Machine Learning Ranks 9 Untested Copper Targets at Fitzroy's Buen Retiro for 5,000-Meter Drilling - Article | Crux Investor

• Fitzroy Minerals and Fleet Space Technologies used 3D underground imaging and machine learning to find 9 possible copper targets at Buen Retiro in Chile. None has been drilled before, and all are within 400 meters of the surface.
• The computer model combined 3D images made using natural ground vibrations with electrical and magnetic measurements, geological models, and about 24,400 copper test results from 140 drill holes.
• The study left out the old open pit where mining had already taken place and areas within 200 meters of existing drill holes. It focused on possible copper deposits hidden beneath layers of gravel.
• Fitzroy has started a 5,000-meter drilling program using diamond drilling and reverse circulation (RC) drilling. The company will use the new drilling results to improve its computer model and choose where to drill next.
• Company leaders believe that much of the property's potential for sulfide copper deposits has not yet been explored. They point to similarities between Buen Retiro and the large iron oxide copper-gold deposits at Candelaria and Mantoverde.

About the Company

Fitzroy Minerals (TSXV: FTZ | OTCQX: FTZFF | FSE: C3Y) is a company that searches for and develops copper projects in Chile. It believes these projects have strong potential to grow in value. Its 2 main properties are the Buen Retiro Copper Project near Copiapó and the Caballos Copper Project in Valparaiso.

9 Possible Copper Targets Found Beneath Gravel

On October 22, 2026, Fitzroy announced the results of a study at Buen Retiro that used underground imaging and machine learning (ML). The study was carried out with Fleet Space Technologies. Machine learning is a way of teaching computers to find patterns in large amounts of information. The computer model identified and ranked 9 possible areas to explore. In each area, the rocks underground appear to have features very similar to rocks where copper has already been found at Buen Retiro. None of these 9 areas has been tested by drilling, and all are within 400 meters of the surface.

Much of the project area is covered by gravel deposited by moving water. This makes it harder for geologists to study the rocks at the surface and choose places to explore using traditional methods. Fleet's 3D model helps Fitzroy follow areas where copper is already known to occur and find new possible targets as deep as about 400 meters. The model also covers areas toward Fitzroy's Sierra Fritis properties to the east, which the company owns completely.

How Underground Imaging and Drilling Data Were Used

Fleet started with a regional ambient noise tomography (ANT) survey that Fitzroy ordered in early 2026. ANT uses natural vibrations in the ground to create a 3D picture of rocks below the surface. The first survey found deep rock structures and areas where rocks had been changed by geological processes or broken into fragments, called alteration and brecciation. Together with a later, more detailed survey, it helped map underground geology much deeper than Fitzroy's existing drill holes.

Fleet then carried out another ANT survey over the main area of interest, using measurement points closer together. This gave the company a clearer picture of faults and cracks in the rocks. Geologists believe these features help control where fluids can move through the ground and where copper deposits may form. The detailed survey also made it easier to compare the ANT images with earlier underground measurements and drilling results.

The machine-learning model then brought together 4 sets of underground measurement data, all in 3D, along with geological information and drilling results. These measurements included ANT data and Fitzroy's induced polarization (IP) chargeability, IP resistivity, and magnetic susceptibility surveys. These methods measure how rocks respond to electrical and magnetic fields. The drilling information included about 24,400 copper test results from 140 drill holes.

How the 9 Targets Were Chosen and Ranked

The computer model studied how underground measurements were connected to the copper levels found in existing drill holes. It then predicted where copper levels might reach or exceed 0.5% in blocks measuring 25 meters across the study area. The model also gave each block an uncertainty value to show how sure it was about the prediction. The 0.5% copper level was used to identify blocks that might meet the target grade. It does not mean every block is expected to contain exactly 0.5% copper.

Fitzroy removed blocks in the old open pit where mining had already taken place, blocks in the mapped gravel cover, and blocks within 200 meters of an existing drill hole. This helped the company focus on places where new exploration could be useful. The remaining blocks were grouped into 9 separate areas. Each area contains blocks with similar computer predictions that fall within a range considered highly reliable. Fitzroy believes these groups may be important from a geological point of view. The company will use the targets to support its own geological studies and drilling plans, not to replace them.

The Possibility of Finding More Copper Sulfide Deposits

The 9 machine-learning targets give Fitzroy a way to explore more of the large iron oxide copper-gold (IOCG) system at Buen Retiro. Until now, most work has focused on oxide copper deposits. The new targets could help the company investigate other kinds of copper deposits, including copper sulfide deposits, across the wider property.

Merlin Marr-Johnson, Fitzroy's President and Chief Executive Officer, explained why the company believes the area is worth exploring and how the new computer model could help:

"We believe Buen Retiro has many geological features in common with the major iron oxide copper-gold (IOCG) deposits at Candelaria and Mantoverde. However, very little work has been done to test whether the wider property contains copper sulfide deposits. Until now, we have mainly focused on oxide copper deposits and whether they could support a mining operation that could start relatively soon without requiring a large amount of money upfront.

We now have a tool to help us test the wider exploration potential of this large IOCG system in one of Chile's most productive copper-mining regions. By combining advanced 3D pictures of the rocks underground, machine learning, and our large collection of drilling results, we can find and rank possible copper targets hidden beneath the gravel. If this machine-learning approach works, it could become a low-cost way to discover new deposits and create value for shareholders."

What the Results Do and Do Not Mean

The 9 areas are possible exploration targets, not confirmed copper deposits. They are places where underground measurements and copper test data show patterns similar to areas at Buen Retiro where copper has already been found. The results do not prove that copper is present or show how much copper might be there, what grade it might have, how large a deposit might be, or whether the mineralized rocks continue across an area.

How reliable the computer model is depends on the quality of the information used to build it and how much of the area that information covers. Its predictions generally become less reliable farther away from existing drill holes and deeper underground. Fitzroy plans to use the computer targets together with traditional geological studies and other exploration information.

Dr. Scott Jobin-Bevans (P.Geo., Ph.D., PMP) checked and approved the technical information. He is an independent geological consultant working with Fitzroy and is recognized as a Qualified Person under the Canadian mining reporting standard NI 43-101.

What Happens Next

Fitzroy has started a drilling program totaling 5,000 meters to test the newly identified areas. The program uses both diamond drilling and reverse circulation (RC) drilling, which are 2 different ways to collect information about rocks underground. Fitzroy and Fleet are reviewing the directions in which the holes should be drilled and the order in which the targets should be tested. As new drilling records and laboratory results become available, the companies will compare them with the original list of targets and update the machine-learning model.

Every new drill hole will test a particular target. Its results will also help check whether the computer model's predictions were correct or need to be changed. The companies will use this information to make better decisions about where to drill next. They aim to make each updated version of the model more accurate so it can guide future drilling at Buen Retiro.