Guojin Hwen Ask Non-ferrous Metals: Supply Contraction and AI Demand Resonance, Non-ferrous Sector "CommodityStock" Value Transmission in Progress.

This year, non-ferrous metals have undoubtedly been one of the hottest sectors. Looking at the stock price changes, as of October 14, 2025, they have already risen by over 70%, ranking first among all industries. So, what are the core drivers of non-ferrous metals this year? From an industry perspective, it does have solid intrinsic logic support such as supply contraction, but a single industry logic alone may not be able to determine the overall trend. From a macro perspective, top-down industry comparisons play a crucial role, but it is easy to overlook many details at the industry level. Therefore, as the first part of the dialogue between the strategy team and the industry team, we will raise questions from different perspectives and mutually answer the confusions about non-ferrous metals. The audio of the conversation will be available on Jin Men Financial. The following four questions are the responses of the strategy team to the questions raised by the metal team: Metal Team: How long will the restocking market last? When will the investment style start to change? Strategy Team: Looking at the inventory cycle, there was a restocking market in the United States in March-May 2024. Although it did not transmit to upstream resources at that time, industrial metals represented by copper have both industrial and financial attributes, making them sensitive to signals for downstream product restocking. Prices of upstream resources also showed a significant increase during this period. However, in terms of sustainability and elasticity, this round of restocking in the US was relatively weak overall (manufacturers' inventories started weakening in June 2024), mainly due to the suppressive effect of high-interest rates on the manufacturing industry. At that time, the US economy was in a phase where manufacturing and service sectors were mismatched, with a differentiation pattern of "strong service sector, weak manufacturing sector," where the strong service sector led to persistent service inflation, while the interest rate-sensitive manufacturing sector was weak due to interest rate suppression, leading to a decline in commodity inflation. However, once entering the restocking rebound cycle, it is easy to trigger expectations of policy tightening, which, in turn, limits the strength of restocking. However, the sustainability of this latest restocking cycle may exceed market expectations, closely related to the changing logic of the current US economic structure. Unlike before, the service sector is showing signs of weakening, indicating late stagflation or even possible recession; meanwhile, the manufacturing sector is in the early stages of recovery, with the latest data showing no further increase in price indices and even a marginal decrease, indicating a marginal improvement in manufacturing activity. With the onset of the Fed's interest rate cuts, the manufacturing demand suppressed under the high-interest rate environment may gradually be released, and thus, we may see the start of downstream restocking cycles in the future. Compared to the weak restocking market in 2024, it is expected that this round of restocking cycle will not only start quickly but also last significantly longer. Of course, the main reason for the deviation between manufacturing and service sectors is the shift in fiscal policy towards social welfare and transfer payments in overseas countries over the past few years. However, this orientation is undergoing significant changes now, as major European and American countries are beginning to shift towards encouraging business investments or direct government support for military, infrastructure construction. In terms of investment style, we are more inclined to believe that we have entered a rebalancing process. We focus on two aspects. Firstly, after the start of overseas interest rate cuts, the transition from inflation sensitive commodities to service inflation through the "re-inflation" process may have a potential impact on both cyclical profit expectations and changes in liquidity environment. Secondly, for domestic investors, the earnings indicators of A-share listed companies are more crucial, one of the core indicators being China's export price index. Currently, the PPI might follow the same pattern as in 2005-2006, where external demand led to stabilisation, resulting in orders for export products increasing in price, causing export sellers to replenish raw materials, leading to stable domestic bulk prices and delayed PPI recovery. This "demand abroad, inventory first" transmission sequence makes export prices a leading indicator for the entire chain. At present, signs of recovery are seen in overseas equipment demand and the "anti-internal consumption" movement is continuing to advance domestically. With structural optimisation under domestic supply constraints combined with recovering overseas demand, the sustainability of Chinese corporate profits is something to look forward to. It is expected that in the next 2-3 quarters, overseas markets may welcome a "re-inflation trade", leading to a wider market style switch. In the early stages of commodity inflation rebound, the market style should be in a rebalancing phase. Metal Team: How to understand the macro-level driving logic of "resources manufacturing technology"? What leading indicators can indicate turning points in the non-ferrous metals market? Strategy Team: From a stock market investment perspective, assets that have been prominent for a period of time usually correspond to sectors in the GDP that are easier to expand or are more profitable in the distribution of listed company profits. Additionally, looking at stock price patterns, given that 60 trading days equate to one quarter, which is also an earnings cycle for companies, it often refers to their long-term slow change or short-term drastic change levels that can exceed the 60-day moving average. Therefore, this cycle can be an important reference for observing sector rotation. Specifically, in terms of resources, manufacturing, and technology: Resources: As consumables for corporate production and operations, resources are directly linked to production, investment, and construction activities. When companies lean towards production or investment in construction, resource consumption significantly increases, as was seen in the upward cycle of the new energy industry in 2020-2021, such as lithium carbonate during the new energy industry's upward cycle. Manufacturing: Highly dependent on resource supplies, manufacturing may see simultaneous price increases when resources are sufficient and when downstream demands are positive. When the manufacturing industry faces supply clearance stages, or when the industry belongs to emerging sectors (supply expanding rapidly but demand growing even faster), the manufacturing sector may exhibit considerable excess returns. Technology: When the middle manufacturing lacks supply expansion and demand uptrend logic (due to slow down in production, investment, and construction), if the technology sector sees technological breakthroughs, it tends to become the market focus and may gradually become an economic growth point. However, some technological industries might transition to "technology manufacturing" after they mature. In this process, the phase matching between cycles also exists, such as the capital expenditure cycle used frequently in industry analysis. Typically, the mining sector's production capacity construction takes longer than manufacturing, further influencing the matching of supply and demand. Looking at the domestic market, from 2023 to 2024, upstream resources and dividend sector stock prices performed better, mainly because China's electricity consumption growth exceeded that of industrial value-added production, resulting in an energy consumption-output disparity, where energy consumption growth outpaced output growth, squeezing the profit margins of manufacturing companies. During this period, the power system directly benefited from the growth in energy demand, followed by the non-ferrous metals sector, although the summer rate of sharp non-ferrous metals stocks in the domestic market was not as high. From the overseas perspective, due to the weak manufacturing activity caused by the high-interest rate environment, there was less demand for resources as manufacturing revived, causing the non-ferrous metals sector to be less favored by the market. It was only in the second and third quarters of 2024 that this trend changed: dividend assets such as coal and electricity started to decline as domestic manufacturing companies began purging operations, and some corporate profits began to stabilize. However, due to the slow pace of the purging process, the sector lacked elasticity. On a global scale, a different scenario was seen: the US, as a service sector-led economy, experienced consistent growth over the past two years due to the differentiation pattern of "strong service sector, weak manufacturing sector". Because US tech company profits are highly correlated with service sector prosperity (the US service sector is actually the profit for tech companies), and coupled with breakthroughs in US tech sectors, various factors combined to drive the continuous profit growth of US tech companies. This trend eventually transferred to the A-share market, where during a time when the domestic economy was in a phase of slowing resource and manufacturing growth and weak demand, the impact of overseas tech trends was reflected, leading to highly coordinated growth in A-share technology and overseas tech sectors resulting in significant excess returns during this period. The future turning point may see the renewed rise of global resources, driven by the upward recovery of the global manufacturing cycle. It is recommended to focus on manufacturing PMIs or the ratio of manufacturing PMI to service PMI. Taking the copper-gold ratio as an example, historical data shows a highly positive correlation between global manufacturing PMI and the copper-gold ratio. Currently, the value of the copper-gold ratio is significantly lower than during past periods with similar PMI levels, indicating that with an upward shift in manufacturing activity, industrial metals represented by copper may demonstrate higher recoverable elasticity. In the medium term, global manufacturing investment activities are expected to launch in full swing. Our focus should not be on when the non-ferrous metals sector will have a downward turning point, but rather on when the resources will lead the upward turning point. Considering that commodities often rise due to supply-demand contradictions, when upstream price increases significantly squeeze the profits of middle and lower companies, it signals that the economy has transitioned into a stagflation stage. From an overseas perspective, when commodity inflation spreads to service inflation, it will trigger stronger expectations for policy tightening or demand suppression effects. This may indicate that a downward turning point for resources is near. Metal Team: From a demand perspective at the macro level, what changes could affect the demand for industrial metals represented by copper? Strategy Team: Looking at the current economic environment, while global economic resilience has been strong in the past 2-3 years, it has been mainly service sector-led, with manufacturing investment activities relatively subdued. Under these circumstances, even though the global economy continues to grow, weak manufacturing investment activity has resulted in relatively low physical consumption per unit of GDP, meaning that the demand for metals lacks strong support. However, this situation may undergo significant changes in the future: with global central banks initiating loose monetary cycles, the global economy is expected to shift towards a manufacturing-driven logic, with the global manufacturing cycle gradually recovering over the next 6 months. As manufacturing activities outpace service sector activities, there is an increase in total resource consumption per unit output. According to our estimation, in years when manufacturing activities are strong, the average resource consumption per unit output is 4.02% higher compared to years when service activities are strong, leading to an average additional resource consumption of $1.09 trillion per year. Therefore, even without considering economic growth, the structural change in the global economy and the transition of the driving logic may lead to increased metal demand. Moreover, considering the annual global economic growth, the rising demand for metals may be even greater, reinforcing the broader trend in global demand for metals due to structural changes. In the medium term, the transfer and rebuilding of global production capacities may lead to an increase in physical resource consumption per unit GDP from both the production and demand ends. Historical data suggests that countries with higher levels of investment have experienced significant positive expansions in per capita income, with per capita income showing an inverted U-shaped relationship with final physical consumption per unit. When people have low living standards, they can only satisfy basic survival needs; as per capita income rises, people have the means to consume manufacturing products such as cars, household appliances, etc., which entail more physical consumption; as per capita income continues to rise, surplus income may be directed towards service-oriented consumption, leading to a decrease in final physical consumption per unit output. As of now, per capita final physical consumption in major emerging markets remains at a low level, and in the background where central banks have started cutting interest rates, global industrial chain transfers and new market capacity constructions may significantly accelerate, leading to increased investment needs and subsequently, increased consumption demand. This will create a resonance expansion of "investment-consumption", and the medium-term recovery of the global manufacturing cycle is expected to become a critical support force for metal demand. Metal Team: If AI development is seen as a source of "structural inflation", can raw materials represented by copper, aluminum, and nickel enjoy valuation premiums? Strategy Team: When discussing structural inflation, it means looking at which sectors are outperforming the average inflation levels. Taking the liquor sector as an example, from 2016 to 2020, its wholesale price outperformed the CPI. When real estate prices continued to rise, residents experienced wealth effects, and fast-growing real estate assets led to increased financing needs and economic activity demand, requiring liquor as a medium, leading to increased consumption and rising prices. As wholesalers anticipated price increases, they started stocking up, creating a positive cycle of "price increase-stockpiling," supported by financial expansion, which fueled high premiums in the industry. To some extent, the major commodities in the domestic black sector have also benefited from this. When traders anticipate price increases and hoard goods, they can seek higher potential returns by capitalizing on downstream acceptance of higher prices. Historically, price movements in the steel industry have been impacted by falling steel prices/real estate prices, indicating increasing acceptance levels for the black industry during real estate boom cycles, forming occasional structural inflation elasticity. These last two years, with financial capital entering a contracting cycle and most industries experiencing profit pressures, upward price movements have actually led to faster demand contractions, reducing business acquisition intentions. Thus, superficially, price dynamics represent a balance of supply and demand, but there is also another underlying logic; once financial capital starts to contract, demand for financial assets decreases significantly, eventually leading to the disappearance of structural inflation spurred by financial expansion. When we apply this logic to AI development, sectors related to AI computing have attracted considerable interest and investments from global financial capital, leading to increased sales and profit margins for related companies. In the future, a few possibilities may exist: first, if US financial capital starts contracting, due to any reason, the lack of financial expansion may affect structural inflation. Second, if AI development could lead to structural inflation and if raw materials such as copper, aluminum, and nickel become essential in the AI industry chain, it is possible they may leverage higher valuation premiums. However, whether these raw materials do benefit from such premiums depends on their position in the industrial chain. Value in the industrial chain often flows towards the scarcity points rather than the most valuable creation points. For instance, in the development of the new energy industry post-2023, despite efforts to improve efficiency in the solar sector, solar panels did not receive significant premiums, while resources such as coal and thermal power benefited more due to their scarcity. Therefore, potential premium valuations for raw materials like copper, aluminum, and nickel in the AI industry chain depend on whether they become scarce resources in the AI industry. Our stance on this is still being observed. However, from the current perspective, there are more gaps emerging in US computing infrastructure, moving towards developing more resources in the electricity and infrastructure sectors. This will lead to an increased dependence on commodity demand from these industries. Metal Team: Looking back this year, the non-ferrous metal market has shown more AI related logic. Which key metal categories are involved, and how does AI impact their demand? Strategy Team: This year, the rapid advancement of AI infrastructure has become a significant driver for non-ferrous metal demand. The construction of AI infrastructure primarily involves data centers, power systems, and heat dissipation processes, with copper being the most significantly impacted metal due to its wide-ranging applications in various aspects of AI infrastructure such as data center power supply, cables, transformers, and heat dissipation equipment are essential base metals. Tin is used in PCB soldering materials and electronic interconnect materials in the AI industry, and although its absolute usage may be small, its high unit value allows for significant benefits from processing related activities. According to the Berkeley Lab's 2024 US Data Center Energy Consumption report, it is forecasted that global data centers will consume 325-580 TWh by 2028, requiring approximately 150 GW of electricity. Considering the substantial increase in GPU shipments and single-machine energy consumption, the expected rise in electricity demand will directly drive power grid expansion and investment in electricity infrastructure, further boosting copper consumption. Using the higher range estimate of 580 TWh for electricity consumption, the AI construction in the upcoming five years is expected to drive consumption of approximately 1.42 million tons of copper, contributing an annual of 240,000 tons increment. Overall, the continuous expansion of AI infrastructure is expected to be one of the core drivers for future copper demand growth over the next decade. Risk Warning 1) Weaker-than-expected economic recovery domestically: If subsequent domestic economic data shows a weaker-than-expected trend, the assumption in this text regarding corporate capital returns may not be applicable. 2) Significant downturn in overseas economies: If overseas economies experience an unexpectedly sharp downturn, the global manufacturing cycle recovery may be temporarily halted, resulting in a slowdown in demand for physical assets. 3) Risk of supply exceeding expectations: The global non-ferrous metal supply-demand setup generally leans towards a "tight supply" state. If mining production exceeds expectations or related control policies are relaxed, there could be detrimental impacts on metal prices. 4) Risk of AI industry development falling short of expectations: The continuous expansion of AI infrastructure may become the core driver for metal demand growth over the next ten years. If AI industry development does not meet expectations, there is a risk of metal demand growth not meeting expectations.