HBP# TB0698

A09-23-0015

William E. Youngdahl Kannan Ramaswamy

Te Digital Transformation of Freeport-McMoRan: Te Strategic Use of Agile, AI and Data Analytics

“Working in a 100-year-old organization in a 10,000-year-old industry, how do you go about building a team that really starts to leverage innovation and build this kind of world-class next-generation data science capability?”1

— Jamie Milne, Freeport-McMoRan’s Head of AI and Analytics Programs, Public Sector, WWT

As 2023 drew to a close, Richard Adkerson, the CEO and Chairman of Freeport-McMoRan (FCX: NYSE) had steered the company for 20 years at its helm. Describing him as “the elder statesman of the copper industry,” Te Economist2 observed that he had “seen it all, from short-term booms and busts to the China-led supercycle, and from industry fragmentation to consolidation.” Adkerson believed that the time for another upswing in copper was near, but this time, he expected that things could get quite challenging. He had told Te Economist, “[t] here is just a scarcity of actionable investment opportunities in the world today,” suggesting that demand spikes could not be addressed by bringing new reserves of copper ore online since copper mines took almost a genera- tion from exploration to commercialization. Indeed, the best of times seemed to lie just ahead, with demand for copper expected to skyrocket on the wings of the energy transition underway from fossil fuels to renewables. However, some believed this could portend the worst of times when copper miners would fnd it challenging to meet demand and any unmet demand would mean loss of revenues and perhaps even trigger the search for alternatives to copper. It was against this backdrop that Adkerson had chosen to address the issues of operational efciency and product yield at Freeport’s vast copper operations spread out across the world.

Freeport had chosen to ride the crest of the Fourth Industrial Revolution (4th IR), investing signifcantly in high-technology tools such as artifcial intelligence and data analytics to eke out more copper from remaining ore in its current mines. Te eforts had started to pay of for the most part, with efciency increases at some of the largest mines that the company operated. Starting with the copper mine in Bagdad, Arizona, the company had incremen- tally expanded the approach to encompass the entire U.S. operations under the Americas’ Concentrator program, an initiative that promised to bring efciency gains to all its copper mines across both North and South America. While charging ahead with the new technologies seemed like a logical strategy, it brought the worrying prospect of imitability. BHP, Grupo Mexico, Codelco, and Rio Tinto, all major competitors of Freeport, were deploying similar technologies and approaches at their mines to enhance metal recovery, increase operational efciencies and even prospect for new reserves. Could technologies such as AI, machine learning, and data analytics become the bedrock of competitive advantage in the future for Freeport? As one analyst remarked, “No doubt, there will be a lot of hype in the years ahead about AI’s ability to change the landscape in mineral discovery but history proves there’s nothing new under the sun.”3 With access to such technologies becoming more common and easy, how would Freeport ensure that it can cloak such technologies with the intangible sources of advantage using its human capital to ensure sustainability? Would it be able to fend of its competitors by rapidly scaling the deployment of AI and related technologies across its mines? What challenges would it have to overcome to hone its execution skills in this regard? Would the lessons learned from its Americas’ Concentrator program be transportable across borders to other regions such as Asia? Tese were likely some of the questions that both business and technology leaders within Freeport pondered as they sized up the immensity of the challenges ahead. It was probably a source of additional concern that Freeport’s shares had barely stayed even for the year while its cross-town rival Southern Copper’s shares had appreciated 20% in the same period. Was this the sign of challenging times for Freeport ‘s competitive advantage in the industry? How would the company ensure its dominance in copper continued?

Copyright © 2023 Tunderbird School of Global Management, a unit of the Arizona State University Enterprise. All rights reserved. Tis case was prepared by Professors William E. Youngdahl and Kannan Ramaswamy for the purpose of classroom discussion only, and not to indicate either efective or inefective management.

For the exclusive use of M. Corella Zelaya, 2024.

This document is authorized for use only by Mariana Corella Zelaya in MIS_520_Summer_6_Tomblin taught by Michael Tomblin, Westcliff University from Jul 2024 to Aug 2024.

Te Copper Industry: Prospects and Challenges Copper is perhaps one of the most crucial metals that has powered human civilization over thousands of years. Remarkably, this metal still remained vital to human progress and was seen as one of the most important elements that would drive the global economy, especially as the world was transitioning from fossil fuels to renewable energy. Ranked third in terms of value in global trade in metals behind iron ore and gold, copper accounted for roughly US$183 billion in 2022. Worldwide production of copper stood at roughly 21.9 million metric tons (MMT) in 2022 according to the International Copper Study Group, a UN-chartered intergovernmental body that studied demand and supply of the metal. Many analysts expected that demand for copper could skyrocket as the green energy transition unfolds, with some, such as mining company BHP estimating that demand would double in the next 30 years, and consulting company McKinsey estimating that demand would grow to 36.6 MMT, representing an increase of 67% of current world production even as early as 2031. S&P Global, a peer consultancy company, ofered a rosier forecast of copper demand at 50MMT by 2035, which would amount to doubling copper produc- tion in just a decade. It appeared that there could be a bonanza on the horizon for copper producers. Much of the optimistic demand projections were founded on the assumption that the increasing emergence of a renewable energy economy at a global level, will generate an insatiable appetite for copper. It was expected that by 2030, the demand for copper from solar, wind, and electrifcation would alone require the entire output of copper mined in the U.S., a volume that was expected to double by 2050 according to the International Renewable Energy Agency.

Roughly 40% of the world’s copper originated in Chile and Peru, which were two of the dominant producers, with other countries such as China, United States, and the Democratic Republic of Congo bringing up a strong second tier of producers. Although copper deposits had been discovered in other countries such as Mongolia, major new fnds were extremely rare. Exploration for new reserves took an inordinate amount of time because the process was not as technology-enabled as in peer industries such as oil and gas, where seismic imaging was routinely used in the prospecting process. Once reserves had been identifed, the frms had to run the gauntlet of the regulatory process, which included very stringent environmental clearances before any mining operations could commence. In the current global scenario with an increased emphasis on ESG (Environment, Social, and Governance) im- peratives, miners took years to obtain the requisite licenses. Te Economist reported, “It takes years to go from licensing to operating an oil well, [but] it can take a generation to develop a “greenfeld” copper mine.” Given this slew of constraints, miners often tended to invest signifcant resources in developing their proven reserves versus prospecting for new ones. Tankfully, the life of a copper mine was much longer than an oil well, over 100 years or more in some locations. However, here too there was a constant battle to identify the highest grades of ore possible and as mines in both Chile and Peru had shown, declining ore quality was a reality most miners had to live with.

If the demand for copper were to increase as had been projected, a situation that was hardly guaranteed, miners would have to bring production online very quickly or miss the opportunity to serve the new customers. Given the constraints in fnding new deposits, it was expected that the mining frms had but two choices. Te frst was called brownfeld development, entailing the development of existing reserves at established mines, a process that typically took about six to eight years to start operations. Te second would be the technology route, which entailed deploying new approaches to increase yields in existing mines. For example, the leaching process used in copper mines was a signifcant source of copper waste that could be addressed through better chemical processes. It was in this second pathway that many of the miners were evaluating the use of advanced digital technologies such as data analytics, digital twinning, and artifcial intelligence to improve process efciencies and enhance yields.

Freeport-McMoRan Freeport-McMoRan Inc., a multinational mining company and one of the world’s largest copper mining com- panies, was headquartered in Phoenix, Arizona, USA. Te company’s origins can be traced back to 1912, when the predecessor organization, Freeport Sulphur, was founded in Freeport, Texas. Te company initially special- ized in sulfur mining and changed its focus to copper during the mid-20th century. Te transformation from sulfur to copper mining was not just a shift in production but a fundamental change in the company’s business strategy and future direction.

Over the years, Freeport underwent a series of mergers and acquisitions that contributed to its expansion. Te acquisition of McMoRan Oil and Gas in 1981 resulted in the modern name that the company is known by today. Te most signifcant of these acquisitions, however, was the 2007 acquisition of Phelps Dodge, a major copper producer in Arizona. Tis move made Freeport-McMoRan the world’s largest publicly traded copper producer at A09-23-0015 2

For the exclusive use of M. Corella Zelaya, 2024.

This document is authorized for use only by Mariana Corella Zelaya in MIS_520_Summer_6_Tomblin taught by Michael Tomblin, Westcliff University from Jul 2024 to Aug 2024.

that time. Te acquisition was notable not just for its scale but also for its timing, as it occurred during a period when commodity prices were soaring. Te years immediately following the acquisition saw a period of fnancial volatility, largely due to the 2008 global economic downturn. Commodity prices, including copper, plummeted, afecting the company’s revenues and proft margins. Te economic conditions forced Freeport-McMoRan and other mining companies to tighten their belts, focusing on cost-cutting and operational efciencies.

In 2012, the company diversifed into the oil and gas sector with the acquisition of afliate companies, Plains Exploration & Production Company and McMoRan Exploration Co. Tis decision was met with skepticism, as it deviated from the company’s core competency in copper and gold mining. Eventually, the company divested these assets to focus more on its primary business, owing to a downturn in oil prices and pressure from investors.

Te global economic downturn that began in 2008 impacted copper prices, putting fnancial strain on Freeport-McMoRan. Te situation became even more precarious when oil prices began to decline signifcantly around 2014, negatively impacting the oil and gas assets that the company had acquired just two years prior. Te company faced a substantial debt burden following its acquisition spree, notably the 2007 purchase of Phelps Dodge and the 2012 diversifcation into oil and gas with the acquisitions of Plains Exploration & Production Company and McMoRan Exploration Co. Te company faced additional pressure from declining commodity prices, particularly copper (see Exhibit 1) and oil, which impacted its revenues and cash fows. To navigate these challenges, Freeport-McMoRan embarked on a strategy focused on debt reduction and asset sales.

Exhibit 1. Average Annual Copper Price (USD/pound)

Source: “Copper Prices – 45 Year Historical Chart.” MacroTrends, www.macrotrends.net/1476/copper-prices-historical-chart-data.

Freeport-McMoRan sold of several key assets in 2016. Among the most signifcant was the divestiture of its 56% stake in the Tenke Fungurume copper-cobalt mine located in the Democratic Republic of Congo. Te sale fetched the company $2.65 billion, providing a much-needed boost to its liquidity for debt repayment. A09-23-0015 3

For the exclusive use of M. Corella Zelaya, 2024.

This document is authorized for use only by Mariana Corella Zelaya in MIS_520_Summer_6_Tomblin taught by Michael Tomblin, Westcliff University from Jul 2024 to Aug 2024.

Around the same time, the company also ofoaded certain oil and gas assets, including deepwater properties in the Gulf of Mexico, for approximately $2 billion. Tis sale was crucial as it helped Freeport-McMoRan distance itself from the oil and gas sector, allowing it to refocus on its core mining business. In addition to these sales, the company further bolstered its balance sheet by selling a 13% stake in the Morenci unincorporated joint venture, a signifcant copper mine in Arizona, to Sumitomo Metal Mining for $1 billion in cash. Tese asset sales collectively played a pivotal role in improving the company’s fnancial health during a challenging period.

By the end of 2022, the company produced approximately 4.2 billion pounds of copper, a nearly 10% increase compared to 2021. It operated 10 copper mines across the world (see Exhibit 2).

Exhibit 2. Freeport-McMoRan’s Mines

Mine Location Metals

Morenci Arizona, USA Copper and molybdenum

Bagdad Arizona, USA Copper and molybdenum

Saford/Lone Star Arizona, USA Copper

Sierrita Arizona, USA Copper and molybdenum

Miami Arizona, USA Copper

Henderson Colorado, USA Molybdenum

Climax Colorado, USA Molybdenum

Chino New Mexico, USA Copper

Tyrone New Mexico, USA Copper

Cerro Verde Peru Copper and molybdenum

El Abra Chile Copper

Grasberg Minerals District Indonesia Copper and gold

Source: Freeport-McMoRan’s 2022 Annual Report.

Copper Processing Copper processing is a complex endeavor characterized by fairly high capital and labor costs. It begins with the mining of the ore, which is less than 1% copper, and concludes with sheets of 99.99% pure copper called cathodes. Te most common types of ore, copper oxide and copper sulfde, undergo diferent processes, namely hydrometallurgy and pyrometallurgy, due to their distinct chemistries. Copper oxides are abundant near the surface but are considered low-grade ore with a lower concentration of copper. Despite requiring more ore to be extracted and processed, this process is less expensive, allowing oxides to be mined at a proft. Conversely, copper sulfde ores, though less abundant, contain higher amounts of copper, and while the processing costs are higher, more copper can be extracted from sulfde ores. After the ores are mined and transported, they are sent to a concentrator. Here, the ore is subjected to a series of physical processes to concentrate the copper minerals and separate them from the gangue (worthless rock).

Oxide ores are generally processed using hydrometallurgy, involving three steps: heap leaching, solvent extraction, and electrowinning. Heap leaching uses percolating chemical solutions to leach out metals, suitable for low-grade ore. Te leaching reagent (dilute sulfuric acid) dissolves the copper from the ore, and the resulting “pregnant” leach solution contains 60-70% copper. Solvent extraction causes the copper to move from the leach solution into the solvent, leaving impurities in the leach solution. Electrowinning is the fnal step, where an electrical current passes through the copper solution, plating copper ions onto a cathode as 99.99% pure copper.

Sulfde ores are processed using pyrometallurgy, involving froth fotation, thickening, smelting, and elec- trolysis. Te crushed ore is further processed at a mill and then subjected to froth fotation to separate copper

A09-23-0015 4

For the exclusive use of M. Corella Zelaya, 2024.

This document is authorized for use only by Mariana Corella Zelaya in MIS_520_Summer_6_Tomblin taught by Michael Tomblin, Westcliff University from Jul 2024 to Aug 2024.

minerals from the gangue. Te froth is then poured into large tanks called thickeners, and the solids are fltered to remove excess water. Te fnal product of this stage is copper concentrate, containing 30% copper, which is then sent to the smelter. High temperatures are used to purify the ore in a series of smelting steps, producing molten anode copper, which is 99% pure copper. Electrolysis is the fnal process, refning the copper anodes in a tank full of an electrolyte solution, producing copper cathodes that are 99.99% pure copper.4

Te Bagdad Project Freeport-McMoRan (FCX) chose its Bagdad mine for implementing data analytics and AI due to its status as one of the company’s oldest and highest-cost operations. Te Bagdad copper operation in Arizona had been mined since 1882 and had an average reserve copper grade of 0.32%, one of the lowest in the industry. Te declining ore quality and the associated increase in operational costs made Bagdad a prime candidate for the implementa- tion of innovative solutions to improve efciency and cost competitiveness. Freeport’s journey to embracing AI and data analytics in its mining operations signaled a signifcant shift in the mining industry toward what was considered the “age of the operator.” In this new era, companies aimed to maximize proftability from low-grade ores that would have been considered waste just a decade earlier. Te implementation of data analytics and AI at the Bagdad mine was aimed at unlocking performance and enhancing the mine’s competitive position by leveraging extensive historical operating data available for the complex. Tis was seen as a way to revitalize the mine and provide a blueprint for a company-wide transformation.

Identifying enhancements in a mine that had already undergone years of productivity enhancements posed a signifcant challenge. However, Freeport-McMoRan was equipped with an abundance of high-quality data to analyze. Approximately a decade earlier, the company’s chief information ofcer, Bert Odinet, had led an initiative to unify the methods each site used to gauge and document individual site performance, and establish a central- ized data repository to house these metrics. With the advent of afordable and dependable wireless mesh networks, Freeport-McMoRan integrated them across all its locations. Tis advancement enabled the company to instanta- neously collect and synchronize performance data every second in the data warehouse, facilitating real-time analysis and correlation. Subsequently, maintenance crews advocated for the incorporation of more network devices and performance sensors on the organization’s trucks, power shovels, and fxed machinery. Te data from these sensors were manually transferred to the data warehouse by the teams, aiming to refne maintenance procedures and opti- mize equipment functionality. Tis advancement enabled the company to instantaneously collect and synchronize performance data every second in the data warehouse, facilitating real-time analysis and correlation.5

Te initiative was met with skepticism and cultural resistance. Te air was thick with doubts about new tech- nologies, approaches, and processes. Te organization, although rich with a culture of innovation and continuous improvement, housed many individuals who were skeptics of these new technological winds. Te initial steps were like walking on a tightrope, balancing between the known and the unknown, proving the value of data science through successful projects to gain wider acceptance. To overcome this resistance, the project leveraged a combination of team composition, experimentation, and agile methodologies. Te leadership team recognized that innovation starts with people. It was one thing to have sophisticated data analytics tools, but quite another to have a team that knows how to wield them efectively. Te organization needed to nurture an environment where people felt safe to innovate and bring forth new ideas. Successes were celebrated, creating a culture of positive reinforcement, and failures were seen as learning opportunities rather than setbacks. It wasn’t just about installing new software or up- dating old machinery. It was about shifting mindsets, about encouraging each employee to think like an innovator, to see the data not as numbers on a screen but as the keys to unlocking unprecedented efciencies.

Te leadership team recognized the need for a specialized workforce. Tey adopted a two-pronged strategy for talent management in the feld of data science. Initially, external experts with in-depth technical knowledge in data science were brought in to guide the analytics projects. Tese experts were profcient in modern technologies and methodologies, serving as the core technical team for data science initiatives. Concurrently, an internal survey was conducted to identify employees with a strong interest or basic skills in analytics. Although these employees were not experts in data science, they had a solid understanding of the company’s business operations. Tey were subsequently trained in more advanced tools like Python, SQL, and machine learning platforms, moving them away from basic data analysis tools like Excel.

Te external data science experts ofered the technical skills required for advanced analytics, while the upskilled internal employees provided insights into the company’s business operations. Te result was an orga- A09-23-0015 5

For the exclusive use of M. Corella Zelaya, 2024.

This document is authorized for use only by Mariana Corella Zelaya in MIS_520_Summer_6_Tomblin taught by Michael Tomblin, Westcliff University from Jul 2024 to Aug 2024.

nizational shift toward a more robust capability in analytics. Te team structure aimed to solve problems more efciently by combining diferent sets of expertise. Tis setup also facilitated quicker adaptation to new analytics methods and technologies.

Metallurgists and plant operators were integrated into site-level development teams to ensure a practical evaluation of digital approaches. Tis cross-functional approach helped to quickly identify and rectify faws in the AI models at an operator level, ultimately gaining the trust of operators and metallurgists. By the time the AI tools were ready for deployment, there was already strong buy-in from the teams. Additionally, it became clear that the knowledge of operators was essential for guiding the development of AI algorithms.

Experimentation played a crucial role in proving the value of the project. Te team focused on creating digital solutions that could improve every aspect of operations and be easily scalable to all of Freeport’s mines. Te goal was to prove the value of the implemented solutions and demonstrate their impact on operational efciency. Agile methodologies were also instrumental in providing structure for cross-functional experimentation and iterative work. Agile work methods allowed the team to operate faster and better, focusing on developing minimum viable products that could be continuously improved. Tis approach con- trasted with the traditional goal of perfecting a solution before deployment. Agile coaches were brought in to train teams in agile meth- odologies, ensuring the entrenchment of this capability within the organization.

Te Agile Manifesto, published in February 2001, focused on four values and 12 principles for agile software de- velopment. It was created by 17 software developers seeking an alternative to more linear product development processes. Te four main values were:

1. Individuals over processes and tools. It valued team collabora- tion and teamwork over work- ing independently and doing things “by the book.”

2. Working software over compre- hensive documentation. De- veloping functional software was prioritized over additional work like documentation.

3. Customer collaboration over con- tract negotiation. Agile teams allowed customers to guide the software development process, prioritizing customer collabo- ration over the fner details of contract negotiation.

Exhibit 3. Te Values and Principles of Agile Software Development

Agile Values Agile Principles 1. Individuals and interactions over processes and tools

1. Our highest priority is to satisfy the customer through early and con- tinuous delivery of valuable software.

2. Working software over comprehensive documentation

2. Welcome changing requirements, even late in development. Agile processes harness change for the customer’s competitive advantage.

3. Customer collaboration over contract negotiation

3. Deliver working software fre- quently, with a preference to the shorter time-scale.

4 . Responding to change over following a plan

4. Business people and developers must work together daily through- out the project. 5. Build projects around motivated individuals. Give them the environ- ment and support they need, and trust them to get the job done. 6. Te most efcient and efective method of conveying information to and within a development team is face-to-face conversation. 7. Working software is the primary measure of progress. 8. Agile processes promote sustain- able development. The sponsors, developers, and users should be able to maintain a constant pace indefnitely. 9. Continuous attention to technical excellence and good design enhances agility. 10. Simplicity—the art of maximiz- ing the amount of work not done— is essential. 11. Te best architectures, require- ments, and designs emerge from self-organizing teams. 12. At regular intervals, the team reflects on how to become more efective, then tunes and adjusts its behavior accordingly.

Source: Manifesto for Agile Software Development, agilemanifesto.org.

A09-23-0015 6

For the exclusive use of M. Corella Zelaya, 2024.

This document is authorized for use only by Mariana Corella Zelaya in MIS_520_Summer_6_Tomblin taught by Michael Tomblin, Westcliff University from Jul 2024 to Aug 2024.

4. Responding to change over following a plan. Agile project management was fexible, allowing teams to quickly shift strategies and workfows without derailing an entire project.

From the four values, 12 principles focused on satisfying customers through early and continuous improve- ment and delivery, welcoming changing requirements, delivering value frequently, breaking the silos of projects, building projects around motivated individuals, maintaining a sustainable working pace, and regularly refecting and adjusting the way of work to boost efectiveness (see Exhibit 3).6

Agile enabled what the team called “one win at a time”.7 Tat frst win was a data model that scrutinized three years’ worth of sensor data from the Bagdad mine’s concentrating mill. Despite prior projections by Free- port analysts that the mill was operating at peak capacity, the team examined the possibility of further enhancing efciency without additional capital investment. Te insights obtained from this model instigated substantial alterations in the methods Bagdad’s staf used for copper extraction. Te team created an AI model named TROI (Troughput, Recovery, Optimization, and Intelligence) to predict the processing plant’s behavior and copper recovery rates under various conditions. Te model optimized settings for maximum copper production and is- sued recommendations at intervals of one to three hours. Te company also developed a model called CHLOE (Crusher Hauling Loading Optimization Engine) for analyzing data from its mining operations. Te creation and fne-tuning of algorithms did not eliminate the need for human intervention, however. Lulu Raymond, a metallurgist at Bagdad, observed, “TROI doesn’t always give fully acc