement in computer processing, the more data can be stored, the better the quality and fidelity of that material maintained, the faster the ability to process the data, the longer data can be kept and the larger the capacity of what technology will be capable of. There is also the added bonus that as all this improves, the costs come down. We can see the consequences of these advancements just in how eDiscovery has evolved over the years.

As these exponential gains compound, they reshape not just consumer technology but every field that depends on digital information—and few areas feel the impact more acutely than e-Discovery. The legal profession, which must find, preserve, review, and interpret increasingly vast volumes of electronic data, sits squarely at the intersection of Moore’s observation and its real-world consequences.

When applied to e-Discovery, Moore’s law creates both upward and downward pressures that shape how legal teams manage data. The growth in processing power drives efficiency and reduces costs, but it also accelerates the creation, storage, and retention of information—dramatically increasing project scope.

Over the past 25 years, the volume of electronically stored information (ESI) has exploded. Thisexpansion stems from several factors:

1) Proliferation of Data Types

Early e-Discovery dealt with a narrow universe: scanned documents, word-processing files, spreadsheets, and email. Today, virtually everything organizations touch is digital and discoverable. Modern data sources include:

• Cloud-based SaaS platforms
• Collaboration tools (Teams, Slack, Google Workspace, etc.)
• Smartphones and mobile apps
• High-resolution photos, videos, and screen captures
• Voice memos and audio messages
• Structured databases and enterprise systems

Text files are comparatively small, but high-resolution images, embedded media, video messaging, and rich-content business files dramatically inflate dataset size. A PowerPoint from 2010 might have contained a few text slides; a 2024 deck often includes dozens of images, GIFs, embedded video, layered graphics, and animations—each significantly increasing file weight.

This also is impacted by the size of individual files themselves. For instance, the average smartphone photo increased from ~500 KB in 2008 to ~3–6 MB today due to higher sensor resolution and richer formats.

2) Changes in Storage Behavior

In the past, limited storage and concerns about system performance encouraged users to delete outdated or draft material. Today, storage is abundant and inexpensive, so the default behavior has reversed: people keep everything. The more capacity we have, the less we delete—and the more data must then be collected, processed, reviewed, and analyzed during discovery.

As a result:

• Multiple document versions are preserved as part of the drafting process
• The same file may exist in numerous locations (local drives, email folders, shared drives, cloud repositories, and colleagues’ inboxes)
• Emails accumulate rather than being regularly purged
• File shares and cloud spaces expand without meaningful pruning

3) Data Collection Practices

Early collections of data usually consisted of “drag and drop” collections where practitioners would go to the computer to be collected, hook up an external hard drive, and through various applications collect the necessary data. What was collected as far as potential relevance was left to the discretion of the person doing the collection. With increasing judicial standards, forensic collections and imaging, and evolving data best practices, it became standard practice to collect and process everything, and sort through it as part of an initial process called “Early Case Assessment.” It’s there where screens and deduplication would be applied to reduce subsequent review volumes.

These volume increases directly affect processing and hosting costs. Data volumes that would have been extraordinary twenty years ago (e.g., 500 GB) are now commonplace for an individual custodian across phones, laptops, messaging platforms, and cloud accounts. Growing quantities, richer file formats, higher resolution media, and retention practices collectively contribute to expanding discovery budgets.

The overall result of our increased data volumes are that the average custodial collections have grown from 5–10 GB in the early 2000s to 200–500+ GB in many modern matters (varies widely by industry). It is not uncommon today for large-scale document review projects collecting overall data volumes in the double-digit TB volume. Coincidental to the increased data volume and our increased efficiencies in dealing with them, the average cost of storage has dropped from ~$10/GB in 2000 to ~$0.02/GB today (varies by provider).

One of the most striking illustrations of computational progress is the comparison between the Apollo Guidance Computer and a modern smartphone. The Apollo computer had:

• A 15-bit word length
• Just 2,048 words of RAM
• 36,864 words of read-only core memory
• A weight of ~70 pounds

A modern smartphone contains over a million times the memory and operates at speeds that make Apollo-era computation appear almost static. The Apollo computer weighed 70 lbs. and was the size of cabinet drawer; a smartphone fits in my pocket and weighs about a pound.

This exponential progress—predicted by Moore’s law—has transformed the business world, and points to what e-Discovery technology can achieve. So, as Moore’s law increases the storage capacity on a microchip, it also shrinks the size required, speeds up the processing, increases the quality of results, and allows for greater innovation by compounding technological benefits and greatly increasing the scope of what computing can do. This all helps to increase work efficiency.

You can see the capabilities of this functionality by just looking at the features used to reduce data volumes over the past 25 years. First, there were simple and then complex search terms; initially a search across one million documents would take some time, but as the tech advanced, the speed improved, and the time required to run decreased. Then came more complex features, such as deduplication, clustering, concept searching and other assorted applications. These take more computing power than simple searching does, but as integrated circuits became more capable, it became much more common, faster and prolific.

Soon after came machine learning, a more complex system analyzing text within individual documents and comparing to exemplar documents for relevance. Commonly referenced as “TAR” (technology assisted review) early applications were again, slow, costly and required a lot of computing power and server resources. As technology grew, the quality of TAR engines did as well, costs came down and reviews sped up. It should come as no surprise that the time that saw the biggest increase in TAR use was in 2015, with legal standards firmly established; this also coincided with the largest decrease in computational costs associated with TAR use across many industries.

All of this is to say that as our data volumes have increased, so has our capacity to utilize technology to decrease the size via functionality. Machine learning at scale for a document review requires a substantial amount of memory and computing capacity, but it is tiny considering the next generation of A.I. powered enhancements and applications. Large Language Models (LLMs) that power GenAI require an exceptional amount of data storage and speed to function adequately.

But from what we have seen within just the last two years conforms to what we have seen throughout the decades since Gordon Moore first postulated his eponymous law; the capacity increases, the speed and quality continue to go up, and the price comes down substantially. The accelerating pace of AI development mirrors Gordon Moore’s observation almost perfectly. IDC estimates that global data creation will reach ~180 zettabytes by 2025, up from 2 zettabytes in 2010.

Moore’s law may be slowing, but the growth of data is not—and the future of e-Discovery will be defined by how effectively the legal industry adapts to new waves of computational innovation. The next era will be shaped not only by exponential hardware improvements, but by the unprecedented influence of artificial intelligence, automation, and eventually quantum computing.

AI is already reshaping information governance by transforming how organizations classify, retain, and secure their data. Instead of relying on manual policies or user-driven decisions, AI systems can increasingly analyze content in real time, identify sensitive material, flag risky behavior, and apply retention schedules automatically. This shift will fundamentally redefine what “governance” means: proactive, intelligent data hygiene rather than reactive, manual cleanup. As organizations implement smarter IG, the data entering the e-Discovery pipeline will be more structured, context-rich, and easier to cull early—dramatically improving downstream efficiency.

Collections will evolve as well. Rather than sweeping up every data source for later sorting, AI-driven systems will allow targeted, context-aware, and continuously updated collections that understand relevance at the source. Algorithms will assess communication patterns, interpret semantic meaning, and identify the most likely responsive materials before a review team ever touches a document. Collection will shift from a static event to a dynamic, iterative process powered by real-time analytics.

On the review side, AI and large language models will transform how documents are analyzed, summarized, categorized, and compared. Instead of relying solely on keyword searches or manual review teams, next-generation systems will interpret content the way a human does—reading across documents, understanding nuance, identifying issues, and generating first-level work product automatically. Lawyers will increasingly supervise machines rather than sift through raw documents themselves, allowing them to focus on strategy, interpretation, and advocacy rather than classification.

Looking further ahead, quantum computing promises another leap. While still emerging, quantum systems could eventually analyze massive datasets instantaneously, enable extremely precise pattern recognition across billions of documents, and perform calculations impossible for classical machines. In the legal domain, quantum tools could make near-instant privilege detection, perfect deduplication, and ultra-fast contextual analysis a reality. They may also reshape cybersecurity, as quantum encryption and quantum-safe protocols become essential components of protecting client data.

All of these innovations point to a legal landscape in which the core challenges of e-Discovery—volume, variety, velocity, and veracity—are met with tools that match or exceed the pace of technological change. The firms and organizations that thrive will be those that embrace this transformation, invest in modern information governance, and collaborate closely with technologists to build workflows designed for the data realities of the next decade.

Gordon Moore’s observation was never just about transistors—it was about the compounding impact of technological progress. In e-Discovery, that progress continues to redefine how we find the truth in an expanding universe of information. The future belongs to those prepared to harness it.