Mi Technovation Berhad (5286.KL) - Deep Dive Research Report

Prepared 13 July 2026

Section 1: What the Company Does

Mi Technovation makes two things that sit right next to each other on a semiconductor assembly line but are usually made by completely different companies: the machines that sort and inspect finished chips, and the tiny metal balls that solder those chips onto circuit boards.

Start with the machine. After a silicon wafer is fabricated and diced into individual chips, each chip has to be tested, graded by quality ("binned"), and picked into the right tray before it ships to a phone maker or a module house. Mi's flagship product, the Mi Series wafer-level chip-scale packaging (WLCSP) die sorter, does exactly this. Its distinguishing feature is a proprietary horizontal turret that moves chips through the sort-and-inspect steps faster and more gently than the conventional rotary pick-and-place arms competitors use. This one platform family accounts for more than 90% of the equipment unit's sales. When a chipmaker ramps a new mobile radio-frequency chip or a wearable sensor, it buys Mi die sorters by the dozen.

Now the ball. A modern chip does not have legs anymore; it has a grid of microscopic solder spheres on its underside (a ball grid array, or the balls deposited directly in wafer-level packaging). Those spheres are a precision-engineered material: exact diameter, exact alloy, exact sphericity, or the chip does not join reliably to the board. Mi makes them through its Taiwanese subsidiary Accurus Scientific, one of the top five solder-ball suppliers in the world.

The reason both businesses live under one roof is a deliberate bet by the founder. Oh Kuang Eng, a University of Malaya mechanical engineer, started at Hewlett-Packard in 1995, built and sold an automation firm (AGS Automation) by 2006, co-founded an assembly-equipment maker (DPE) in 2007, and folded that business into Mi Equipment in 2012-2013. He listed Mi Equipment on Bursa Malaysia's Main Market in June 2018 and renamed it Mi Technovation that December. The pivotal move came in April 2021, when he acquired Accurus and turned a one-product equipment company into a company that sells both the tool and the consumable. His stated ambition, repeated in interviews, is to build "a diversified semiconductor solutions provider" rather than a single-product equipment vendor.

The company's own framing is that it wants to own more of the semiconductor back-end value chain - the equipment, the material, the power device, and now the vehicle powertrain - rather than ride the boom-and-bust of any one of them.

That framing explains everything that follows: a company that has bolted on a power-semiconductor unit and a vehicle-technology unit on top of its two profitable core businesses.

Section 2: Business Segments

Mi Technovation now reports across four business units. Two of them make essentially all the money today (equipment and materials); two are venture-stage bets (power semiconductors and vehicle technology).

Semiconductor Equipment Business Unit (SEBU) - the founding business

SEBU designs and builds the WLCSP die sorters described above, along with vision inspection machines, precision bonding machines, final-test handlers, and the smart-factory automation that ties them together, plus the spare parts and service that follow an installed machine for years. Manufacturing and R&D sit in Malaysia (Penang - Bayan Lepas and Batu Kawan) with satellite operations in Korea, Taiwan, and China.

The core capability is the horizontal turret sorting architecture and the process knowledge to handle bare, fragile die at high throughput without yield loss. This took over a decade to refine, and it is why customers who standardise on Mi sorters tend to keep buying them. Within its niche - high-throughput WLCSP die sorting for mobility and wearables chips - Mi is a genuine specialist rather than a generalist competing head-on with the giants. It wins on throughput-per-dollar and on being nimble for a specific package type; it loses where a customer wants a full back-end line from a single vendor (bonder, sorter, tester all from one supplier), which is where ASMPT and Besi are stronger. Roughly 55-65% of group revenue in recent quarters. Management treats it as the anchor and talks about a "die sorting and binning platform" as the key demand driver into the AI and advanced-packaging cycle.

Semiconductor Material Business Unit (SMBU / Accurus Scientific) - the earnings engine

SMBU manufactures solder spheres and adjacent consumables (solder ingots, solder paste, ESD packaging), run under the Accurus brand with plants in Tainan (Taiwan, the headquarters, roughly 20 production lines at high utilisation), Ningbo (China, around six lines running below 70%), and a presence in Singapore. In Q1 2025 the group added a roughly US$10 million interconnects facility in Senai, Johor, Malaysia.

The capability here is metallurgical: producing spheres to tight diameter and sphericity tolerances in specialty alloys, qualified into customers' packaging processes. Because a solder ball becomes part of the customer's qualified bill of materials, switching suppliers means re-qualifying the joint - slow and risky - which gives Accurus stickiness a pure commodity would not have. It exists as a separate unit because it was acquired (2021), has a different customer base (packaging and assembly houses rather than equipment buyers), different economics (a recurring consumable sold by volume, not a lumpy capital tool), and its own competitive set. Crucially, although SMBU is often the smaller revenue line, it contributes an outsized share of group earnings - roughly 45-55% - because solder balls are a higher-margin recurring product than machines. Competitors are the other global solder-ball leaders: Senju Metal (Japan), Shenmao Technology (Taiwan), and MK Electron (Korea). Management describes SMBU as the steadier, cash-generative counterweight to lumpy equipment orders.

Semiconductor Technologies Business Unit (STBU, formerly SSBU) - the power-semiconductor option

STBU is the group's move into silicon-carbide (SiC) power semiconductors - the power modules and devices used in EV inverters, chargers, and renewable-energy systems. The group invested about US$30 million in a Hangzhou, China R&D-and-manufacturing facility, launched in October 2023. This is a strategic option rather than a profit contributor: it is small, still ramping, and pushes Mi from back-end assembly into actual device design, a much harder and more capital-intensive game dominated by Infineon, onsemi, STMicroelectronics, and Wolfspeed. Management positions it as a long-dated bet on the electrification cycle, not a near-term earnings driver.

Vehicle Technologies Business Unit (Ohima) - the newest bet

Launched in 2026 under the "Ohima" brand, VTBU develops intelligent EV powertrain systems, electric drive components, and autonomous-driving platforms. In the Q1 FY2026 result, management confirmed that VTBU revenue "materialised as planned" with early development of electric powertrain systems. It is the most speculative unit, taking Mi outside semiconductors entirely into mobility systems - a large adjacent market but one with different customers and no installed base yet.

Section 3: Products and Business Detail

The equipment catalogue. The centre of gravity is the Mi Series WLCSP die sorter, a machine that picks tested die off a wafer, inspects each one with machine vision, grades it into a quality bin, and places it precisely into carrier tape or a tray. The horizontal turret is the differentiator - conventional sorters swing die on a rotary head or a linear pick-and-place gantry, and Mi's horizontal-turret motion is what management credits for higher throughput and gentler handling of thin, fragile die. Around this sit vision inspection machines (defect detection on packages and die), precision bonding machines, final-test handlers, wafer-fabrication tools, and testing instruments, plus smart-factory automation software for Industry-4.0 line control. The recurring layer is spare parts, components, and maintenance service on the installed base - the annuity that comes after each machine sale. Management's shorthand for the current demand pull is the "die sorting and binning platform."

The materials catalogue. Accurus makes solder spheres in a wide range of diameters and alloy compositions, used in ball grid array (BGA) and wafer-level packaging to form the electrical and mechanical joint between chip and substrate. Around the core sphere it sells solder ingots, solder paste, and ESD-safe packaging bottles as complements. The high-value growth product management repeatedly flags is its specialty alloys - custom solder formulations for higher-reliability or higher-performance joints, adopted increasingly in mobility, wearables, and high-performance-computing packages.

Manufacturing footprint. Equipment is built primarily in Penang (Bayan Lepas and Batu Kawan), with technology centres in Korea, Taiwan, and China. Solder balls come from Tainan (the high-utilisation flagship, ~20 lines), Ningbo (~6 lines, run below 70% and therefore holding latent capacity), and the newer Senai, Johor interconnects site. The power-semiconductor unit runs from Hangzhou. Singapore serves as a corporate and coordination hub.

Geographies. Sales concentrate in the East-Asian semiconductor heartland: Taiwan and China are repeatedly cited as the demand centres for equipment (advanced-packaging capacity expansion), with Korea and the broader Asia-Pacific packaging base buying solder balls. This is a company whose customers sit almost entirely in Asia's chip-assembly corridor.

Milestones that changed the business. Listing in June 2018 (as Mi Equipment). The April 2021 Accurus acquisition, which turned a single-product equipment maker into a dual equipment-plus-materials company and remains the single most important corporate decision. The October 2023 launch of the US$30m Hangzhou SiC facility. The Q1 2025 opening of the US$10m Senai interconnects plant. And the 2026 launch of the Ohima vehicle unit.

Section 4: Customers

Mi sells to two distinct customer bases that rarely overlap.

The equipment unit sells to chipmakers and outsourced assembly-and-test houses (OSATs) that run WLCSP lines - the factories that package mobile radio-frequency chips, power-management ICs, sensors, and wearable-device silicon. The buying decision here is made by process and equipment engineering teams, and the criteria are throughput, yield, handling gentleness on thin die, cost of ownership, and how well the tool drops into an existing line. Sales cycles are long: a customer evaluates a demo tool, qualifies it against its own die and yield targets, then places volume orders as it expands capacity. Once a fab standardises on Mi sorters and trains its technicians on them, re-tooling to a rival platform is disruptive, which is the source of repeat orders.

The materials unit sells to packaging and assembly houses that consume solder balls as a production input. The decision-makers are materials-qualification and supply-chain engineers, and the binding criterion is qualification: a solder sphere alloy and size gets designed into a package's qualified process, and changing supplier means re-running reliability testing on the joint. That qualification is the switching cost - it is why Accurus holds share and why the relationship is recurring and volume-based rather than project-based.

On concentration, the company does not publicly name marquee accounts, and specific customer-concentration percentages are not disclosed in the material I could access, so I will not invent them. What management does disclose is end-market exposure: mobility and wearables, HPC and memory (the AI and high-bandwidth-memory pull), and automotive and renewable energy. The geographic concentration in Taiwan and China is itself a form of concentration risk.

Contract structure differs by unit and matters for how you read the numbers. Equipment is lumpy capital-order business - revenue arrives in waves as customers expand capacity, which is why equipment quarters swing. Materials is recurring consumable revenue that flexes with customers' production volumes and, at times, with order pull-ins (in Q2 2025 management explicitly attributed a solder-ball surge partly to customers pulling orders forward amid tariff concerns). The blend gives the group a lumpy capital line balanced by a steadier consumable line.

Section 5: Competitive Landscape

Mi competes in two different arenas, and it is a small fish in both by headline size, but a real specialist in its chosen niches.

In back-end equipment, the giants are Besi (Netherlands), ASMPT (Hong Kong-listed, Singapore-rooted), and Kulicke & Soffa (US), with Cohu, Disco, Hanmi, and Hanwha also active. These firms make bonders, die-attach, and full assembly lines. Mi does not try to out-range them; it competes on a specific product - high-throughput WLCSP die sorting - where its horizontal-turret architecture gives it a throughput-and-cost edge for that package type. It wins when a customer wants the best sorter for mobility/wearables die; it loses when a customer wants a single vendor to supply the entire back-end line, because the majors bundle. The barrier protecting Mi is narrow but real: the process knowledge to handle thin die at high speed without yield loss, plus the installed base and qualification at existing customers. It is not the moat of a Besi in hybrid bonding, but it is enough to sustain a defensible niche.

In solder-ball materials, Accurus is a top-five global supplier competing with Senju Metal (Japan), Shenmao Technology (Taiwan), and MK Electron (Korea). Here the barrier is metallurgical process control plus customer qualification lock-in. This is a more consolidated, higher-barrier market than commodity solder, and it is why the materials unit earns better margins than the equipment unit.

In SiC power semiconductors (STBU), Mi is a new entrant against incumbents an order of magnitude larger - Infineon, onsemi, STMicroelectronics, Wolfspeed. It has no scale advantage and no installed base; this is a venture position, not a competitive threat to anyone yet.

The honest read: Mi has defensible specialist positions in die sorting and solder balls, not a wide moat across the board. Its edge is depth in two niches, not breadth.

Section 6: Industry

Mi rides the semiconductor back-end - the assembly, packaging, and test stage that happens after wafers are fabricated. Two structural forces drive its demand.

First, advanced packaging. As transistor scaling slows, the industry increasingly wrings performance out of how chips are packaged and interconnected rather than just how small the transistors are. That shift pushes spending toward back-end equipment and toward the materials (solder balls, interconnects) that make advanced packages work. The AI and high-bandwidth-memory (HBM) build-out is the sharpest expression of this: HBM stacks and advanced logic packages need more, and more precise, interconnect. Management has repeatedly named "HPC & memory" as a growth driver for its solder-ball unit, which is a direct read-through to the AI packaging cycle.

Second, mobility and wearables. WLCSP is the packaging style used heavily in phones and wearables, and every new device generation refreshes demand for die sorters and solder balls. Management named the mobility and wearables segment as the driver of its equipment strength through FY2025.

The advanced-packaging equipment and materials market is in a multi-year expansion, with industry research houses pointing to advanced packaging as the fastest-growing slice of back-end spend on the back of AI, HPC, and heterogeneous integration. Mi sits in the East-Asian core of this supply chain - Taiwan, China, Korea, and Malaysia - which is where the world's chip assembly is concentrated.

Cyclicality is the defining feature. Semiconductor equipment is famously boom-and-bust: capital orders surge in an up-cycle and evaporate in a down-cycle (the group's own equipment unit swung to a loss in a weak quarter of FY2024 before rebounding sharply in FY2025). The materials side is steadier because it tracks production volume rather than capex, which is precisely why owning both dampens the swing. Two industry-level wildcards recur in management's commentary: US-China tariff and trade dynamics (which in 2025 pulled some solder-ball orders forward as customers hedged) and foreign-exchange, since the group reports in ringgit but earns heavily in US dollars and other currencies, making USD/MYR moves a swing factor on reported profit.

Regulation and geopolitics matter more than typical for a company with Chinese manufacturing (Ningbo solder balls, Hangzhou SiC) selling into a semiconductor supply chain that sits at the centre of US-China export-control tension. That is an industry condition Mi cannot control.

Section 7: Growth Triggers

All triggers below are drawn from management's quarterly-result commentary and the analyst briefings that followed each release. Malaysian issuers do not publish call transcripts, so these are cited to the results-release date.

• SEBU double-digit revenue growth guided for FY2025, driven by mobility and wearables demand for Mi Series die sorters. (Q2 FY2025 results, ~14 Aug 2025; repeated at Q3 FY2025 results, ~13 Nov 2025.)

  Management stated it was "cautiously optimistic" on the 2H25 outlook for the equipment unit, "anticipating a double-digit revenue growth for FY25."

• SMBU (solder-ball) growth led by specialty-alloy adoption plus HPC and memory demand - the AI/HBM packaging pull. (Q2 FY2025, ~14 Aug 2025; reaffirmed Q3 FY2025, ~13 Nov 2025.)
• Die sorting and binning platform demand into advanced packaging cited as the sustaining driver for equipment. (Q4 FY2025 results, ~late Feb/3 Mar 2026.)
• Vehicle Technologies Business Unit (Ohima) revenue beginning to materialise, with development of electric powertrain systems and related components confirmed as on-plan. (Q1 FY2026 results, ~late May 2026.)

  Management stated VTBU revenue "materialised as planned in the first quarter with development of electric powertrain systems."

• SMBU solder-ball strength continuing into subsequent quarters, supported by high-performance-computing and memory plus seasonal mobility and wearables demand. (Q1 FY2026 results, ~late May 2026.)
• Senai, Johor interconnects facility (~US$10m) in production from Q1 2025, adding materials capacity. (Company disclosure, Q1 2025 onward.)
• Ningbo latent capacity - the China solder-ball plant runs below 70% utilisation, leaving room to grow volumes without major new capex. (Analyst briefing context, FY2025.)
• Hangzhou SiC power facility (~US$30m) ramping as the long-dated electrification option. (Launched Oct 2023; ongoing.)

Section 8: Key Risks

Equipment-order cyclicality. SEBU is capital-order business, and it has already shown it can swing hard - the unit ran at a loss in a soft quarter of FY2024 before rebounding through FY2025. The mechanism is simple: when customers pause capacity expansion, orders vanish quickly and there is little backlog to cushion the fall. This is a high-probability, recurring drag rather than a tail risk, and it is the single most important reason the group bought a materials business to offset it.

Foreign-exchange translation. The group reports in ringgit but earns substantially in US dollars and other currencies. Management has explicitly attributed quarter-to-quarter profit swings to forex: a stronger dollar produced gains in Q4 FY2025, while a forex loss dragged Q2 FY2025 net profit down even though revenue grew strongly. The mechanism is translation and remeasurement, not operations - but it makes reported profit noisy and can mask or exaggerate the underlying trend in any single quarter.

In Q2 FY2025, management flagged that net profit contracted "despite strong revenue growth" specifically because of a foreign-exchange loss - a reminder that a good operating quarter can still print a weak profit line.

Geographic and geopolitical concentration. Manufacturing in Ningbo and Hangzhou, plus heavy sales into Taiwan and China, place Mi at the centre of US-China semiconductor trade tension. Export controls, tariffs, or a hard decoupling could disrupt both its China operations and its cross-strait customer base. In 2025 tariff fears actually helped (customers pulled orders forward), but the same dynamic can reverse into a demand air-pocket once pull-ins are digested.

Customer and end-market concentration. With demand concentrated in mobility/wearables and a Taiwan/China customer base, a slowdown in smartphone or wearable refresh cycles, or the loss of a large packaging customer, would hit disproportionately. The company does not disclose customer concentration, so the risk cannot be sized precisely - which is itself a disclosure gap worth noting.

Capital diverted into unproven ventures. The group is funding two venture-stage units (STBU SiC, VTBU Ohima) that consume capital and management attention while contributing little revenue, and it trimmed dividends in 2025 to do so. If SiC power and EV powertrains fail to scale, that capital earns a poor return and the diversification thesis weakens. This is a moderate-probability, moderate-impact risk: the bets are affordable against group cash generation, but they are outside Mi's proven competence in semiconductor back-end.

Materials commoditisation. Solder balls are a differentiated but ultimately physical consumable. If specialty-alloy differentiation narrows or Chinese solder-ball capacity floods the market, the materials unit's margin advantage - the reason it drives half of group earnings - could erode.

Section 9: Walk the Talk

The six reporting periods used, most recent first: Q1 FY2026 (~late May 2026), Q4/FY2025 (~late Feb/3 Mar 2026), Q3 FY2025 (~13 Nov 2025), Q2 FY2025 (~14 Aug 2025), Q1 FY2025 (~late May 2025), and Q4/FY2024 (~24 Feb 2025). The most recent is within 90 days of today.

The story across these six is a management team that set a specific, modest bar early and then cleared it, with the honest complication that reported profit was repeatedly whipsawed by forex.

Coming out of a weak FY2024 - where the equipment unit had actually swung to a loss in one quarter - management framed FY2025 as a recovery year and, by mid-year, put a number on it: double-digit revenue growth for the equipment unit in FY2025, with modest growth for materials.

At the Q2 FY2025 result (14 Aug 2025), management was "cautiously optimistic," "anticipating a double-digit revenue growth for FY25" for SEBU and "modest growth" for SMBU.

They delivered on the direction decisively. By the Q3 FY2025 result (13 Nov 2025), the group posted a record quarterly profit and reversed the prior-year loss, with nine-month profit up sharply, and management reaffirmed the equipment unit was "on track for double-digit growth for FY25." The full-year FY2025 result confirmed it: group revenue and net profit both rose materially year-on-year, and the equipment unit had clearly recovered. So the central FY2025 promise - double-digit equipment growth and a return to health - was kept.

The place where the narrative needs care is profit versus revenue. Management was consistently accurate on the operating trajectory (revenue and segment growth landed roughly where they guided), but the profit line was noisy for reasons they flagged in advance rather than hid. Q2 FY2025 is the clean example: revenue grew strongly, yet net profit fell, and management attributed the miss to a forex loss rather than pretending the quarter was weak operationally. The reverse happened in Q4 FY2025, where a favourable dollar move flattered profit. To their credit, they named forex as the swing factor both times rather than claiming credit or dodging blame - a mark of straight talk.

On the newer bets, management has been careful not to overpromise. The vehicle-technology unit (Ohima) was described at the Q1 FY2026 result as revenue "materialising as planned" - deliberately understated language for a venture-stage unit, not a hype-laden launch. Similarly, the SiC power unit has been positioned as a long-dated option, not a near-term earnings driver, which is consistent with how it has actually contributed (little, so far).

The plain assessment: this is management that does roughly what it says on the operations, sets conservative-to-modest expectations, and is transparent about the forex noise that muddies the reported bottom line. There is no pattern of missed promises or quietly dropped guidance across these six periods. The open question is execution on the two new units, where there is not yet a track record to judge.

Section 10: Shareholder Friendliness Index

Dividends. Mi pays semi-annual dividends, and the pattern has been variable rather than a steady climb. By calendar-year totals: roughly 4.0 sen in 2023, an elevated ~8.0 sen across 2024 (including a stronger year-end payout), then a step-down to roughly 3-4 sen across 2025, and a 1.0 sen first interim declared for 2026 (ex-date 11 June, paid 6 July 2026). The notable feature is that dividends were trimmed in 2025 even as profit hit records - management redirected cash into the Senai interconnects plant, the Hangzhou SiC facility, and the new Ohima vehicle unit. So the dividend is real and recurring but explicitly subordinated to reinvestment, and it should be read as a growth-reinvestment policy rather than a payout-maximising one.

Buybacks and dilution. The company is actively repurchasing shares under a standard Bursa general mandate. In MoatMap's recent-window data, Mi bought back 170,000 shares on 2 July 2026 at RM4.77-4.79 (about RM0.81m), and the disclosed cumulative treasury position stands near 1.67% of capital - meaning the group has been accumulating treasury shares over an extended programme, not just in July. I could not independently verify the year-by-year buyback figures for 2023-2024 through external search (Bursa's announcement portal is not readily indexed), so I will not put precise older numbers on it; what is verifiable is an active, ongoing buyback in 2026 and a treasury stake around 1.67%. Against that, share count has been broadly stable at roughly 890 million shares, with the buyback modestly shrinking the free float rather than option grants inflating it - a mildly shareholder-positive dilution picture.

Verdict: Neutral. Mi both pays a dividend and buys back stock, but it deliberately trimmed the payout in 2025 to fund expansion, so it is best classified as a reinvest-first company that returns some capital rather than one that prioritises returning capital to shareholders.

Section 11: Insider Activities

Per the guidance for this gated venue (Bursa Malaysia), the MoatMap disclosure database is the sole source for recent insider dealing; the exchange portal returns auth-blocked stubs and was not used.

All insider activity in the last 12 months is one person selling, and it has a clean, disclosed explanation.

(Bursa Changes in Director's Interest, Section 219 CA 2016, 26 May / 29 May / 3 June 2026. The MoatMap feed lists each transaction twice - once as a director-interest change and once as a dealing notice - so the three underlying trades total ~300,000 shares.)

Sells - the why. Heng Kok Lin, the 63-year-old executive director who ran the materials business unit under Accurus (the unit that drives roughly half of group earnings), sold about 300,000 shares across late May and early June 2026 for a combined ~RM1.46 million. The reason is disclosed and not ambiguous: he chose not to seek re-election at the 2026 AGM and formally retired as an executive director in June 2026, transitioning to a "BU Fellow" mentoring role. The trades are classic pre-retirement diversification by a long-serving insider stepping back from the board, not a signal about the business outlook. The sizes are also small relative to the company (each trade is ~0.01-0.02% of shares outstanding). This is the kind of sell that is noisy rather than negative.

Buys. There were no open-market insider purchases in the window. There is, however, a company-level counter-signal: Mi itself was buying back shares over the same period (170,000 shares on 2 July 2026), which is the corporate equivalent of insiders putting money in.

Net assessment. Insider activity is narrow (one departing director), fully explained by retirement, and small in size. It is not cluster selling, not broad-based, and not tied to any adverse disclosure. Set against an active company buyback, the net read is neutral - a retiring executive trimming a personal holding, with no bearish signal about the business and a mild positive from the company repurchasing its own stock.

Section 12: Scenarios

Bull case. The advanced-packaging cycle keeps accelerating. AI, HBM, and heterogeneous integration drive packaging houses in Taiwan, China, and Korea to expand capacity, and every expansion pulls in more Mi die sorters and more Accurus solder balls at the same time - the two core units firing together. Specialty alloys become a bigger, stickier share of the materials mix, lifting margins, while the Ningbo plant's spare capacity absorbs volume growth without heavy new capex. Then the optionality pays off: the Hangzhou SiC unit reaches commercial scale on the back of EV and renewable demand, and the Ohima vehicle unit lands its first real powertrain programmes, giving Mi a third and fourth growth leg outside its cyclical semiconductor base. Two years out, Mi looks like a diversified back-end and power-systems group whose earnings are less hostage to any single chip cycle, and the dividend cut of 2025 reads as a well-timed reinvestment.

Base case. The two core units carry the company. Equipment continues its recovery on mobility, wearables, and advanced-packaging demand, delivering the double-digit-type growth management has guided, while solder balls grow steadily on HPC, memory, and mobility. Reported profit stays choppy quarter to quarter as USD/MYR swings add and subtract from the bottom line, but the underlying trajectory drifts upward. The SiC and vehicle units keep consuming modest capital and generate small, slowly building revenue without yet moving the needle - a long-dated call option that neither pays off nor blows up in the window. Dividends stay variable and modest as cash keeps flowing into expansion, and the buyback continues to nibble at the share count. Mi remains a solid specialist in two niches with two speculative bets attached.

Bear case. The semiconductor capex cycle rolls over. Packaging customers pause expansions, equipment orders dry up quickly with little backlog to cushion them, and SEBU swings back toward the kind of loss it printed in a weak FY2024 quarter. At the same time, 2025's tariff-driven order pull-ins in solder balls reverse into a demand air-pocket, and Chinese solder-ball capacity pressures Accurus pricing, compressing the very margins that make materials the earnings engine. US-China export controls or tariffs disrupt the Ningbo and Hangzhou operations and the cross-strait customer base. Forex turns against the ringgit-to-dollar translation at the wrong moment, amplifying the reported decline. And the capital sunk into SiC power and the Ohima vehicle unit - both outside Mi's proven competence - fails to scale, turning the diversification story into a drag on returns rather than a hedge. In that world, Mi is a small, cyclical back-end supplier that spent its up-cycle profits on bets that did not land.