Mineral Resources are Abundant
This week, I am continuing with the theme of sustainability, particularly as it relates to a rising population and standard of living. This week’s topic is on the depletion of natural resources, and whether “running out” of critical natural resources is a threat to civilization’s sustainability in the foreseeable future (the post’s title gives away the answer). I will review some basic concepts related to mineral availability and why pessimistic forecasts of depletion are generally wrong. Then I will apply the principles to forecasts for copper depletion.
Speaking of sustainability, I regret to say that, at least for the time being, I will be slowing my writing pace to one post every two weeks. Writing these posts is one of the highlights of my week, and I am pleased that in nearly five years, I have not missed a weekly post except when traveling or hospitalized. But with the posts having gotten longer and more heavily researched, and with various other professional commitments, I cannot keep up this pace of writing anymore and fulfill my other obligations.
As for this sustainability series, my tentative plan is for two more posts. In two weeks, I will consider whether existential risk and global catastrophic risk is a threat to the growth of civilization. In four weeks, I will conclude with an examination of theoretical limits to scaling.
Understanding Reserves
The most important point to take from this section is that “reserves”, “resources”, and the “ultimate recoverable resource” (URR) are not the same thing. But because the distinction between these ideas is frequently misused in the literature, we need to disentangle them.
Meinert, Robinson, and Nassar (2016), working for the U. S. Geological Survey, are among the very many authors who explain the difference. Briefly,
Reserves are those mineral deposits which are known to exist with a high level of certainty and which can be recovered economically with today’s technology and prices.
Resources are those deposits which are believed to exist, but not necessarily with high certainty, and which may or may not be recoverable economically with today’s technology and prices. All reserves are resources, but not all resources are reserves.
The ultimate recoverable reserve (Meinert, Robinson, and Nassar (2016) use the term “all there is”) are those deposits that have been or will eventually be recovered, whether or not we currently know about them or can recover them economically with today’s technology. All reserves and resources, and all of the mineral that has been harvested in the past, are part of the URR, but the URR is generally much more than this.
For a more detailed delineation of reserves, see Brobst, Pratt, and McKelvey (1973), who introduced a nomenclature that is still in wide use today in the USGS, equivalent agencies in other countries, and in the private sector, albeit with variations. Vincent McKelvey was the director of the USGS from 1971 until he was pushed out by the Carter administration in 1977. McKelvey was a noted optimist on mineral resource availability and a bit of an ideologue, and these factors put him into a long feud with M. King Hubbert, of Hubbert’s peak fame, as recounted by Priest (2014). McKelvey lent his name to the McKelvey diagram for classifying minerals, a modern rendition of which is below.
All recoverable minerals are classified as identified or undiscovered. The identified minerals in term are divided into demonstrated, which have been at least partially observed and quantified, and inferred minerals, which have not been observed but can reasonably be expected to exist based on knowledge of geology. Demonstrated minerals are, in turn, subdivided into measured resources, which have been measured with enough precision so that the amount of the reserve can be stated with high confidence, whereas indicated resources have not been measured with as much precision, but enough to offer a reserve estimate with reasonable confidence. Undiscovered minerals are also divided into two groups: hypothetical minerals are those that might exist in regions that are unexplored but have similar geological characteristics to those with known minerals, and speculative resources—which may be analogized to Donald Rumsfeld’s unknown unknowns—are those in regions that are neither explored nor understood well geologically. Different agencies will offer different numbers to make all of these concepts precise.
The McKelvey diagram is not static, in that as exploration proceeds, minerals can move from speculative to hypothetical to inferred to indicated to measured, or they can fall from the box entirely if exploration demonstrates that they do not exist. Similarly, minerals can move between the subeconomic and economic categories (usually from the former to the latter) as technology and market conditions change.
Wellmer (2022) explains more deeply the economics of how reserves work and how they are regulated by market mechanisms. As a rule of thumb, in a healthy market, reserves should be at least 25 years of annual production or 50 years of annual production, depending on the geologic nature of the mineral. If reserves are too low, there is a market signal for the industry to explore more and boost reserves. The industry will not explore if reserves are too high, since it is not economically sensible to invest in exploration for a product that can only be produced in the far future.
The metric used here is the R/P (reserves to production) ratio, which is the length of time that reserves will last at current production rates. Because of the market regulatory effect, R/P tends to remain constant over time, despite depletion and increasing levels of production. For this reason, R/P tells us nothing about how long a resource will actually last. And yet, too often do depletion forecasts treat R/P as though it does. Wellmer (2022) observes that The Limits to Growth (Meadows et al. (1972)) makes this mistake. To pick one of many examples of this error, Scholz and Wellmer (2021) observe that The Limits to Growth reported an R/P of 23 years for zinc, using 1970 values, so that a naive extrapolation would see zinc fully depleted in 1993. By 2020, R/P for zinc was 21 years. However, over that time, world zinc production increased from 5.6 million tons to 12 million tons, and so in absolute terms, zinc reserves doubled over 50 years.
Forecasts that treat reserves or resources as the URR are almost always too pessimistic, and yet this rookie mistake continues to be repeated, as we will see with our case study of copper.
Peak Copper?
Copper is one of the oldest natural resources to be exploited. Pompeani (2021), using carbon-14 dating, estimates that copper working commenced around the Great Lakes region around 7500 BC, inaugurating the Chalcolithic (Copper Age).
After millennia of production, Northey et al. (2014) project a world peak in copper production in the late 2020s or early 2030s, with production nearly ceasing by 2100. The forecast is based on several runs with different assumptions of their Geologic Resources Supply-Demand Model (GeRS-DeMo), which had been developed a few years earlier and applied to lithium, phosphorous, coal, oil, gas, and various other minerals. The model consists of a supply module, which entails mines opening and producing at a constant rate for a number of years, with ramp-up and ramp-down periods; and a demand module, which is based on population and GDP projections with historical values of copper demand per capita at various per-capita GDP levels. The model also allows for interaction between supply and demand, so that unmet demand can induce miners to produce earlier than they would with lower demand.
However, and here is where the major flaw is, the model uses a static value of the URR. It is based on the survey Mudd and Weng (2012) of copper reserves, with some additions between the publication of the two papers, and also additions allowing for improvements in the milling recovery rate—the fraction of ore that is retained as valuable product after processing. The resulting world URR is 2097 million metrics tons (Mt) of copper. This is close to the 2100 Mt of copper reserves estimated by Johnson et al. (2014), a fact sheet compiled by the U. S. Geological Survey. However, that same USGS report also estimates an additional 3500 Mt of undiscovered resources, yielding a URR of 5600 Mt, far more than the value used by Northey et al. (2014). In other words, Northey et al. (2014) conflate known resources with the URR.
Declining Ore Grades
A seemingly conclusive piece of evidence from Northey et al. (2014) to support a near-term depletion hypothesis is declining ore grades. The grade of an ore is the proportion of the total mined material that is the desired mineral. Mudd and Weng (2012) find that the world average copper ore grade was 0.62% at the time, meaning that for every ton of earth processed, 6.2 kilograms of copper are recovered. If grades are lower, than a larger amount of earth must be moved to recover a given amount of product. For example, Calvo et al. (2016) find that from 2003 to 2013, average copper ore grade declined by 25% in Chile, and the energy required to produce a ton of copper increased by about 12%.
But Rötzer and Schmidt (2018) challenge the common narrative that declining ore grades are a result of depletion. Instead, they argue that declining ore grades are a result of increasing production. High grade deposits are still in commercial production, and they find that low grade deposits are being mined in addition to, not in replacement of, high grade deposits. Thus with increasing production, average grade can decrease without high grade deposits declining.
Better technology has converted low grade deposits from what would have been considered useless overburden in 1900 into valuable ore by 2000. Rötzer and Schmidt (2018) argue that the Jackling process of open pit mining, which enabled the economical production of lower grade ores, was the most important development in the 20th century for copper mining. Conway (2023) has more detail on the Jackling process. Rötzer and Schmidt (2018) also highlight the froth flotation process of separating copper from the rest of the material, improved mechanization, and various other advances.
A Near-Term Peak is Unlikely
To illustrate how greatly the failure to treat reserves as dynamic can skew a forecast, let us consider another application of GeRS-DeMo from Giurco et al. (2012). From the abstract of that paper,
The projections (based on current estimates of ultimately recoverable reserves) indicate that peak production in Australia would occur for lithium in 2015; for gold in 2021; for copper in 2024; for iron in 2039 and for coal in 2060.
The forecast peak for lithium was just three years after Giurco et al. (2012) was published. Here is lithium production in Australia through 2022.
The gold projection looks a little better, with a peak in 2020 as of 2023.
More recently, Guj and Shodde (2025) assess the copper reserve at 3533 million tons. Since 933 Mt had already been produced, this gives a URR of least 4466 Mt, more than twice the value of Northey et al. (2014). Guj and Shodde (2025) find that there were about 850 Mt of resource growth from 2010 to 2023, of which about a third came from deposits discovered after 2010 and the rest from deposits discovered before. How can reserves in deposits discovered before 2010 continue to grow? Because as mining technology advances, some ores that were not previously considered to be resources are upgraded. I should mention that the pre-2010 data is based on a sample of 52 deposits, not all deposits, and so those numbers are incomplete. How incomplete they are, I cannot say.
It does not appear that geologic constraints will cause a peak in copper production in the 2030s, or probably in the next few decades after that. But Vershinina and Rollin (2024) note that growing concerns about environmental and human rights issues around mining, a set of issues that are collectively labeled as “environmental, social and governance” (ESG), may be a limiting factor. The degree to which ESG concerns, or good old-fashioned NIMBYism, will or should inhibit copper mining is beyond the scope of this post, but I do think these issues are part of a shift in the understanding of limits to growth being imposed by environmental degradation rather than resource scarcity.
Copper’s Bright Future
Concerns about political challenges notwithstanding, there are many reasons to expect copper abundance for decades to come.
According to the International Energy Agency, world copper demand was 26 Mt in 2023 and is forecast to grow to 36 Mt/year by 2040, driven in part from the need for copper for electrification in a low-carbon world.
Wang et al. (2021) estimate that expanded recycling could reduce copper production’s energy needs by 15%, so that would increase recycling from 40% to I guess around 50%. They also see potential to increase yields (the ratio of pure copper to finished product) of copper semis by 5 percentage points, or 7.4% more copper semis from a given input. They also see potential to decrease copper demand through material efficiency and substitutes for some uses.
For McKinsey, Crooks et al. (2023) identify several emerging technologies which will further create copper resources and convert resources into reserves. They include better flotation, grind-circuit roughing, coarse particle scavenging, sulfide leaching, and AI process control. The report argues that these technologies will increase copper production by 20% over what it would otherwise be.
Then there are wildcard technologies. Petersen et al. (2016) assess that nodules in Clarion-Clipperton Zone in the Pacific Ocean could provide 226 Mt of recoverable copper, or about a quarter of present terrestrial reserves, and there may be 5000 Mt of recoverable copper overall on the ocean floor. Deep sea mining remains controversial and does not yet occur on a commercial scale, but Zhang et al. (2024) document progress toward that.
Diallo, Kotte, and Cho (2015) highlight technological developments that may allow recovery of copper and other metals that are dissolved in seawater. The mass of the oceans is, give or take, around 1.37 quintillion tons, and the oceans contain 0.09 parts per million of copper. That works out to 123,300 Mt copper, dwarfing all estimates of the terrestrial URR. Finally, Łuszczek and Krzesińska (2020) analyze the chemical composition of chondrites, a type of meteorite that originates from S-type asteroids, and from them estimate that the asteroid 6 Hebe, which contains 0.5% of the mass of the asteroid belt, contains 1,139,000 Mt of copper. Asteroid mining is not feasible at present, though.
The Abundance of Natural Resources
To my knowledge, the world has run out of only one mineral resource. Cryolite, chemically Na3AlF6, is used in the Hall-Héroult process to dissolve aluminum from bauxite. However, the mineral is rare on Earth, and the only commercial mine was the Ivittuu mine in Greenland, which closed due to depletion in 1987. Now only synthetic cryolite is used for aluminum smelting.
There is an apparent paradox between the obvious finiteness of the Earth and the fact that no major mineral has run out or appears to be in imminent danger of running out. Predictions of resource depletion keeping come back because it seems evident that at some point they must come true. But unless the forecast properly understands reserves, resources, and the URR; and it models economic feedbacks to exploration and production; and it accounts for technological development, then it is best regarded as one in a long time of foiled forecasts.
Quick Hits
In last week’s post on planetary boundaries, I stated the following,
Thus civilization is contained in the Holocene epoch, a relatively stable climactic era with conditions favorable to agriculture as we know it.
A reader questioned whether the Holocene climate has in fact been stable, at least by the standards of previous epochs. The assertion that the Holocene climate has been stable was made in the main paper, Richardson et al. (2023), and perhaps I should have thought twice about taking this claim at face value. Wikipedia documents many major climate changes in the preindustrial Holocene, including the 8.2 kiloyear cooling event; the 4.2 kiloyear event that brought droughts in some parts of the world, flooding in others, and has been implicated in the collapse of civilizations at that time; and the transition of the Sahara into a desert. I do not know how all this compares to previous epochs.
It cannot be stressed enough how central the population issue was to the early environmental movement. Known for signing other major pieces of environmental legislation, President Richard Nixon established the Population and the American Future commission, chaired by John D. Rockefeller III. The commission suggested what we should recognize by now as the usual population control measures, including antinatalist structures in the tax code, immigration restriction, and legalized abortion. However, by 1972, for a variety of reasons Nixon had cooled to the idea of population control and gave the commission report a tepid reception. Hoff (2010) tells the story of how this happened, and more broadly, how the population control movement declined in the early 1970s.
One thing I find especially interesting about Hoff (2010) is how the Nixon administration shifted the commission’s focus from overall population numbers to the distribution of the population. The distribution question consisted of two pieces. First was the “urban problem”, the subject of the Daniel Patrick Moynihan’s controversial 1965 report, The Negro Family: The Case for National Action. Second was the related issue of white flight from city centers and the resulting urban sprawl, which came onto the radar of the environmental movement at around that time. Thus, while urban sprawl and overpopulation are clearly two different concerns, the connection between them is more than superficial resemblance.
Nandagiri (2021) describes the problem of “voluntary” family planning as part of a development or population control agenda, and she calls for a decoupling of these two things. She says what I have tried to say before: there is a contradiction in terms if “empowerment” is used to support some broader social agenda.
Robert Zubrin makes the case for a renewed sense of purpose at NASA and a focus on a crewed mission to Mars.
The video game industry is facing its most severe downturn since the 1980s. In an hour-long analysis, the YouTuber NeverKnowsBest posits why this is happening. He also explores three non-reasons: greedy corporations, a public backlash against “wokeness” in gaming, and a declining in game quality.
Again, I am not planning a post for next week, and I will see you in two weeks.