How JX Advanced Metals Financed Its Push Into the Materials Behind AI

An AI system needs more than powerful processors. Its data must move, its power must remain stable and its servers must shed heat. JX Advanced Metals is expanding across the materials behind those functions. Its latest example, announced September 10, is a plan by group company Toho Titanium to increase production of small-particle nickel powder used inside multilayer ceramic capacitors, or MLCCs.^[10]

The industrial story is tied to an unusual financing decision. JX coupled convertible bonds with a repurchase of its own shares, seeking both a change in its relationship with ENEOS Holdings and resources for expansion. Understanding that combination matters: the headline bond amount is not the amount available to build new production lines.^[1]

A financing with two destinations

Sources: JX’s May 20 and June 18 disclosures. Approximate amounts follow the company’s rounding.^[2][3]

JX’s May 20 explanation put the initial conversion price at ¥4,860 and described the bonds as zero-coupon. Payments above face value helped leave funds after the proposed buyback. The ¥82.7 billion figure was the company’s estimate for growth investment, not operating profit or a statement that the money had already been spent.^[2]

The distinction changes how the transaction should be judged. Cash used to acquire existing shares changes ownership and capital structure. Cash invested in production can eventually increase supply and earnings. They can belong to one strategy, but they are different uses of money with different tests of success.

The final May 19 terms distinguish issuer payments from investor offering prices. For the 2029 bonds these were 110.75% and 113.25% of face value; for the 2031 bonds, 111.50% and 114.00%. The filing estimated net proceeds of ¥277.593 billion and earmarked the residual for target and crystal-material capacity or rare-metal resources by March 2028. The investor offering price should not be used to calculate the company’s receipts.^[15]

Why buy shares while preparing to expand?

In its May 11 announcement, JX said ENEOS had retained more than 40% after JX listed on the Tokyo Stock Exchange’s Prime Market on March 19, 2025. JX wanted greater flexibility in capital policy and considered reducing that holding a management issue. When ENEOS indicated a wish to sell, JX chose a tender offer, citing the possible market impact of a large disposal. These are management’s stated reasons, rather than evidence that the transaction necessarily improved market valuation.^[1]

The June 18 results recorded 57,300,112 shares repurchased at ¥3,401 each, for ¥194,877,680,912 excluding expenses. The final count was slightly above the planned amount because of allocation rounding after excess applications. The completed result is the appropriate measure of the buyback, rather than the earlier authorization ceiling.^[3]

A repurchase can lift earnings per share by reducing the number of shares over which a given profit is divided. It can also change return on equity without a corresponding improvement in factory performance. Investors therefore need to distinguish a better capital ratio from better operations. The strongest outcome would combine the two, but neither establishes the other.

Zero coupon still involves a trade-off

A convertible bond gives its holder a contractual route from debt into shares. For the issuer, that can reduce immediate cash interest payments. For the investor, the conversion right supplies exposure to a possible increase in the share price. The financing exchanges that opportunity for capital today; describing it simply as free money misses the bargain.

JX said that full conversion at the initial price would involve fewer shares than the planned repurchase. That is a comparison with the position before the combined transaction. Relative to the position immediately after a buyback, subsequent delivery of treasury shares puts shares back into circulation. The starting point and the conversion assumptions matter.^[2]

If holders do not convert, the issuer must prepare to repay debt. If they do, the debt burden can fall while ownership changes. Neither outcome should be treated as predetermined. A responsible assessment asks how the company would fund repayment as well as how shareholders would fare under conversion.

This creates two clocks. One measures the time needed to install equipment, establish reliable production and collect customer payments. The other runs toward the bonds’ maturities. Capacity expansion can require working capital before it produces cash: inventory and production costs arrive before the customer’s payment. A promising order environment does not remove that timing gap.

The optical link: why indium phosphide matters

Indium phosphide, usually abbreviated InP, is a compound semiconductor used in light-emitting and light-receiving devices. Those devices convert between electrical and optical signals. JX’s July 2025 investment announcement said the company had handled the material for more than 40 years and identified it as a prospective earnings pillar within its semiconductor and information-communications materials businesses.^[9]

On June 16, JX announced a policy of investing up to ¥120 billion over four years. Including earlier announcements, it described an approximately ¥150 billion investment scale, targeting capacity seven to ten times fiscal 2025’s level at Isohara and Hitachinaka. It linked funding to the bond proceeds remaining after the buyback. These are investment and capacity plans, not completed spending or realized output.^[4]

The business logic is understandable. Faster computation has less value when equipment must wait for data. Spending to move information can therefore accompany spending on processors. But an increase in traffic does not translate mechanically into an equal increase in substrate revenue. Architecture, component design, material use and selling prices all affect the connection.

For JX, the opportunity lies in turning expertise in crystal materials into a larger, dependable supply business. The challenge is to expand without assuming that every customer forecast will arrive on schedule. Capacity is valuable when customers need qualified products; unused capacity still carries costs.

The materials used to make the chips

Sputtering targets sit further back in semiconductor production. In a vacuum chamber, ions strike a target, releasing atoms or molecules that deposit as a thin film on a surface such as a silicon wafer. The target is a manufacturing material, rather than a finished AI processor.^[7]

JX describes its capabilities in purification, casting with low inclusion levels, powder metallurgy and microstructure control. These point to why the business cannot be understood simply through the tonnage of metal sold. Manufacturing consistency and control of unwanted material are part of what the customer buys.^[8]

The company’s March 10 Hitachinaka announcement specified approximately ¥23 billion of target-production investment. That amount is within the previously announced ¥150 billion new-factory budget and includes an earlier ¥6.6 billion investment. Adding all three figures would double-count expenditure. The separate InP program also uses a ¥150 billion total, but its scope and announcement history differ.^[5]

A July 23 Korean expansion adds approximately ¥4 billion of processing investment, with phased operation planned from the second half of fiscal 2027 and capacity roughly double fiscal 2025’s level. JX describes Korea as a downstream processing base near advanced semiconductor customers. The significance is the connection between production and customer-side processing, rather than a single isolated factory.^[6]

Japan.co.jp’s assessment is that expanding several stages of supply can be more useful than maximizing only one. Extra upstream material cannot automatically become a delivered product if the next processing step is constrained. Equally, idle downstream equipment does little without suitable feedstock and customer demand.

Power and cooling widen the opportunity

Toho Titanium’s September plan calls for phased equipment operation at Chigasaki from 2027 and a doubling of small-particle nickel capacity. The material serves internal electrodes in high-end MLCCs. The company links demand to AI data centers, but a capacity announcement does not establish future shipments or profits.^[10]

TANIOBIS is also expanding upstream production of functional tantalum powder in Thailand, with phased startup planned in the first half of 2027. JX’s June release explains two connections: tantalum targets help form layers supporting reliable fine copper wiring, while tantalum capacitors serve power circuits. The material reaches the AI supply chain through more than one component.^[11]

In August, Tatsuta Electric Wire & Cable announced a 30% increase over fiscal 2025 capacity for leak-detection systems at its Kyoto factory, with full operation scheduled for the end of that month. Its explanation connects demand to liquid cooling in high-performance server facilities. This is a further example of established manufacturing capabilities addressing the physical requirements of computing.^[12]

Product breadth should not be confused with complete diversification. If one wave of data-center construction drives demand for several materials, a slowdown in that wave could affect several businesses together. Different products can share the same underlying investment cycle. That is an analytical risk, not a prediction that the current expansion will reverse.

A history longer than the AI boom

The group traces its beginnings to Fusanosuke Kuhara’s opening of the Hitachi Mine in 1905. The mine closed in 1981; the Isohara plant began operation in 1985. A 2010 merger reshaped the group, the current Japanese company name followed in 2016, and the English name became JX Advanced Metals Corporation in 2024.^[13]

The relevant continuity is technical. Refining, shaping and controlling metals can support increasingly demanding applications even as the end markets change. The current AI opportunity draws on that accumulated capability. It is nevertheless a new commercial test: a history of making materials does not guarantee the successful ramp-up of every new product or production process.

The measure that matters after the financing

JX’s August 6 presentation raised its fiscal 2026 consolidated operating-profit forecast from ¥190 billion to ¥232 billion. Its explanation included a ¥40 billion improvement from market conditions, alongside a ¥6 billion negative effect from reduced mining production caused by bad weather in Chile. The overall upgrade cannot be read as a pure measure of AI-materials performance.^[14]

The next evidence to watch is practical: commissioning dates, the yield of saleable output, volumes and prices, and cash generated after investment. A production line can open on time while taking longer to earn an adequate return. Conversely, better output from existing facilities can matter before a major new factory contributes.

The financing has created room for expansion while addressing an ownership issue. Its lasting value will depend on what follows. JX must turn capital into materials that customers can use, and those deliveries into cash that supports the business through its investment and debt cycles. The decisive evidence will come from both the factory floor and the cash-flow statement.