Trek 2000 International Ltd

Sources: S&P Global Market Intelligence via StockAnalysis.com for FY2021–FY2023 items; company filings for FY2023–FY2025. FY2024 uses the audited, restated comparatives. Values in USD millions unless stated.

Income statement

FY2021 and FY2022 EPS are derived from net profit and weighted average shares (313.83m and 313.09m).

Margins and growth (%) — derived

Revenue CAGR FY2021–FY2025 was -14.8% (derived). The three-year CAGR from the FY2022 trough to FY2025 was +5.7%.

Long-run revenue context

FY2016–FY2020 data is from TipRanks. TipRanks also shows FY2024 revenue of US$27.11m, which conflicts with the company's reported US$19.85m; the company figure is used (see Appendix A).

Balance sheet

FY2021 and FY2022 NAV per share is derived. Goodwill is nil; intangibles are entirely capitalised development costs, since patents and trademarks (cost US$3.83m) are fully amortised.

30 June 2026 balance sheet (1HFY2026 announcement):

• Total assets US$47.36m and total equity US$35.63m.

• Cash and short-term deposits US$24.43m, including a pledged deposit of US$0.28m.

• Unquoted investments US$7.58m.

• Inventories US$5.33m; receivables US$4.99m; payables and accruals US$11.37m.

• Zero borrowings; NAV per share 11.46 US cents.

Cash flow

1HFY2026: operating cash flow was -US$1.30m, reflecting a working-capital build of about US$4.0m. PPE capex was US$0.43m and buybacks US$0.20m.

Ratios

Net debt / EBITDA: not meaningful (net cash position; EBITDA negative).

Interest coverage: not meaningful. The only finance cost is lease interest (US$5k in FY2025).

ROIC: S&P figures are distorted because invested capital is near zero once cash is excluded, and operating income excludes investment gains.

Working-capital days: company-disclosed for FY2023–FY2025; FY2021 and FY2022 not disclosed. 1HFY2026 days were 34 / 14 / 32, a cash conversion cycle of about 16 days.

Commentary on trends, inflections and drivers

Revenue. Trek's revenue base has collapsed and partially reset. It fell from about US$240m in FY2016 to US$41m by FY2018, after forensic audits exposed large related-party trading with T-Data Systems (about US$185.7m of transactions, per the July 2017 RSM report). It then drifted to a FY2022 trough of US$16.6m. FY2023 and FY2024 brought modest recovery from ICS volumes. FY2025 was flat (-1.1%) because management deliberately turned down lower-margin projects amid tariff uncertainty. The decisive inflection was 1HFY2026, when revenue of US$33.4m exceeded all of FY2025. Management attributes this to AI-driven demand for memory modules during a global memory-chip shortage.

Gross margin. Margins are structurally thin, consistent with module integration and resale rather than IP licensing. They turned negative in FY2022 (-14.4%) amid inventory write-downs and the memory downcycle. They recovered to 9.1% in FY2023, slipped to 5.2% in FY2024 (management cited stiff competition and a sluggish consumer market), and rose to 12.1% in FY2025 through project selectivity. They held at 10.6% in 1HFY2026 despite the revenue tripling.

Profit quality. Every profitable year since FY2021 relied on non-operating items:

• FY2023: a US$3.1m gain on disposal of an unquoted investment.

• FY2025: a US$6.1m gain on selling an unquoted equity stake for US$6.5m. The company says the price included consideration for strategic cooperation and technical support given to the investee.

• FY2025: a US$0.72m share of associate profit.

Stripping out the FY2025 disposal gain and the US$0.19m associate revaluation, FY2025 pre-tax profit would have been about -US$1.8m (derived). FY2025 G&A also quadrupled to US$3.4m on higher professional fees, which likely relate to the Ron Sim dispute and governance work, although the company does not specify. 1HFY2026 is the first period in the review window where trading operations alone produced a clear profit: segment results excluding Corporate were about +US$1.5m.

Balance sheet and cash. Liquidity has been preserved throughout. Net cash plus investments never fell below US$20m and stood at US$33.6m at end-FY2025. The investment mix has shifted from quoted securities (US$17.1m at end-FY2023) to bank deposits (US$26.9m at end-FY2025) and unquoted funds valued at Level 3 on manager-supplied NAVs. Operating cash flow swings widely because of working capital: -US$7.0m in FY2024 when receivables and inventory rose, +US$7.6m in FY2025 on unwinds and US$5.2m of extended payables. A further reversal (-US$1.3m) came in 1HFY2026 as the company built inventory for demand.

Capital allocation. For eight years no dividends were paid. Buybacks were small but accelerated (8.18m shares in FY2025, 1.71m in 1HFY2026). Notably, the S$11.58m raised in 2015 is still entirely unutilised more than a decade later. It was earmarked half for R&D (originally medical diagnostics) and half for working capital.

Management guidance (1HFY2026 announcement, 12 August 2026)

Management says it is "cautiously optimistic" for the next 12 months. It expects AI-driven demand to continue alongside a memory-chip shortage that will keep disrupting supply chains. Its stated priorities are selective, profitable projects, cost discipline, continued innovation in core storage and AIoT, and continued AIRES development. The Executive Chairman says the company remains optimistic about 2026 on AI memory demand and will keep investing in AI-enabled technologies and expanding in high-growth markets. The aviation revenue expectation of more than US$15m through FY2027 was reiterated in May 2026. No numeric revenue or profit guidance is given.

Consensus

No sell-side consensus is available; this is not publicly disclosed.

Bull case

1. Cycle tailwind with operating leverage: 1HFY2026 gross profit of US$3.5m against roughly US$2.1m of half-year operating expenses shows that, if TrendForce's view of rising prices into 2027 holds, Trek can generate genuine operating profit. 2H2026 could match or exceed 1H given TrendForce's +10–15% DRAM and +15–20% NAND 4Q26 contract-price forecast.

2. Downside protected by cash: about US$32m of liquid assets against an EV of only about S$18m means the market values the operating business at a fraction of annual sales (about 0.3x EV/Sales). A sustained dividend and buybacks could close the gap.

3. Optionality and a cleaner shareholder register: Azure Capital's entry, the end of the Ron Sim dispute, a strengthened ARC chair and possible aviation revenue (more than US$15m through FY2027) provide re-rating catalysts not in the base business.

Bear case

1. Cyclical, low-quality earnings: core operations lost money in every year from FY2021 to FY2025. Reported profits relied on one-off investment gains (US$6.1m in FY2025), and FY2022 showed how quickly a downcycle produces losses (-US$13.3m).

2. Concentration and margin deficit: 74.9% of 1HFY2026 revenue came from Taiwan, and gross margins (10.6–12.1%) trail module peers by 9–35 percentage points. Trek lacks pricing power and is exposed to the loss of a single large buyer or a price reset.

3. Governance discount is earned: the founder convicted in 2022 is still on payroll, the Chair and CEO roles are combined, committees were briefly non-compliant, segment disclosure is opaque, and S$11.58m of 2015 proceeds is idle after more than ten years. All of this argues for a structural discount to NAV-plus-earnings value.

Catalysts and monitorables (next 12 months)

• FY2026 full-year results (expected around late February 2027): watch the 2H2026 revenue run-rate, whether gross margin holds above 10%, inventory and receivable days, and whether a final dividend is declared. One aggregator lists an estimated 12 November 2026 date, which does not fit Trek's half-yearly reporting.

Memory pricing: TrendForce 1Q27 contract-price outlook, as a lead indicator of Trek's revenue.

Aboard AI: first commercial orders, certification milestones, and recognised revenue against the more-than-US$15m target.

AIRES: order flow beyond the US$350k Cambodia contract, and whether segment results start to be measured rather than allocated.

Capital allocation: use of the 2015 proceeds, any acquisitions, buyback execution against the 30.24m mandate, and a formal dividend policy.

Shareholder register: Kioxia's and Creative's intentions; Azure Capital's engagement; directors' option exercises.

Governance: any separation of Chair and CEO roles; appointment of a female director; related-party employment disclosures.

Analyst verdict

Trek 2000 International is a cash shell with a cyclical memory-trading business inside it. It has just enjoyed the best half-year in its post-scandal history. The 1HFY2026 numbers are real and important. Revenue tripled to US$33.4m, gross profit rose nearly ninefold, and, for the first time in the five-year window, trading operations rather than portfolio gains drove profit. Combined with US$32m of liquid assets, no debt, a reinstated dividend and the end of the Ron Sim dispute, Trek is fundamentally healthier than at any point since the 2016–2018 governance collapse.

But the improvement is borrowed from the memory cycle, not built on durable franchise economics. Trek's gross margins are a fraction of its Taiwanese peers' in the same upswing. One geography supplies three-quarters of revenue. Its own disclosures make it impossible to tell whether AIRES or Customised Solutions create value. The strategic narrative of IP, AI avionics and smart energy remains largely prospective: AIRES produced a quarter of a million dollars of revenue in the half, and aviation has produced nothing reportable yet. Governance has improved at the board level, but the founder's continued employment, the combined Chair and CEO role, and a decade-old pool of unspent equity proceeds keep the discount justified.

At about S$0.195, roughly 1.3x book and an EV near S$18m, the market already prices in a sharp fall in earnings from current run-rates. The stock offers asymmetric exposure for investors who believe memory tightness persists into 2027 and that management will return surplus cash. It is a tactical, cycle-linked holding rather than a structural compounder. The key test is whether FY2026 results confirm a positive operating line through a full year, and whether the board commits to a formal distribution policy. Absent both, the September–October 2026 price retracement is likely to continue whenever memory-price momentum fades.

---

APPENDIX A — DATA DISCREPANCY LOG