When Mount Vesuvius erupted in A.D. 79, it buried a collection of papyrus scrolls in the Roman town of Herculaneum.
The heat carbonized the scrolls, leaving them so fragile that physically opening them could destroy the text.
For centuries, much of the library remained effectively sealed.
Then, in 2026, the Vesuvius Challenge announced that PHerc. 1667 had become the first Herculaneum scroll to be virtually unwrapped and read from end to end.
The scroll itself still hasn’t been unrolled…
Instead, researchers combined high-resolution X-ray imaging, virtual unwrapping techniques, and machine learning to reconstruct its surface and detect the ink hidden inside.
The ability to read an ancient manuscript isn’t exactly market-moving news…
But it provides an unusually clear demonstration of a much larger economic shift:
Better computing can turn previously inaccessible information into a valuable asset.
For investors, that creates a second wave of the AI trade—one built around the companies whose existing data and physical assets become more valuable as our ability to analyze them improves.
AI is changing the economics of discovery
Finding a major oilfield or mineral deposit has traditionally required an enormous amount of money, time, and uncertainty.
A company surveys a region, collects geological samples, processes the results, and drills exploratory holes that may produce nothing.
Mining projects can take well over a decade to move from discovery to production. S&P Global estimates the average development timeline for a new mine is about 17 years.
Oil exploration carries a similar challenge. Companies can spend years studying a basin before committing hundreds of millions of dollars to drilling a single deepwater well.
Advanced computing changes the equation by helping companies identify the strongest targets before committing enormous amounts of capital.
Think of an old medical scan: New technology won’t change the underlying condition, but it may reveal something previously invisible.
The same principle applies to exploration data.
Modern computing systems can compare enormous volumes of geological information, identify relationships that are difficult for humans to recognize, and repeatedly reinterpret the same dataset as analytical methods improve.
As a result of these improved tools, a seismic survey, geological map, or drilling record can become more useful years after it was originally collected.
That turns data from a one-time expense into an asset that can appreciate as computing power advances.
And we’re already seeing what that can unlock.
How old exploration records helped find a major copper deposit
KoBold Metals, backed by Bill Gates and Jeff Bezos, was built around making mineral exploration more scientific and data-driven. More specifically, the company uses computational models to analyze geological maps, surveys, drilling results, and other exploration records.
The company applied this approach to Zambia’s Copperbelt… which helped it zero in on the Mingomba project—an area that previous generations of mining companies had already studied.
The company is now advancing plans for a mine, expected to cost more than $2 billion and eventually produce around 300,000 metric tons of copper per year.
That’s a big deal.
S&P Global projects global copper demand will rise from roughly 28 million metric tons in 2025 to about 42 million by 2040, driven by electrification, grid expansion, AI infrastructure, and defense. Without significantly greater investment and production, it estimates supply could fall roughly 10 million metric tons short of demand by 2040.
The world needs more copper discoveries… And that means it needs a more efficient way to find them.
KoBold offers an early example of how data and computing could help close that gap.
Big Oil has been investing in this advantage for years
TotalEnergies has been using its Pangea supercomputers to improve oil and gas exploration since 2013.
Each new generation has given the company more power to process seismic surveys, produce clearer images of underground geology, and make faster decisions about where to invest—or where to avoid drilling.
By 2019, Pangea III had reached 31.7 petaflops of computing power—the equivalent of roughly 170,000 laptops. TotalEnergies said the system could reduce geological risk, shorten project studies, and improve field operations at a lower cost.
The company has since added Pangea 4, a more energy-efficient hybrid system that combines an on-site computer with additional capacity in the cloud. And in 2026, it announced plans for Pangea 5 (expected to enter service in 2027), which will increase its computing power sixfold and further expand its use of advanced seismic processing and AI.
Exxon Mobil is doing something similar.
Its Discovery 6 supercomputer is designed to process advanced seismic imaging significantly faster than its previous system. Exxon says the combination of high-performance computing and new imaging technology could reduce certain seismic-processing jobs from months to weeks, improve well placement, and increase resource recovery using less capital.
These companies are spending billions to find and produce energy. Even a small improvement in where they drill can have an enormous financial impact.
That’s why computing has become such an important competitive advantage: Better interpretation of existing data leads to better billion-dollar decisions.
When better data meets a frontier opportunity
The next stage of the AI boom may favor companies that own more than physical assets.
It may favor the companies that also control the information needed to understand those assets.
BluEnergies (BLUGF) offers a timely example.
The company is working alongside TotalEnergies to evaluate potential drilling targets across three blocks in the Harper Basin, offshore Liberia.
A central part of the project involves reprocessing 6,167 square kilometers of 3D seismic information originally collected by TGS in 2013.
That survey provides a three-dimensional picture of the rock formations beneath the seafloor.
Now, more than a decade after the information was collected, the partnership is running it through newer processing methods to produce clearer images and look for stronger signs that the geological structures could contain oil.
The companies are also gathering information from the seafloor to help confirm what they see in the seismic images.
In plain English, they’re combining an older underground map with newer analytical tools and fresh field evidence to decide which targets—if any—justify the enormous cost of drilling.
As of BluEnergies’ July 6 update, TGS had completed more than half of the seismic reprocessing program. The company said the work remains on schedule and is intended to improve the definition of potential drilling prospects.
It’s important to be precise about what that means: BluEnergies has not discovered a producing oilfield. The project remains speculative, and drilling will ultimately determine whether the geological interpretation is correct.
But the potential advantage is clear: Better processing can help the partnership identify which targets appear strongest before it commits to drilling. That gives BluEnergies exposure to a potentially enormous discovery without starting an exploration program on its own from scratch.
The opportunity Wall Street may be underestimating
AI-powered discovery could spread across dozens of sectors:
- Mining companies can compare geological records and target drilling more efficiently.
- Energy companies can reprocess old seismic libraries and improve subsurface models.
- Drug developers can screen enormous collections of compounds before moving into expensive laboratory and clinical work.
- Agricultural companies can analyze decades of weather, soil, and crop information.
- Materials scientists can simulate combinations that would be too expensive or time-consuming to test individually.
As computing improves, that information may reveal more than when it was first created.
This changes how investors should think about data: The most valuable database may not be the newest or largest. It may be the one connected to a scarce physical asset where one additional insight could create enormous economic value.
The obvious AI trade has been the infrastructure layer: chipmakers, cloud companies, and data center operators. And those businesses remain essential…
But the next wave of opportunity could be identifying which industries—and which overlooked assets—become more valuable because of it.
On Wall Street Unplugged, Frank regularly breaks down the overlooked industries, assets, and companies positioned to benefit from the next stage of the AI boom. Subscribe for free so you never miss an episode.
MARKETING DISCLOSURE: BluEnergies Ltd pays Curzio Research Inc for marketing services. Access the full disclosure here.
