Key Takeaways
- 01 Capital discipline is reshaping commodity supply curves into 2026.
- 02 AI infrastructure capex is becoming a primary macro variable.
- 03 Central bank policy across AU, CA and US is converging on neutral.
The smart mining evolution taking shape at Barrick begins with a familiar operating problem: decisions made in one department affect work elsewhere. A maintenance delay can disrupt processing. A missing spare part can extend that delay. Changes in ore supply can then leave the original production plan out of step with conditions.
Barrick wants those connections to become easier to understand and manage.
Its North American business has selected Avathon as a strategic technology partner, with plans to bring operational data and mining knowledge together through artificial intelligence.
The announcement outlines a broad ambition. It does not report completed deployment across every asset or measured improvements in production, safety or costs. Those results will need to follow implementation.
Figure 1: Barrick’s Cortez gold complex in Nevada. Credit: Barrick Gold, via The Northern Miner.
What Is Barrick Trying to Change Across Its Operations?
The partnership aims to connect work across the mining value chain, from identifying geological opportunities to keeping equipment running and materials available.
That is a wider brief than introducing an AI tool for one task. Exploration, planning, safety, processing and procurement are all included in the proposed operating model.
Avathon chief executive Pervinder Johar describes the opportunity as bringing separate functions and systems into a connected operating system. Barrick’s ambition is to give employees a better view of how their decisions affect the wider business.
The official partnership announcement sets out the planned scope and confirms that human expertise remains central to the approach.
For example, a maintenance recommendation becomes more useful when planners can also see the likely production interruption and whether replacement parts are available. That illustrates the intended connection between functions; it is not a reported Barrick case study.
The Barrick gold assets AI programme is therefore about improving the information behind operating decisions. Whether that produces lasting gains will depend on how well the platform works within everyday mine routines.
How Would an AI Operating System Work in a Mine?
Avathon’s platform is designed to bring information from different operational systems into a shared structure.
At its centre is a Computational Knowledge Graph. In plain terms, this maps relationships between equipment, processes, people, operating limits and data. AI agents then use those relationships to support decisions and coordinate work.
The proposed AI operating system mining approach has three main elements:
- Connected information: Data from different functions is brought together so that related issues can be considered alongside one another.
- Continuous analysis: The platform looks for changing conditions, possible failures and opportunities to improve performance.
- Coordinated action: Recommendations can inform planning, maintenance and other workflows, with people retaining judgement and accountability.
The practical difference lies in context. An equipment warning on its own says something may need attention. A warning connected to maintenance schedules, parts inventory and production plans gives a team more information about what to do next.
That does not remove the need to check recommendations against conditions on site. The announcement specifically retains human operational control.
It also leaves implementation details open. It does not disclose a site-by-site timetable, integration costs or the sequence in which applications will become operational.
Where Could the First Applications Make a Difference?
The initial applications cover six areas, each with a recognisable operating purpose.
Safety monitoring is intended to identify hazardous conditions and unsafe behaviour earlier. Production applications would connect ore flow, processing choices and maintenance constraints. Reliability tools would monitor equipment condition and recommend maintenance actions.
The proposed work falls into three connected groups:
- Safety and reliability: Computer vision could flag hazards, while equipment monitoring could help teams anticipate failures and arrange repairs.
- Production and planning: AI would support scheduling and processing decisions using operating data and constraints, with the aim of improving throughput and recovery.
- Supply chains and exploration: The platform would connect material requirements with maintenance needs and apply machine learning to geological and production information.
| Application | Intended benefit |
| Safety monitoring | Earlier identification of hazardous conditions |
| Production and recovery | More consistent processing performance |
| Asset reliability | Fewer unplanned equipment interruptions |
| Supply chain intelligence | Better material readiness |
| Mine planning | Plans that reflect operating conditions |
| Exploration | Better-informed targeting decisions |
These are expected uses, rather than results already demonstrated under the partnership.
Recovery and throughput also mean different things. Throughput measures how much material is processed; recovery concerns how much valuable metal is extracted. Raising one does not automatically improve the other.
That makes coordination important. A useful recommendation must account for the operating conditions and limits of the whole process, rather than favouring a single measure.
Figure 2: Underground work at Goldrush, Nevada. Barrick says mining professionals will retain operational judgement, accountability and control under the proposed AI approach. Credit: Barrick/Nevada Gold Mines.
Who Remains Responsible When AI Recommends an Action?
Barrick’s mining professionals will retain responsibility for operational decisions.
The partnership combines AI analysis with human expertise. The platform is intended to provide predictions, reasoning and decision support, while employees apply their knowledge of the mine and its conditions.
That boundary is particularly important in safety-related work. Identifying a possible hazard is useful only if the information reaches the right people and leads to an appropriate response.
Barrick president and chief executive Mark Hill said the ambition is to connect the value chain while helping people make better decisions more safely.
The announcement also refers to internal governance and safeguards for responsible development and deployment. It does not provide a detailed description of those controls.
For context, Barrick’s Nevada Gold Mines overview describes a network of mines and processing facilities. Such a setting helps explain the appeal of coordinating information across operations, although the partnership announcement does not specify an installation schedule for individual facilities.
The key question is how effectively employees can understand, check and use the system’s recommendations during normal work.
What Would Show That the Partnership Is Delivering?
The next useful disclosures would move from intended applications to evidence from operating sites.
A platform selection establishes direction. It does not establish the size of the eventual benefit.
Three types of evidence would help readers assess progress:
- Deployment details: Which applications are operating, where they are being used and how they fit existing work processes.
- Measured outcomes: Changes in equipment availability, unplanned downtime, recovery or other clearly defined operating measures.
- Human oversight: How recommendations are reviewed, errors are addressed and employees are trained to use the tools.
Comparisons will need context. A production improvement may also reflect different ore, maintenance work or equipment changes. Credible reporting should explain what contribution the AI system made.
For Mining Herald readers, this smart mining evolution is worth following because it addresses connections that shape daily mine performance. The strongest evidence will be whether teams can act on better information and demonstrate repeatable results.
Also Read: Why Critical Minerals Are Key to the Global Energy Transition
FAQs
- What has Barrick selected?
Its North American business has selected Avathon’s Autonomy Platform as a strategic technology partner. - Will AI control operations independently?
Mining professionals will retain operational judgement, accountability and control. - Which activities could use the platform?
Expected applications include safety, maintenance, processing, planning, supply chains and exploration. - Have performance gains been confirmed?
The announcement outlines intended benefits but provides no quantified results from the partnership.
Disclaimer
Prepared for Mining Herald for general information only. This article does not constitute investment advice. Expected technology benefits remain subject to implementation and operating results. Readers should review company disclosures before making investment decisions.
Filed under
About the author
Jonathon Brown
Jonathon Brown began his career as a broadcaster, working across markets in British Columbia before moving into financial journalism. Since 2017, he has specialised in stock market reporting, covering emerging companies across the healthcare, technology, mining and consumer sectors. He brings more than 15 years' experience to his reporting. A graduate of Vancouver Island University and the British Columbia Institute of Technology, Jonathon is focused on delivering clear, balanced reporting for investors.
