For Owners · Industrial Leadership
Grow, don't Chase.
AIA Engineering | Ahmedabad | May 2026
The ground is moving
Five disclosures from the last 18 months. Each one is a number a CFO can defend in a board meeting.
$6.5M / yr
Newmont · Cadia, NSW
An ML controller on the HPGR feeding the SAG mill lifted throughput 80 t/hr. Cadia is almost certainly an AIA-supplied mine.Q2 2024 earnings call · Cadia Site Tour, Nov 2024
$18.9M
BHP · Escondida concentrator
Azure ML hourly recommendations to operators. 3 GL of water and 118 GWh of energy saved since FY22. BHP opened an Industry AI Hub in Singapore in May 2025.BHP Insights, Aug 2024
~$400M
Freeport-McMoRan · Bagdad
McKinsey ML found Bagdad's ore was seven types, not one. +90,000 t/yr copper. A new concentrator without the $1.5–2B capex.McKinsey case study, 2024
Inside the ball
Moly-Cop · Digital Engine
Wireless sensors inside the SAG ball itself. Three AI acquisitions in 24 months. Edikan Gold: 27% throughput uplift in 8 months.SAG 2023 Vancouver · MILL Trakka acquisition
40 yrs of data
Magotteaux · 2024 Annual Report
"Building on more than 40 years of data, Magotteaux developed machine learning models to predict grinding media alloys performance." Plus a five-year strategic partnership with Codelco.Magotteaux Integrated Annual Report, 2024
62%
Mining AI capital flow
Of mining AI investment now happens through M&A, not internal R&D. Productised AI vendors are being acquired. The window to build the moat is narrowing.Mind the Bridge × BHP, Oct 2025
Strategic encirclement
AIA's public AI footprint: zero. No Chief Digital or Technology Officer disclosed. No AI line in FY26 capex (₹120–130 Cr, of which ~₹30 Cr is solar). MillART is good metallurgy, but it is DEM simulation, not ML.
What we can do about it
Each one is a defensible play. Not a tool purchase. Each is anchored in an asset only AIA has: forty years of mine-by-mine wear data across more than a hundred unique ore and media combinations.
Move 01 · Wear analytics
A predictive wear product on AIA's own data.
XGBoost models published in 2024 predict SAG liner wear at ~5% MAPE. The residual gap is alloy and ball-property data. AIA owns precisely that, mine by mine. Ship a wear-prediction service before Magotteaux scales theirs.
Moat40 years of proprietary mine-by-mine performance records.
Move 02 · Generative R&D
An AI pipeline for ore-specific alloy design.
Microsoft's MatterGen (open-sourced Jan 2025) and DeepMind's GNoME design novel inorganic crystals from target properties. Pair with MAGMASOFT and CALPHAD to produce ore-specific high-chrome candidates competitors cannot replicate without your performance data.
MoatOutcomes data from a hundred-plus operating mills.
Move 03 · Foundry CV
Computer vision on the casting line.
YOLO defect detection and 3D anomaly models on CT scans are off-the-shelf. The lowest-execution-risk pilot AIA can run, inside its own foundries, with measurable yield uplift in one quarter. Builds internal AI muscle before any customer-facing work.
MoatYour physical lines, your defect labels, your cost data.
Move 04 · Customer engagement
A customer portal that predicts wear and quotes itself.
Caterpillar's Cat Helios runs on 16 PB across 1.6M connected assets. AI-handled B2B interactions cost $1–2 vs $15–25 rep-assisted. Convert AIA field service from a cost centre into a continuous data-collection engine.
MoatDirect mill-floor relationships your OEMs do not have.
For Owners
shalin@masalagames.com