Anglo American plc (AAL) Earnings Call Transcript & Summary

Afternoon, all. Thanks very much for joining us. Today is an update on our technology journey. Technology is a key part of our core capabilities. It's integral to our holistic sustainability, and it is making our world-class assets even more competitive. Tony and the team will take about 65 minutes to run through the slides then we'd be very happy to take questions. Stephen Pearce will join us for that. And I'd like to make you think that in many ways this presentation is building on the sustainability presentation which we hosted from a couple of weeks ago. Next slide is from the lawyers. And with that, Tony, over to you, please.

Thanks, Paul. Welcome, everybody. It's a day that we really look forward to be able to present, if you like, our update to you. Before I start the presentation proper, can I just introduce the team of presenters that I have with me today: firstly, Matt Daley, Group Head of Mining; and Dave Palmer, the Group Head of Business Improvement; Rohan Davidson, Group Chief Information Officer; Arun Narayanan, the Group Chief Data Officer; Donovan Waller, Group Head of Technology Development; and finally, Pierre Herben, the Group Head of Carbon Neutrality. So we'll get started. So thanks, everybody. Firstly, all our work is really built around our fundamental company purpose. That's to reimagine mining to improve people's lives. And I think that's the context that you -- that we will place everything around. Technology and innovation are central to how we are creating a fundamentally more sustainable business. If we step back to around 2015, when we started FutureSmart: We talked about changing the footprint of mining; changing the inputs around water; around like energy, the way it looked; the way that we interacted with the people around us and with our workforces. If we look at some other adjectives, we also talked precision. We talked innovation. We talked step change ambitious and transformational. We talked about reliability, performance, best in class and basically setting ourselves up for future 15 to 20 years of leadership and success. And I think also, and it needs to be noted, we're talking about end-to-end systems, not pieces of a system but basically goes across our entire business. It's clear that the boundaries between FutureSmart and our sustainability work and program actually don't exist. The 2 areas are inherently interconnected and same of -- and part of the same purpose and ambition. Just how we choose to describe them to the external world changes, but in the end, you can slice and dice it all sorts of ways. It's just the same thing. FutureSmart today continues to develop, continues to mature and morph at real pace. And again I'd like to thank you and take the opportunity to update you and bring you along on our journey. The -- we'll take you through some broad headings: firstly, operational excellence; digitalization; innovation; and carbon neutrality. They are the core of the broader program that I talked about and ambition. There is a lot of other activity and a lot of other focus areas, but that's too much for today's program and perhaps a topic of a future event with you. With that, I'd like to hand over to the Group Head of Mining, Matt Daley, I've just introduced, who will take you through delivering our full potential through P101. Over to you, Matt.

Excellent. Thank you, Tony. So today, we're really excited to share some of the significant progress we've made in our full-potential P101 program, a program underpinned by the Anglo American operating model. Now P101 takes a systems approach looking at the value chain to ensure our improvement initiatives are targeted, leveraged and accelerated for maximum impact. Now when we commenced this program late 2017, we completed a self-assessment using available industry benchmark data to establish our baseline position against our industry peers. We had some strong areas, some not so strong, but by and large, we were sitting around the 60th percentile. Now fast-forward to the start of 2021, and across the company, we have moved up the curve significantly, a great result, but our work is not done. The foundation is strong and confidence levels are high that we will close the gap to benchmark and beyond over the next 18 to 24 months. So let's start with a couple of operational examples in mining and processing to explain how we're delivering these results. So first up, mining. Now tactical mine design is foundational to equipment productivity and hence moving material safely at the lowest unit cost. Utilizing 3D design tools to visualize equipment operating parts, we are optimizing equipment activities and interactions and providing ideal geometries to deliver benchmark productivity. Now we have taken this approach into life-of-mine planning as well as short- and medium-term design, ensuring across all time frames planning decisions deliver the right material to the right place at the right time whilst locking in P101 equipment performance over the mine's life. Now this work has been a key contributor to seeing shovel productivity increase since start of the program. The before and after photo you see on the screen is an example of one of our operations in South Africa, where best-in-class tactical mine design has produced a very different-looking pit. We have different geometries, eliminating multiple vertical interactions, enabling step changes in productivity. On to blasting. Now another key area of focus in the P101 program has been drill and blast. Now again with a focus on value, blasting has been key to improved equipment performance and concentrated milling rates. It also is delivering improved safety and social performance at our operations. So 2 key areas of focus. The first is blast design. And Los Bronces is a great example, where we've delivered a 14% improvement in fragmentation through design changes targeted to deliver optimal downstream milling benefits. Changes to blast design have also allowed us to eliminate trim or buffer blasts that are usually designed to sit against a pit wall. Now this improvement has reduced the frequency of our blasts, providing more productive hours in a week, and improved our shovel loading rates. Now the other key area of focus has been drill and blast execution. The best designs in the world are only useful if we can execute them with precision, so to support this, we've rolled out across the entire group a standard real-time, in-field digital platform that tracks compliance through all aspects of drilling blast performance and enables designs to be modified using real-time accurate data as field conditions change. Now this platform is also integrated with our autonomous drilling fleets. Next slide, please. So the final example for mining, and I should mention that this is just a small subset of the P101 work programs underway, is how we are using digital twins to optimize haul roads and improve operating performance. Now digital twins allow us to model theoretical truck and operator performance and measure it against actual in-field performance. We can then simulate modifications to roads before making changes in the field, or we can address training and performance with operators to improve productivity. Now such training is delivered through truck simulators that have been tailored specifically to our P101 objectives, and we combine that with in-field coaching and mentoring. Now the example on the slide from Mogalakwena illustrates the power of these tools. We were able to deliver a 23% reduction in haul cycle times. That kind of improvement provides optionality. Either use that additional capacity to move more material to increase revenues. Or if there's a higher-value case, we focus on efficiencies and parking up equipment, dropping unit costs and hence future capital requirements. Next slide. I would now like to talk through some of the P101 examples for processing, where the focus has been on the key value drivers of throughput, run time and recovery. I'll start with throughput. Now Minas-Rio in Brazil is a great example where we've seen a 7% improvement in plant performance driven by systematic debottlenecking and circuit reconfiguration. So what do we mean by this? Well, basically it's changes to pumping and piping to modify the circuit, ensuring we eliminate bottlenecks and set the plant up to run as efficiently as possible utilizing the existing infrastructure. This means we are banking the improvements for close to 0 capital costs. Run time, another key driver. And we're seeing the application of our P101 asset maintenance strategies and tactics provide more operating hours per year. Now the Moranbah plant in Queensland is a great example, where a 12% improvement in run time has been delivered through this targeted P101 focus. Next slide, please. Now recovery in processing plants is critical, and the P101 program is focused on delivering recovery in line with the full ore body potential. On the previous slide, I talked about the system reconfiguration and debottlenecking at Minas-Rio to improve throughput. Now what is really great about this program of work is it also delivered a 2% improvement in mass recovery. When added to throughput, this saw an overall 9% improvement in plant performance post the P101 work program. And finally, I'm just going to touch on revenue. Now Kumba is a great example of where multiple disciplines have come together to deliver an outcome. In this example, we had geology, geotech and mining involved with new blast designs and then processing and maintenance engineers implementing modifications to the plant crushes and screens. Now when all combined together, this delivered a 5% improvement in the lump-fine ratio across the 2 mines at Kumba. Now with lump iron ore attracting a higher premium, we saw an increase in revenue for the site post this work. Next slide. So best practice principles. Now we all know that improvement programs can deliver short-term benefits due to the added management attention and focus, and then sometimes we see performance drop away when attention is devoted to the next issue. Now the P101 program is different. It was designed, from day 1, to change the way we work. It is fully integrated into the operating model. We've ensured best practice principles are embedded into management routines, and they have been documented and supported by flexible modern training programs that are part of our governance practice and standards. P101 is becoming our normal. And to that end, we believe this program uniquely positions us to take the same P101 approach as technology is deployed and we redefine the industry benchmarks. We are planning for new ways of mining and processing as part of the FutureSmart program, and we are going to be positioned at the top of the curve as the industry changes. Next slide, please. Now the final slide here just attempts to further highlight some of the P101 examples. We have seen improvements in truck performance that deliver increase in material movement or further reduce our haulage costs. We've seen loading performance across the group improve; example of that there in our platinum business that saw a 25% increase in annual volumes at Mogalakwena with their rope shovel, which allowed us to park up other high-cost loading equipment. A very similar story across our processing plants, we've seen improvements across the board in key focus areas of run time, throughput and recovery. Now the operational improvements are important that we're celebrating, but what really matters is how this translates into value. And at that point, I'm going to hand back to Tony O'Neill. Thanks, Tony.

Thanks, Matt. As an aside: Matt and a couple of cohorts, Andrew Marxsen and [ Andrew Mcclelland ], have taken P101 as an idea on a catchy slogan from concept in 2017 to part of the everyday language across the business today. It's been a fascinating exercise of when you can click at the right idea with an organization. Today, the operations look at and learn from each other. And there's [ not, without a ] little bit of competition between them, all helping the equation. The benefits of P101 at $2 billion, and that's rounded and gross, clearly are significant. Capital intensity has improved roughly 30%. And planning and analysis, as Matt said, of each aspect of the operations are at levels that we haven't seen before. And that's giving us a really different insight and understanding of the business, particularly around value flows, pluses and minuses; and further opportunity, particularly around the area of another layer of stability again. And that drives, if you like, further value unlock. It's all built around the operating model. And with that, I'd like Dave Palmer, please, the Group Head of Business Improvement, to talk about the operating model and value release, please. Over to you, Dave.

[indiscernible] very much, Tony. So as you just heard from Matt, we've made significant progress in the implementation of the P101 program, which is underpinned by the deployment of the Anglo American operating model right across the business. These operational improvements across safety, productivity and cost performance have delivered material value. As we presented in the annual results earlier this year, we've now delivered over $2 billion of gross annual improvement as at the end of 2020 through a variety of initiatives that span right across our commodities. Now COVID has certainly had an impact, and it continues to do so. However, we've taken a range of actions to ensure we can stay the course. On our current projections, we still see significant further value from the P101 program over the next few years, with significant contribution coming from a number of key initiatives across throughput at Los Bronces, further optimization at Minas-Rio as well as significant improvement from the longwalls put in Australia as we return to normal operations. So with all this in mind, we foresee the gross potential actually delivering up to $3 billion by the end of 2022. Now as we continue to ramp up the improvements that we are delivering, we also need to look at the challenges that we face in converting this to the bottom line. And we need to understand, how do we sustainably unlock that full potential and deliver that full value? So turning our minds to that: In 2020, whilst we delivered over $2 billion in gross improvement across P101, operating model and technology innovation, about half of that has hit the bottom line in terms of net EBITDA improvement for the same period. Now as highlighted in annual results, we still encounter operational instability and timing issues that currently prevent us from fully realizing this potential in terms of net EBITDA benefit, particularly in the same period. So that's not entirely unexpected, considering the rapid rate of improvement that we are delivering. Now to unlock this value, it requires us to stabilize and optimize these complex industrial value chains which are made up of the individual processes and subprocesses, and this is where the operating model and P101 really come to the fore. It's the stability of these subcomponents that is key to long-term sustainable value. Now it's worth noting that every mining company on the planet encounters this; matter of fact, every linear industrial system, for that matter, so whilst the problem is common, the point of difference is how we're responding to it. Our operating model enables us to identify, quantify and address this in a really systematic way to reduce that instability and the variability at every level through the business. That's what allows us to increase performance right the way through the value chain. Now in order to further accelerate this across the organization, we're also engaging in sprint-based interventions at our operations to address this instability, reduce this variation where it matters most. Now it's this systems approach that enables us to bring forward greater value from our improvement initiatives and also allows us to be far more selective in terms of where we build capability and invest in capacity, driving maximum value. We're also leveraging the superior capabilities that we have within our technical and sustainability team to not only deliver the physical improvements, some of which Matt has just taken you through, but to further enhance our capability through the digitalization of our operating model right through the business. And you're going to hear a lot more about that today. Ultimately, we're now better leveraging our people across the group to realize the full potential of business and harnessing the power of process and technology. Now the impact of this is that we will see an increasingly greater proportion of our improvement flowing through to net EBITDA. Now I appreciate all that can sound a bit theoretical, so here's what it looks like in practice. Here's an example of a recent sprint in our Los Bronces operation that was targeting truck payload increases for one of our fleets. You can see on the left-hand side how we stabilized that process by controlling the outliers, steadily increasing that average payload performance as the sprint program progressed, as we move to the right-hand side. How did we achieve this? It's a really systematic approach to the process, so it includes some quite detailed quantitative analysis of key value drivers and understanding reference distributions. And that's what enables us to pursue the high-leverage opportunities. We targeted operator proficiency. We reduced variation through some operator assist technology, and we optimized the equipment configuration to give us maximum capability. So what is the impact of that? Ultimately, increased truck payloads means increased production volumes and reduced haulage costs. So let's have a look at how this all comes together and ultimately how we deliver our part of the $3 billion to $4 billion in EBITDA improvement that is commitment we've made. Now we initially identified that the operating model in P101 would contribute about $1.5 billion to this target. As I said earlier, our gross improvement, we see that now delivering up to $3 billion over the next couple of years. And perhaps most importantly, an increasing majority of this gross improvement is now hitting the bottom line as the operating model becomes more systemic and our targeted stability spreads become embedded across the business. As such, we're confident on meeting the net $1.5 billion commitment for our contribution, with further potential upside. Essentially we've delivered a lot of improvement through the operating model and P101 to the business, and there's plenty more to come. Tony, back to you.

Thanks, Dave. It was fair to say that there was some resistance to the operating model in the organization in the early days. This has certainly turned around under Dave's leadership. There's a real pull from the business, and Dave's challenge now is to get it rolled out even quicker. And he's certainly got an aggressive program this year. If we step back, though. Around 2015, our technical and industrial IT infrastructure was really behind where we considered appropriate to layer our FutureSmart program onto. And how we use data to analyze and optimize our work was really constrained by this. Using the operating model as the base principle, Rohan Davidson, our Group Chief Information Officer, has done wonders transforming our technical infrastructure, enabling our program to proceed at pace. I'll ask Rohan, please, to describe some key outcomes of this program. Rohan, over to you, please.

Thank you, Tony. Today, I'm going to talk about the wonders of advanced process control, what it is, how it improves our business performance and the contribution it makes to our sustainability journey. As part of our digitalization approach, one key component is to take the best technologies across all industries and adapt them to our operations. We've led heavily on the tools and techniques forged in the petrochemical industry, and we've developed our own industry-leading capability and advanced process control. APC automates the implementation of the operating model across our plant control systems, unlocking optimization that has previously been beyond our reach. At its heart, APC employs virtual operators or algorithms in closed-loop control that are responsible for optimizing a single process step to achieve a target output. This might be a mill rotation or maybe sump level or a valve setting. The technology is particularly useful where process steps involve time lags, like long conveyor runs or surge bins or float tanks. Operators in their conventional control systems struggle to accommodate these time lags, but they lend themselves very well to advanced mathematical modeling. In particular, with artificial intelligence and machine learning, we can usually produce models that outperform the best operators running these complex systems. [ At our large plants ], there are many APCs. Take for example our copper operations at Los Bronces. Our [ end state ] for Los Bronces has over 100 of these virtual operators. And with 3 years into the program, we've completed about half. And last year, the half that we've completed contributed over $60 million to the underlying operational performance of Los Bronces through improved recoveries, energy savings and throughput improvements. The graphs on the side here are indicative of the [ sort of ] performance that we can achieve through APC. To read these graphs: The conventional control portion is shown on the left-hand side on -- in red, and where the APC has its hands on the wheel, that's shown in green. At our Minas-Rio operations, APC streamlines our crushing and our metal separation processes. At our copper operations at Los Bronces, we've had some of the best results. Our giant SAG mills have APC models that help us with the grinding process, where there's a 10- to 15-minute residency time for all of the [ raw ] inputs that go into our SAG mills. Keeping throughput constant is important to match downstream metals recovery capacity. And you can see there the APC control, after a period of adjustment, is able to keep throughput in a much more constant range than we had under manual control. In this particular case, we've improved the stability by 32%. Now the APC itself is also looking to optimize energy efficiency and reduce water use. And in this particular case, for every tonne we mill, we've managed to reduce the energy consumption by 17% per tonne processed. As Tony had mentioned, at the start of our transformation journey, we had no underlying standardization across our plant and control philosophies or infrastructures and no way to integrate support services across them. With APC, we've integrated the system across plants, which gives us whole-of-business visibility for performance monitoring, support and for our incident response. Next slide, please. Our goal is to have all automatable processes under APC by 2024. Last year, we reached the halfway mark, and the systems that we've put in place contributed over $170 million to underlying earnings. We're on track to double that record next year. As well as the throughput benefits that drive the financial contribution, APC lets us optimize our energy inputs and our water consumption. Last year, we saved 173,000 gigajoules of energy, dropping over $5 million from our power bill. Most of this came from our Chilean copper operations, where APC controls the grinding mills, with a 7% energy reduction across the entire system. Each time we build one of these controllers, we look for opportunities to solve the same problem that exists perhaps elsewhere in the group. The copper mill logic is being retrofitted to our platinum operations, and the early stages of this project have already reduced our power consumption there by 4% across the system. In Brazil, a 12-month project to stabilize the temperature and the moisture of the feedstock for our nickel smelter has contributed directly an 8 kilotonne annual reduction in our CO2 footprint. And next slide, please. We hear a lot about digital twins. We're in the process of building digital twins for all of our own key operations. We take laser scans, which let us build detailed CAD models. We combine these with drawings and plant master data, and that gives us a digital twin. And these twins let us model different scenarios. They're used, for instance, by our processing teams to explore expansion options and improve flow sheets. What sets our approach apart, though, is the tight integration between the digital twins that we're building and the APC environment. Having real plant data gives us a full virtual asset. Including simulated instrument data lets us run a training environment not unlike a flight simulator but for plant operators, letting them practice recovering stable operations from scenarios they're never likely to experience in the real world. Our rich plant data passes through the APC environment, which gives us context for our data lake and for our analytics solutions, which my colleague Arun will talk to shortly. All of these technologies are baked into the [ commissions ] design of Quellaveco, our first fully digital mine. Thank you. Back to you, Tony.

Thanks, Rohan. As Rohan said, effectively his work has enabled connectivity and digital twins for a large part of our business. It's enabled us to move on a vision we had around artificial intelligence, how to be among the best to effectively simulate and optimize our business in real time to come to life. We also want to be in control of our destiny in this area. Arun Narayanan, our Chief Data Officer, has recruited an outstanding team in the data analytics stays -- space. And our ambition is to lead the industry in this area. So with that: Arun, you've [ got a higher bar to make ].

Thank you, Tony. I'm very pleased to be here. Good afternoon to everyone. I'm going to be unveiling VOX3L. Next slide, please, actually. Thank you. I'm going to be unveiling VOX3L, which is the mining industry's first digital transformation platform for data-driven decisions. It brings together some of the points that were already discussed today. It's the implementation of the operating model with machine learning and artificial intelligence. It encompasses the entire value chain for us, and it helps us really move forward into the future. What is VOX3L, and how does it work? What are the aspects of it? The first point, though, is about the operating model. The operating model tells us to do the right work in the right way at the right time. In that context, VOX3L is all about figuring out what that right is. Next slide, please. So VOX3L tries to bring a lot of things together. First of all, it is a platform built in the cloud. It's a very modern platform looking at all of the data that exists in our operations. Trying to bring that all together is the first step. Rohan referred to a variety of digital twins. That data, plus data coming in from our historians as well as geoscience data as well as external data, is all pulled into a modern scalable infrastructure that we call the data lake. This data lake contains the data of our operations and presents the first opportunity for us to start to integrate things. As we begin to build these integrated models, we are cleaning up the data. We're performing data management tasks at the lowest levels from data quality to the highest levels like data management. And this cleaned-up information allows us to create data-rich applications that are listed on the screen over here. And what are these applications and how they work form the rest of this presentation. It gives us a great opportunity to leverage the advancements happening in the computer science industry to bring machine learning, artificial intelligence, burst capacity for applications through cloud computing and much, much more. Next slide, please. Now VOX3L is very different. It's different because the entire system, the entire value chain modeling has been built specifically by us for mining, as opposed to looking at other alternatives which are specifically targeted just to equipment or just to other solutions. It looks at the full value chain. And also it takes in the expertise, the deep expertise, that sits in the T&S organization and the business units. And it encapsulates that knowledge in the way these applications and these solutions are designed. Now it takes all of the digital twins that were built at individual steps of the value chain, and it pulls together almost a systematic digital twin of the full value chain. And that's really what sits at the heart of VOX3L. Now Tony mentioned this earlier, if we go to the next slide. What we are really trying to do is that we are trying to take a stand on data ownership. The data that we generate as part of our operations is really an asset for us. It's our asset and we have to treat it as we would treat any other asset. We need to protect it. We need to make sure we are using it where it needs to be used, and we need to make sure that we're getting the best return that we can get from that asset. And that's what the VOX3L data program sets out to do: It gives us a pathway to begin to claim ownership of that data and begin to use that data to drive the best decisions that we can get. The other aspect of this is intellectual property. And intellectual property in this system is owned almost entirely by Anglo American. We do work with industry-leading partners like Microsoft for the cloud solution and we buy other components where they are necessary, but they are encapsulated away and the intelligence and the system is owned entirely by us. And what that means for us from a scalability standpoint is that we can essentially productize these solutions and we can deploy them at pace without incurring additional costs to all of our 31 operations. We've evaluated a variety of other options. Clearly you can build a bunch of point solutions. You can look at commercial off-the-shelf software and pull those all together to form digital solutions, but as we compare and contrast those options to VOX3L, it's very clear that VOX3L always comes out ahead of all of these because it gives us the quickest pathway to owning our data. It gives us an intellectual property and scalability of deployment, as well as it gives us an ownership in our own future in terms of how fast and how hard we can drive this program. Next slide, please. Now if we were to think about this as a system view, there are many things that are happening here, but as we introduce the applications, it's important to understand that these applications are integrated at the data level. And they give you opportunities to optimize the full system. And maybe one way to think about it is the first thing that you can get out of it is that you can just fundamentally eliminate waste that exists in the system. Now both Matt and Dave Palmer referred to this in their presentation, but maybe an analogy is a better way to go here. If you look at a system like Uber -- and many people love Uber, the transportation service, for many things, but what it's doing is it's really reducing the number of wasted minutes that passengers are waiting for drivers and drivers are waiting for passengers. And that elimination of that waste is what happens when you build such a holistic system. So similarly, we want to understand what ore bodies exist and what ore bodies should be mined and what ore bodies go through processing plants correctly so that we can just fundamentally eliminate waste from the system. There are specific applications inside the VOX3L ecosystem and these perform specific tasks. And starting off with [ S3 ], the sustainability application: These has -- these applications have capabilities to do, let's say, power analytics, methane gas analytics; the ability for us to track our environmental and our biodiversity impact; and also do greenhouse gas modeling. Moving on to [ G3 ], which is the geoscience application, clearly you'll get a clue, that G for geoscience. The pattern follows through the slide. And the geoscience application allows us to do rapid resource modeling. As we acquire core samples, we are able to use the power of artificial intelligence to quickly analyze and perform this analysis of core, which usually takes us a long time to do. As we rapidly reduce the time it takes us to do these core analyses, we can build ore body models very quickly. The same application, [ G3 ], also has capability to burst into the cloud to do model execution and further refine the ore body definition. Here we are not buying computers, but we are really just using the compute power available to us on -- from Azure on demand to be able to solve these problems. In the [ M3 ] product, for mining, we are able to optimize mine planning and also, in the context of P101, review our operational performance. In addition to that, we can track materials [ spatially ], which plays a role in the full value chain integrated play, but also we can understand what is in our stockpiles and model our stockpiles in detail. One of our flagship products is [ P3 ], and [ P3 ] sets out to change the game in the way we run our processing plants. We can optimize the recipe for our processing plants given the ore body [ feed ]. [ A3 ], for asset utilization and maintenance, gives us a way to make sure that we are using our physical assets to their best possible potential; perform maintenance tactics optimization; and of course, predictive maintenance. The health and safety of our employees is very important. There's actually an entire application dedicated to that, and [ H3 ] application sets out to deliver that outcome. In addition to being responsible for tracking some of our COVID-related responses, the [ H3 ] application digitizes critical controls; makes sure that we can do an automated TARP, trigger action response plan, as well as perform video safety analytics. Now there's a variety of applications over here. These are being built over time. And it's a journey for us to get these deployed everywhere, but some of these applications are actually live and they're available for our use at our sites. And this journey continues on for us as we deploy these applications. I want to share with you some specific examples of these applications being used at some of these sites in [ P3, A3 and H3 ]. And I'm also going to share with you some upcoming features in [ S3 and X3 ]. So starting off with [ P3 ]. What is [ P3 ], and what is it trying to do? The best way to think about it is it's like Google Maps for a processing plant. It helps you understand, as you have done before on paper, what your control charts looks like. It helps you see these electronically; and helps you understand, in this particular case, performance process review, which is a digital rendering of different controls, and understanding when those controls are outside the bounds of your KPI. The metallurgist can quickly review those actions and make remedies and make sure they are back on track to achieve the KPIs that they set out to have. What's maybe very special over here is that we are able to detect the state of the processing plant. We call it context. And as physical properties change, such as input or mineralogy or the ability to -- or someone changes the [ set ] points, the system automatically detects a context. And it's able to give you the KPIs in the sort of setting of that context. It's like understanding what your rev should be for a specific gear, but as your gear changes, your rev changes in a car. The same analogy applies over here. And just like Google Maps, it gives you the context of where you are, but what's coming and what's very exciting is recipe optimization. It's going to tell you what to do as well and give you the best recipe for the specific ore body feed that you're getting. And this is done with a context of what your ESG performance looks like. The other example that I have is the [ A3 ] application. The [ A3 ] application is at the Anglo Converter Plant in Rustenburg. And this is talking about maintenance. Typically, if you perform maintenance on a schedule, what can happen is that you can either over-maintain or under-maintain the asset. Under-maintenance means unexpected shutdowns, and over-maintenance can result to wasted operational efficiencies. What predictive maintenance does is allows us to monitor the health of the asset digitally and trigger alarms much in advance, allowing us to plan and execute the maintenance work as part of the operating model principles. This also makes it safer. And this particular example -- in the last 2 months, we've had 5 catches in the Anglo Converter Plant, which have saved us more than 121 hours of downtime. It's detected everything, from water levels that are unfavorable to nozzle and leaks and to furnace points that were not set to the right level and allowing us to make remedies to them much, much more proactively than we would have done otherwise. All of this ensures that we are able to counter the headwinds that we're facing. The next example that I want to share is [ H3 ], and this is the health and safety product and specifically the COVID response part of the health and safety product. Now there is a global dashboard on understanding the impact of coronavirus within our employees and our operations that we have built, but in addition to that, we have also built a comprehensive track-and-trace program to understand how the virus is spreading because we have deployed wearable devices for our employees. This allows us to proactively not only manage the health of our employees, but it has also made operational differences to us, specifically in one case in Namdeb where we were able to understand that one member of a fleet -- of a crew had exposure before the ship went out. So these are tangible examples of operational benefits as well in addition to the health and safety benefits from the same application. Now the next one that I'm going to talk about is an upcoming feature for the sustainability product, for [ S3 ]. Now sustainability sits at the core of what we are doing within the whole platform, and we are really trying to fundamentally drive waste out of the platform. One method of doing that is to model greenhouse gases in relation to incremental production steps, so this much production cost, this much greenhouse gas versus that much cost, that much -- and this gives us an opportunity to understand what decisions we need to take to balance our throughput and our financial outcomes in the context of greenhouse gases and our commitments that we have made there. Actually you can see that Google is also going down the same path. There is a green routing option that will soon be available in Google which will let you say, "I want to take a trip that's 2 minutes longer," but it is 10% more greener. And that's exactly the same concept that we are bringing in through [ S3 ], where we say this last 2% in production is costing me 20% of my greenhouse gases. And do I pursue that, or do I pursue something else? And it gives us that visibility in a very clear and understandable manner. Moving on to [ X3 ]. Now [ X3 ] is one of the most critical products in our portfolio. The X stands for integration; and it's central enough to our strategy that the [ X3 ] product actually makes its way into the name of the platform, VOX3L, itself. And [ X3's ] fundamental goal is to deliver the operational planning component. So the operational planning is part of the Anglo American operating model. And the digital operational planning tool or DOP delivers the promise of giving you a way to use the power of computers to get a much better plan out of the system. DOP builds value driver trees. Using these value driver trees, we are able to assess different scenarios and rank these scenarios for the best outcomes that we want based on whatever objective, function or balance of them that we are looking for. So you can either optimize the throughput or greenhouse gases or balance them out. And this operational planning tool has many advantages in the way the picture is drawn. First of all, it uses the data from the data lake, so everybody is working on the same holistic data set and we are all together in the same picture. Secondly, it gives you the ability to model different things in the computer system and find out the best answer for you for that given scenario. And thirdly, it ties very closely with the feedback portion of the operating model, which are all the applications that I spoke about, so [ G3, M3, P3 ]. They're all providing feedback as to how well the previous round of the operational planning went. So together, you can see that this is really laying the foundation for us to step into a version of the operating model that is fundamentally powered by digital technologies and for us to be able to do scenario planning in a way that was never possible before. This is pulling all of the information together from the full value chain and is driving this. So looking at the technology of this and the way we have built the VOX3L platform and the set of applications that sit in it, we have been quite successful financially. In the 3 years that we have existed as a team, we have spent less than $100 million and returned $150 million in value in the period ending in the last year. And this year, we are accelerating quite a lot and is -- almost planning to spend the same amount of money and to return that same $150 million impact just this year alone. That shows you the speed at which we are accelerating with the program. This program is going to be successful, as it depends on people if we take them along on this journey. Next slide, please. The digital literacy program is a key enabler for that part of it. So what is this program all about? As we begin to build a data-rich organization, we are expecting people to understand these technologies and to be comfortable with it, so to that end, we have rolled out what we call a digital literacy program. It's a lot of training that people can access online at their own pace, in bite-size chunks. And there has been a tremendously enthusiastic response in the takeup of that program. In addition to the basics, there is also an advanced nanodegree program for those who are far more committed. Think of it like a single course in a masters program that allows people to commit to one area of study, like citizen data scientists, and deliver that in a period of time and be much more confident about these technologies. The last part I want to talk about is process and what VOX3L does in terms of how mining processes itself changes. Next slide, please. And VOX3L allows us to enable remote work. So in one example over here is that we have built the integrated remote operating center in Santiago in Chile, and this gives us an opportunity to try to enable a way of working that was not really possible before. It reduces the need to make needless trips to the site. The people at the operating center have the same data that is available to people at the site, and it just brings about a step change in efficiency and productivity. This also is laying the foundation for us to bring in experts, who might be elsewhere, to look at the same problem and give you advice as well. So as another way of thinking about it is that the digital transformation work also transforms the way in which work itself is done. So pulling it all together. What have I spoken about, so far, is that there is a digital transformation that is underway within Anglo American. This digital transformation is promising to deliver both sustainability efficiencies as well as value growth. And the pillars on which it stands are a holistic transformation of people, process and technology. The people transformation is led by the digital literacy program; process transformation, partly through the VOX3L data program, partly through the way remote work is enabled. And the technology is a game-changing, differentiated, productized set of applications that sit on a modern cloud-based deployment of technologies that are available to us. Thank you, Tony.

Thanks, Arun. In -- the starting premise in putting together the data analytics team was that mining wasn't cool and we would struggle to get serious talent to join us. We've actually found the complete opposite. What we've found is that we've been able to attract people from organizations that we thought were cool. And the reason has been -- is really around our purpose. The people have joined us and they -- because they want to change the world. And they see our purpose as a vehicle to do that, that they can leave their impacts on the future for humanity. And we've got some -- an amazing organization. It's been really, really one of the joys to watch. If we move on, though. FutureSmart, where our program effectively began in 2014, has not been left behind while we talk about all these other things. Donovan Waller, our Group Head of Technology Development, many of you will know him, will take you through, if you like, our core -- let me say, traditional innovation program; and bring you up to date as to progress in that area. Over to you, Donovan.

Thanks, Tony. Good afternoon, everyone. For those of you who are new to our calls since our 2019 update, let me give you a short history of where we've come on our technology journey. Our early assessment on our FutureSmart journey was that we wanted a more precise and sustainable pathway for ourselves and the industry, one that would simultaneously achieve profitability and sustainability. As such, we looked outside our industry for help and we embraced open innovation and collaboration. In 2019, we shared with you a number of technologies that showed a lot of promise, and today, what we'll show you is the delivery outcomes and the progress of those technologies as well as some new and some more promising technologies. Over time, we've also become better at discovering new technologies and taking them through the development cycle as well as embedding them into our operations. In addition, as Tony mentioned earlier, FutureSmart has grown to embody the Sustainable Mining Plan; as well as to leverage off the Anglo operating model, P101 and digital. Let's move on to the technologies, bulk ore sorting. In ore bodies, grades occur in -- naturally in uneven distributions. And what we do is we take advantage of this by sensing the grade in the material as soon after the blast as possible, and we reject the obvious waste instead of processing it through the plants. What this means is that we have a higher grade that goes through the plant and then we speed up the mining rate in order to make sure that the plant is kept full. In an ideal situation, you're able to reject up to 40% of the mass for a 20% upgrade. And where we are in this process is that we've completed demonstration units at El Soldado, at Mogalakwena and at Barro Alto. And in fact, we've gone one step further in that we have a full production unit in Mogalakwena. Barro Alto have turned their demonstration unit into a production unit and are in the process of securing another. And we're well under our way in getting Los Bronces up and running [ and a ] full-scale rollout as well. Coarse particle recovery. CPR uses a HydroFloat recovery in order to recover metal at much larger particle sizes. What this means is that the particles spend less time in the mill, and we're able to increase our throughputs by up to 20%. Since particles are large and -- hence, when used in conjunction with our hydraulic dry stack, we're able to then recover more than 85% of the water. Where we are in this process is that -- and some of the images that you see there is that El Soldado unit is currently commissioning as we speak. And the Mogalakwena unit will be up and running towards the end of the Year. We've also managed to bring forward the rollouts at Los Bronces and at Minas-Rio. Looking then at a process that follows on from coarse particle recovery, we look at hydraulic dry stack. What we do is we use the sand created from the coarse particle recovery and we create a sandwich-like structure with layers of sand and conventional tails. The water from the fine tails drains through the sand layers. And as mentioned earlier, we're able to recover more than 85%, and what gets left is a stable dry stack that can be repurposed for other uses. Where we are at the moment is that we're constructing a full-scale demonstration unit at El Soldado. You can see that in the images at the top. And as we go along this process, we're sharing the experience with regulators in order to bring them along with the journey. This is -- really is -- hydraulic dry stack is applicable to anywhere in the business where we have the sand available, which typically comes from coarse particle recovery. It doesn't necessarily have to. Let's move on to hydrogen. In order for us to achieve our 2030 decarbonization targets, we need to displace 1.5 billion liters of diesel that we consume each year. Using green energy to power our operations, we use the excess to generate hydrogen on site. We're currently building the world's largest fuel cell-powered vehicle. The fuel cell-battery combination has been constructed and is currently under test. It'll soon be delivered to South Africa and installed on a truck which is ready for it in order to run before the end of the year. Our aim is to roll out at 7 sites before 2030, and so we have those in planning. And we aim to kick off the first production site, which will be Mogalakwena, in 2024. Microwave is a really interesting, new technology that I don't think you would have heard too much of before. And what happens is that some minerals respond well to microwaves whilst they pass seamlessly through the waste. This differential heating causes stresses and micro fractures in the rock. And this causes the rocks to soften and hence then melt quicker, resulting in an increased throughput. In addition, they tend to break along the ore grain boundaries, and that means that liberation becomes easier. We're currently looking at rolling out a unit on the Los Bronces pebble circuit, but the technology is applicable across platinum as well as copper. So whilst it's wonderful that we have this great portfolio of technologies, what it really says to us is that, if you really want to take advantage of them, you need to be able to get them rolled out quickly. And so what we've learned to do is to be able to ideate and develop the technologies quicker. And we've also found new ways to be able to implement them quicker at sites and at operations. What is really pleasing to me is that every technology that we've developed, so far, has actually been rolled out before the one before it. I'm not sure how long we can keep that up, but it's a really good trend. Microwave is our flagship example. In this instance, from the first laboratory results of having the results on what happened to the ores and the microwaves to us coming to a final design will take us only 1 year. And we anticipate that the rollout would only be a 6-month rollout. Part of what allows us to do this is some of the work that we've been looking at in terms of modular replicatable plants. So traditional mining technologies are developed over around a 15- to 20-year cycle. And so far, we've been able to reduce this down to a 2- to 5-year cycle, and we're still reducing it. So as a result, we're turning our speed of implementations into a competitive advantage. And we've got a whole lot more exciting technologies to come. Thanks, Tony.

Thanks, Donovan. I think it's worth noting that what Donovan talked to you today is a subset or, if you like, probably what we call our second wave. I think, through Donovan's good work and with his team, that we can see, if you like, another wave banking up behind this. That's for a future presentation. We're committed to 1/3 of our sites being physically carbon neutral by 2030 and carbon neutral by 2040. And Mark has certainly been talking extensively about that. I think we all agree the world needs it, firstly, but it seems like a really natural fit with our technical and sustainability ambitions. It's clear from an early stage that we need to develop a detailed granular pathway in which we could execute to achieve the targets because they're really quite onerous. We are very pleased with work to date. Pierre Herben, our Group Head of Carbon Neutrality who's recently joined us, and by the way, we're happy to have him, will take you through a high-level look at some of this work. Over to you, please, Pierre.

Thank you. Thank you very much, Tony. Good afternoon, everyone. I'm very happy to present you, yes, a high-level presentation of our journey towards carbon neutrality. As you said, Tony, last year, Anglo American group announced its ambition to reach carbon neutrality by 2040. And here maybe, before we go further, it would be good to look at the size of the problem. So in scope 1 and 2, if you look at the donut figure that we have here, it represents the sources of greenhouse gas emissions we have across the group. So the first source, you see it in orange, is basically the electricity. So it's the emissions generated by fossil fuels used by the power plants that produce the electricity we are acquiring to power our sites and mine operation. So it's a good 1/3 of the total emissions of our group. Then if we look at the second category, you could say these are the fossil fuel we use directly to power our trucks and for the mobility, like the diesel but also the coal that we use for some heating, to heat some process like smelters. And a third main source of emission, you see it on the left, in blue, fugitive methane, which actually is very specific to coal mines, met coal mines, like in Australia, where actually we are for safety reason ventilating heavily the different mine shafts. And the exiting air contains very diluted sources of methane, quantities of methane, which nevertheless they have a very strong impact on greenhouse changes and climate. So this is actually the 3 main categories. And there are a few other things, but they are very minor. So since last year, we really started to develop a plan, a pathway to really control our destiny and make sure we get there in time. Next slide, please. If we look at the pathway, it's really multifaceted. It's actually a set of, let's say, 6 building blocks, 6 levers that connected together will provide us the path towards carbon neutrality. The first one obviously is about generation of renewable energy. So it's really developing the capabilities for all the sites across the globe to get access to carbon-free electricity. And while we are deploying that, of course, we are integrating the second lever, which is the hydrogen. So producing hydrogen. You have seen a great project already in development for this year in Mogalakwena with Donovan. And actually here it's integrating renewable energy with hydrogen production to replace diesel by hydrogen and start operating our fleet, but when we speak about [ renewable ] energy, we always speak about intermittent sources. We speak about solar. We speak about wind. And you know that [ these sites, ore mines ] needs to operate 24/7, 365 days a year. So what we have done, we have been ideating. We have been really contemplating a series of options really to provide the necessary storage capabilities that will enable us to generate and store and dispatch energy 24/7 more to our operations. The fourth lever, we spoke about methane emission, is how we will capture that methane and neutralize it to avoid emissions or impact on greenhouse gas due to methane emissions. The fifth lever is an important one. It's energy efficiency. And I'm really connecting you back to what you have seen in traditional, I would say, technical innovation from Donovan but also all the digital progress made with Arun on that field where actually it's about producing more with less. So avoiding energy consumption, avoiding emission from the beginning; and also planning better the things in a way that we are emitting less. So it's an integral part, connected part of the journey we are going through. And last but not least, we know that there will always be, I would say, minor but difficult-to-abate emissions. So here as well we have embarked on a set of ideation process to identify nature-based, technology-based options to look at -- to capture and sequester CO2. So the -- and now we check the system and the architecture of the building blocks we have identified, and we believe we are in a very good path towards carbon neutrality. And on the next slide, I'd just like to briefly show you some of the progress we have made since last year. So if we start and focus on Africa: We have developed in South Africa now a detailed, I would say, comprehensive plan to convert all our sites, and we operate in 4 countries, to 24/7 renewable electricity access. We are now initiating the next phase, starting engaging very soon with stakeholders, with partners, with governments to make the next phase and start the deployment phase. Actually you see as well there integrated in that plant the hydrogen, so the hydrogen truck, with the demonstration and the first implementation happening by the end of this year in Mogalakwena. If we move west to Latin America. I'm also here very happy to announce that now the team has concluded a plan and secured renewable electricity access to all our sites, so in Peru, in Chile, in Brazil, from now on in South America. And it will enable as well, that's very nice, to have Quellaveco or a new mine site to start operating directly with renewable electricity. If we focus now on Australia: We are planning the same approach for renewable electricity, renewable energy to really convert our operations towards, I would say, clean energy. And specific to our met coal mine, we have already, I would say, developed a concept study; and identify potential solutions to, I would say, build. And we are now starting a prefeasibility study on a solution to capture and neutralize the methane coming from our ventilation process, with safety as a first priority and greenhouse gas abatement as a second priority obviously. And when you look at the plan with what we have already delivered, we are very confident that we are going to beat our target in time. And we are, so far, on track on the way. Thank you very much. And Tony, back to you.

Thanks, Pierre. We've always felt that strategic mine planning is an absolute key to understanding optionality of any mining business and the value of any mining asset, just how you set it up, the infrastructure; getting the scale and the options right. At a personal level and to be honest, I hadn't seen many examples of this work being done at the right level over my career. And I suspect Mark would have the same view. So we set about developing. This is a core competency within the organization, and capability. Then Matt Daley and Carmen Letton, with their teams, they have developed a suite of strategic plans that in -- that certainly in my view has been top class. It's been really pleasing, so I'd ask Matt. Please, can you take us through, if you like, some of the findings of your work?

Thank you, Tony. So today, we have talked about P101, our data-driven digital transformation with VOX3L, APC, new technologies and our carbon strategy. Now these individual programs are all critical to reimagine mining to improve people's lives, but when you combine and leverage these programs, you really start to see the full value and optionality this work can deliver across Anglo American's portfolio of assets. Next slide, please. So to that end, during 2019, we completed, as Tony mentioned, a series of resource development plans for around 10 of our assets or districts. Now these resource development plans are unconstrained, full-potential plans that incorporate many of the programs of work and technology we talked about today but, of course, in their 2019 state of maturity. Now this work illustrated that, in conjunction with our large and high-quality resource base, there is a potential for a 30% to 50% uplift in NPV from the group of assets that we reviewed. As you can see from the pie chart on the right of the slide, the NPV uplift is driven by a combination of P101, technology development and conventional expansion. Now resource development planning has continued through this year using the latest work that was presented today, and compared to 2019, we are seeing innovation and technology overtake conventional expansion as the preferred pathway to value. So this new pathway is adding even more value earlier at lower capital cost; and importantly, with an improved footprint from a sustainability standpoint, less water, less energy, less carbon. The new preferred pathway is also providing significant, new optionality to our portfolio. It has allowed us to look differently at timing and scale of growth opportunities, asset life extensions; change the thinking on how we should deploy capital and what regional or greenfield projects will be the most value accretive. We strongly believe, and the results are reinforcing that belief, that today's work programs will deliver a different and better future for mining. On that note, back to you, Tony.

Thanks, Matt. So ladies and gentlemen, that pretty much ends our presentation today. I hope it's given you a new look, an up-to-date look of where we are in our technical and innovation space and the linkage that has with our purpose and our sustainability programs. And with that, I will hand over to Paul Galloway, who will look after questions. Thank you.

Thanks, Tony. I hear we had a couple of teething problems at the start there, so clearly investor relation needs to adopt an operating model and a technical upgrade. So apologies if that disturbed anybody. We'll now go to the Q&A. [Operator Instructions] And just to remind you: We also have Stephen Pearce, the Finance Director, on the call to help us as well. With that, if I can ask the operator for the first question, please.

The first question comes from the line of Sergey Donskoy.

Yes. Can you hear me?

Yes, we can hear you.

Yes. I have 2 questions, if I may. Is it possible to quantify, speaking of conversion of your truck fleet to hydrogen, what sort of amount of PGM you will need to convert your own truck fleet across all operations? And what electrolyzer capacity will be required to achieve that as well? And second, it's kind of also, I think, related to technology, although in a somewhat different way. It's not about efficiency improvement. It's about production. Do you think about investing or expanding in DRI, production in your iron ore divisions at Minas-Rio and at Kumba to capture more of the value that some other may reap if you don't do that?

[ Well, Tony, did you get the question ]...

Yes, we've got that, yes. Thank you very much. So Donovan, can you pick up the first part of Sergey's question, on the hydrogen? [ And we'll get ] Tony on the second part, on the iron ore. Thank you.

Yes, certainly. So I think the first thing to say is that the fuel cells that are currently in use, use quite a bit of platinum, but if we rather just have a look at the U.S. Department of Energy suggested amount of platinum that they're aiming for into the future, it talks to around about 4 to 8 ounces of platinum per truck. And when we have a look then at the rather -- at the electrolyzer portion, which also uses platinum, we're looking at around about 31 ounces per truck. So it really just depends on the number of trucks that we're rolling out. So for instance, at Mogalakwena we'll be rolling out 40 trucks.

I see. [Audio Gap] Minas-Rio infrastructure constrain downstream. We continue to look at the options, but the rail at Kumba and, if you like, the pipeline at Minas-Rio are not short-term fixes. They would require, if you like, a multiyear program. We continue to look at them. I know that the crew at Kumba certainly talk to Eskom regularly about rail capacity, but there are no short-term, if you like, fixes in train at this point.

Sergey, is that good?

Yes, quite good.

Next question comes from the line of Ian Rossouw.

Yes. I had a couple of questions just on the sort of expansion options, particularly at Collahuasi. You're obviously looking to approve plans over the coming year. Does this -- some of this technology make you want to rather pull some of these expansion options, maybe not just related to Collahuasi but others as well, and then sort of try to integrate as much as possible into these before you make the decision? And then maybe just related to that, on Quellaveco: Obviously you've had some time now, I guess, due to COVID, I guess, a pause in the construction at Quellaveco. You've obviously added some of this technology to that as well recently with approval of the project. And I just wanted to get a sense of what do you think the ultimate sort of savings on power, water; and maybe just an idea of the longer-term production profile, how that changes versus the original approval guidance.

Thanks, Ian. Okay, Matt, can I ask you to take the first question, on the Collahuasi expansions? And then Tony perhaps on Quellaveco.

Yes, certainly. Thanks, Paul. So really good question on Collahuasi. The technologies you saw today have a really high level of applicability for Collahuasi, concentrate the mines, so bulk ore sorting, coarse particle flotation, hydraulic dry stacking. We are studying these technologies at the moment. Near term for us, Collahuasi opportunities sit in the debottlenecking of the existing plant. We have the mine ability to feed and expanding the capacity in the plant. So first prize, debottlenecking; second price, certainly moving towards implementation of technology. Currently working very closely with the other joint venture holders at the moment [ enable to ] position this technology as being the pathway for expansion for the asset.

But could this potentially delay some of those expansion decisions, do you think?

The -- probably the pathway is really going to be around permitting. So at the moment, there's nothing committed beyond the near term, which was the addition of a fifth ball mill. So that project will go ahead. Anything beyond that, at the moment, we have full flexibility in terms of what we implement. So near term, debottleneck, fifth ball mill; beyond that time horizon, complete flexibility on how we take the asset forward.

Okay. Quellaveco...

Next question comes from the line of Richard Hatch.

Yes. I just had a question on electrification of fleet. I just wonder whether you might be able to talk a bit about electrification of drills and such like and whether you see an opportunity there. And then sort of secondly, on South Africa, I just wonder whether you might be able to give us any sort of steer or guidance over the time frame for kind of sorting out your own power kind of solutions there. And then just finally, just I wonder if you might be able to give us any kind of numbers around CapEx, as to kind of what kind of investment is required to progress this sort of net 0 and such like targets. Obviously group is generating a lot of free cash. Isn't now the time to start putting a bit more into these technologies when you've got the money and it's flush?

Matt, do you want to try and take that one for me?

Yes, certainly. So on the electrification front, 100%. So if you look at our large rope shovels, they're already electrified. If you look at a lot of our large production drill rigs, take Mogalakwena for example, about 80%, 90% of their fleet is already electrified. It's always a trade-off between the value proposition of being electrified and the less mobility you have by having cables through your pit. As we transition into our large modern undergrounds, examples being at Venetia, we're very much moving to an electrified future for that asset. If you go forward with our plans in terms of acquiring the next fleet as we move into production, the intent is to have everything that we can electrified over the next kind of 3 to 5 years. So very much the trajectory. In terms of hydrogen, also looking at options away from the trucks, into smaller fleet. So I think there's very much a future where electrification either through having [ pure fed by cables ] or hydrogen on drills and on support equipment like dozers is very much in our future.

Okay. Look, I think we missed out part of Ian's last question as well, so Tony, I think there's 2 with you, Quellaveco and the second part of this one, please.

Ian, the commercial terms that we already had in place for Quellaveco on power were quite competitive. And then moving to the renewables, I don't think there was a big financial benefit at all, but it enabled us to, if you like, have a green mine from get-go. And that was something that we were very keen on. In terms of coarse particle recovery in -- at Quellaveco, it's at the back end of the process, whereas you get the real benefit from coarse particle recovery in a water sense at the front end. So there isn't a particular -- a saving in water at Quellaveco. The permit was always limited by water. The current mine is permitted to 127,000 tonnes a day. I think the permit, as it stands, limits us to 150,000 tonnes a day at some point in the future. Our aim, though, given where we are, is to get the thing up and running and stable as quick as we can, particularly with COVID; and then look beyond, if you like, pretty quickly from there towards going towards 150,000, but at this point, 127,500 is the guidance.

Richard, have we answered your questions?

It -- you got the first one. I think the second one was just on, firstly, time frame of kind of green power in South Africa and kind of transitioning away from Eskom. And then just a question on the capital piece. Just is now the time to start putting more cash into it given the fact you're generating quite a lot from strong commodity price?

Very good. Thank you. Tony?

Those -- the -- that's a good question. Look, we're at a point where we're about to start discussions with the government and our local communities. I think for us to put a time line around that at this point is a bit presumptuous. They may or may not like what we put to them. We think the plan is solid. It certainly looks very workable to us. It's basically a wind power at East and West Coast, renewables and then storage of energy in water. So some -- it's a well-thought-through concept that hangs together. Hopefully, we'll get some sense of people's reaction to these over the next 2 months. And I think, after that, if we've got, if you like, a direction from the government in particular, I think, at that point, Mark or somebody can talk to the topic.

Richard, all good?

Yes. Well, I guess we'll just have to wait for that decision. And then you can decide whether now is the time to start spending more generally on it.

Yes.

Next question comes from the line of Myles Allsop.

Great. Just maybe a question for Steve thinking about the CapEx. Richard touched upon it, but I don't think we really answered it. And OpEx implications of all the technology and sort of decarbonization of the business. How much CapEx are you going to be spending each year? Do you rely on subsidies? And does it impact OpEx? Is it increased OpEx like we hear from a lot of companies? Or are there options to use this to reduce OpEx? And then maybe just secondly, on your savings, your EBITDA, net EBITDA, of $800 million, what are you net of? Are you net of cost inflation? Or is it just net of the sort of the costs of implementing?

Stephen -- yes.

Paul, I'll jump in, if that's all right, yes. Thanks [ for that ]. On the CapEx, as Tony touched on, in terms of, say, a program like power in South Africa, we'll come back to you once we work out the details and have the discussions with the government. The main things that will influence that is, well, who may we partner with and how much would be on our balance sheet over what time versus others, et cetera. And so we'll guide you as those plans firm out. In terms of things like, say, Donovan's area where we're doing the R&D work and some of the early-stage rapid turnover of trials, et cetera, that's all included in our capital guidance that we give across those 4 [ years ]. And our best estimates of the programs that the team have put together that would be executed over the time line that we give the capital guidance is also included in those numbers. As often said, depending on the initiative and the site, again I'll try to encourage people to think about the benefits that come with a lot of this technology, not just the costs of implementing. So obviously it's about recovery at Quellaveco as we put that plant in. It may be about gray water or energy savings at El Soldado, those sort of things. So these programs do have a very rapid payback in the majority of cases as we look to implement, and so they are really quite positive from an NPV perspective. And it's those numbers you'll see that drop into the EBITDA. Yes, we need to deliver these numbers to help offset the impacts of inflation, for example. So when we put out the original $3 billion to $4 billion target to 2022, we always said we needed to hit that sort of number because we are likely to give about half of it back in inflation over time. And so that sort of is the way that we've constructed the numbers. I hope that helps, Myles.

It's great.

Next question comes from the line of Sylvain Brunet.

My first question is related to both Slide 14 and 17, on the $3 billion to $4 billion EBITDA improvement target. If you could just remind us if this was based on 2017 kind of prices at the time and whether we're right to think that the delivery in the coming year and next year will be essentially accelerated. Because if we look at Slide 14, you delivered $2 billion over 3 years, and effectively the target would be $2 billion over 2 years. And my other question, if I may, is on automation, which wasn't mentioned much. I know it used to be a bit more fashionable in Australia a couple of years back. Is it the case that this is not applicable to your operations? Or do you think the cost of implementation is not compelling enough?

Thanks, Sylvain. Stephen, would you like to pick up the first part? And Tony perhaps the second, please.

Yes. Thanks, Paul. Sylvain, yes, you are right. The delivery of the benefits does accelerate through the balance of '21 and '22. We've always said that, particularly in relation to some of the technology and innovation, as the team have been progressively implementing and rolling out the initiatives that they've discussed today and with our forward program -- then yes, '21 and '22 are very important for the technology and innovation benefits and for the volume benefits, particularly as we deliver Quellaveco in -- through 2022. I think the message that I'd like you to take away from the team's presentation today is, yes, we are confident we will deliver those targets, but there's lots more to come. And we're already -- yes, so that's both in the years beyond 2022 as we continue to roll out next generations and modules, et cetera but also as we get that stability that Dave spoke about as well. I mean I think it's a combination of those 2 things that gives us a lot of confidence in the future as we go forward.

Sylvain, with respect to automation, I think, firstly, Quellaveco mine is actually autonomous. So it's something that we are very interested in. Clearly, as a company in the jurisdictions that we work in, employment and our relationship with our host communities is really more important than automation. So it's an area that we will be ready to move into. Certainly the equipment we buy is autonomous capable and will be going forward. And if we get the right relationships and particularly employment numbers, and outside the gate will be the key to that, then we will push them harder, but I think we have circumstances that aren't like Australia, for example. And you mentioned it. We will be capable, and if we have the right approvals and blessings, we'll go forward with it. In saying that, we've also looked really hard at the performance of a lot of people who have put automation in. There have been some really outstanding success stories. And that's why we're interested in it, but we've also seen quite a number of players who haven't put it in very well at all. So -- but as I said, it's part of our armory going forward if we choose to deploy it.

Next question comes from the line of Sergey Donskoy.

Yes. Just one follow-up on maybe Slide 49 I'm looking at right now, just to understand: When we speak about gross improvement and net improvement, this $3 billion to $4 billion target, with $1.4 billion coming from operating model, P101; and $1 billion coming from technology and innovation, just to make sure, these are gross improvements, or net improvements?

Stephen, go ahead.

Gross -- [ so it will be ] gross of cost inflation. Obviously they're net of headwinds. And the sorts of either timing implications because things are caught up in working capital and haven't been released to EBITDA or because we have seen instability in the plants. So it's a net result that drops to the EBITDA line but before inflation. Hopefully, that makes sense. Can I just add also to that value driver point in terms of automation? Obviously, given the scale in Australia, particularly in places like the Pilbara, buffer limited processing they do, the benefits they get from automation and in fleet automation in particular are probably more important and material for them. In our case, we're mining different things, in a lot of instances, besides iron ore. Then a lot of the benefit we get from automation is in processing, in control and stability through the other parts of the process. And that's, hopefully, again you've got that sense of the [ team ], but that's the value driver that really drives the bottom line for us. Thanks, Paul.

Next question comes from the line of Richard Hatch.

Yes, just a quick one on greenfield growth. So I mean sort of there's quite a few greenfield growth opportunities in the portfolio: PGMs, Twickenham, Der Brochen, Sakatti up in Finland. You've got Chidliak over in Canada, Woodsmith in the U.K., but how are you kind of thinking about all of these different technologies on the greenfield kind of projects? And are there any sort of ones which you'd kind of draw out and say you really can see a great kind of opportunity by implementing some of these technologies?

Thanks, Richard. Tony, can you do that, please?

Thanks, Richard. I mean the applicability of the technologies that we're talking about have pretty much got universal application. I think we've given some clues throughout the presentation, but the -- in platinum in the northern limb there is a lot of interest in how we transform that business, in copper. At basically the major assets there is also a clear application of these new technologies and thinking. So I think around all the assets we actually see opportunity, but probably the major, major ones at this point would be around the northern limb and Collahuasi at the right time.

There are no more questions at this time. Please continue.

Okay, everyone, well, thank you very much indeed for your time this afternoon. It did take longer than we would have normally done these sessions, but we wanted to give good exposure to the team and to let you know all the good work and hard work that is going on within the group. As ever, if you've got any follow-up questions, then please contact myself or Rob Greenberg or Emma Waterworth. With that, thank you very much indeed. And have a very good afternoon. Go safely.