SUNeVision Holdings

Phase A — Understand the business

Lens 1 · Company Overview

SUNeVision is the technology arm of Sun Hung Kai Properties (SHKP, 0016.HK) and, on any measure, the largest carrier-neutral data-centre landlord in Hong Kong — roughly a fifth of the territory's installed IT capacity. Strip away the "software" GICS bucket the screeners file it under and what you actually own is a REIT-like industrial property compounder wearing a data-centre logo: SHKP's balance sheet and land bank, turned into recurring colocation rent.

The business is two lines:

• Data-centre & IT-facilities (~92% of revenue) — carrier-neutral colocation under the iAdvantage brand: rack space, power, cooling, cross-connects and interconnection to the ~100+ carrier/cloud ecosystem in its buildings. Sold as multi-year contracts; revenue is recurring rent + power pass-through, not project-lumpy.
• Extra-Low-Voltage (ELV) & IT systems (~8%) — building-systems integration and structured cabling, largely for SHKP-group and third-party properties. Lower-margin, project-based, effectively a legacy tail.

Revenue FY2025 (yr-end 30 Jun 2025): HK$2,938M, +9.9% YoY. Profit attributable to shareholders HK$979M, +8%; EBITDA HK$2,128M, +15% (company-reported); net cash from operations HK$2,063M, +23%. This is a high-margin annuity: ~57% gross margin, ~72% EBITDA margin on the reported number — the economics of scarce, sticky infrastructure, not of a software vendor.

Customers: carrier-neutral means the tenant base is the who's-who of clouds and carriers with Hong Kong presence — Microsoft/Azure, AWS, Google, Alibaba Cloud, Tencent Cloud, Huawei all operate in the HK market, plus international banks and telcos anchoring the older campuses. The MEGA IDC flagship has "already attracted major cloud service providers and international banks to move in". Exact customer names and concentration are n/a — not disclosed at tenant level; the July-2026 EGM confirms a single large anchor now dominates the MEGA IDC Phase-1 leasing story (see Lens 5), which is itself a concentration flag.

Suppliers / inputs: land + power + construction. The single most important supplier relationship is with its own parent, SHKP — land, buildings and construction services flow through continuing connected transactions re-approved by independent shareholders at the 30 June 2026 EGM. Grid power is supplied by CLP Power (Kowloon/NT). This is a business whose supply chain is unusually internalised (parent land) and external-monopoly-dependent (single grid).

Lens 2 · Supply Chain

Named map, upstream → SUNeVision → end customer:

Upstream inputs

• Land & buildings → Sun Hung Kai Properties (0016.HK). The controlling shareholder supplies the sites and, historically, the shell. "SUNeVision's entire data-centre facilities, ~1.5M sq ft GFA, are owned by SUNeVision or its controlling shareholder SHKP". This is the single largest chokepoint and the single largest moat — see Lens 3. It is also the related-party knot (Lens 10/13).
• Grid power → CLP Power Hong Kong (the Kowloon/New Territories utility monopoly). The binding constraint in HK is grid capacity: securing it can take 4–5 years vs. <1 year in peer markets. Any operator without secured power allocation cannot build; SUNeVision's incumbency (existing substation ties in TKO) is a hard advantage.
• Construction & fit-out → local contractors (competition for contractor services is pushing HK build costs above inflation), plus SHKP's own construction arm via the connected transactions.
• Critical M&E kit → global OEMs — generators, UPS, chillers, switchgear (Vertiv, Schneider, Caterpillar-class vendors; specific vendors n/a — not disclosed). GPU-server availability is now an upstream gating input on the demand side: management explicitly tied the recent order pickup to "improved GPU supply in the region".

SUNeVision (the node): converts land + power + capex into leasable, powered, cooled, interconnected white space under iAdvantage. Assets: MEGA IDC (TKO, flagship), plus the earlier JUMBO / SUPreme / TKO-area campuses.

Downstream / end customers: hyperscale clouds (Azure/AWS/Google/Alibaba/Tencent), international & Chinese carriers, banks and enterprises needing HK latency/data-sovereignty. Hong Kong's role in the chain is the low-latency China-facing gateway — the reason a global cloud pays a premium for HK capacity rather than routing to cheaper Malaysia/Johor.

Chokepoints / single-source dependencies:

1. Grid power (CLP) — the industry-level chokepoint. 4–5yr lead times; SUNeVision's secured allocations are an incumbency moat, but it is hostage to one utility and to government power-policy.
2. Parent (SHKP) for land — internalised chokepoint; benign while interests align, a governance risk if they don't (Lens 13).
3. Single-market geography — 100% Hong Kong. Unlike GDS/DayOne (pan-Asia) or Equinix/DLR (global), SUNeVision has zero geographic diversification. Concentrated on one city's power, land, regulatory and cross-border-data regime.

Names present, chokepoints marked — lens satisfied.

Lens 3 · Competitive Advantages (moats)

The moat is real, specific, and unusually structural for a mid-cap — but it is a local-monopoly-adjacent land-and-power moat, not a technology moat.

1. Scarce, permitted, powered land in the world's most supply-constrained major DC market. HK grid-connection lead times of 4–5 years mean the moat is time: a competitor with capital still needs half a decade to light a comparable site. SUNeVision + SHKP already hold the land and the substation ties. This is the closest thing to a regulatory moat in infrastructure.
2. The SHKP parent — balance sheet, land bank, patient capital. SUNeVision can fund HK$15B+ (MEGA IDC) and "fast-track" Phases 2 & 3 because it sits under Hong Kong's largest developer with a >40% family-controlled anchor. Standalone neoclouds must raise dilutive equity/debt at every turn; SUNeVision does not. That is a durable cost-of-capital and optionality advantage.
3. Interconnection / carrier-neutral network effect. Two decades as the neutral meeting point means ~100+ carriers and the major clouds are already in the buildings. New tenants come for the ecosystem; incumbents won't move their cross-connects. Classic colo switching-cost + network-effect moat (the Equinix playbook, at HK scale).
4. Incumbency / market share (~20% of installed IT capacity, #1 by operational core-&-shell power at ~150MW). Scale → better cross-connect density → pricing power.

Bargaining power: Over customers — high for existing interconnected tenants (they're locked to the ecosystem), moderate for greenfield hyperscale (clouds are large, sophisticated, and can play HK vs. Johor/Singapore). The MEGA-IDC "yield-focused approach" — declining to chase orders on price and accepting slower near-term fill (Lens 5) — is management asserting pricing discipline, i.e. claiming they have the power to wait. Over suppliers — low vs. CLP (utility monopoly), structurally high vs. land (internalised via parent), moderate vs. contractors.

The moat's limit: it does not travel. It is a Hong-Kong moat. The bear (Lens 13) is precisely that a single-city land-and-power moat is worth less in a world where AI compute can be sited in cheaper, faster-to-power neighbouring markets, and where HK's cross-border-data role is a geopolitical variable, not a constant.

Lens 4 · Segments

segments.csv is empty → all figures ``, no research-layer grounding.

By product line (FY2025, yr-end Jun-2025):

Trend: the DC segment has increased its share over time as the campuses filled and MEGA IDC came online; ELV is a shrinking-relative tail. Group revenue trajectory (all ``, HKD M, FYE Jun):

*Note the EBITDA discrepancy: stockanalysis computes FY25 EBITDA at HK$1,895M; the company's own FY25 release reported EBITDA HK$2,128M (+15%). The gap (~HK$233M) is almost certainly definitional — the company likely includes DC-business EBITDA before central/ELV drag or on a segment basis, vs. stockanalysis's group calc. Surfaced, not silently reconciled. I use the company figure for margin narrative and the stockanalysis series for trend consistency.

By geography: ~100% Hong Kong. There is no geographic segment to break out — which is the story (Lens 2/13).

The deceleration to watch: FY2025 revenue growth decelerated to +9.9% from +14.0% — the first sub-double-digit print in the series. Bulls read this as the temporary air-pocket before MEGA IDC fills (1H-FY26 growth was soft on "delayed order confirmations" as the company held the line on yield ). Bears read it as the top of the cycle for a maturing single-market operator. The whole call hinges on which reading is right — and the July-2026 anchor-order news (Lens 5) is the first hard evidence for the bull side.

---

Phase B — Measure performance

Lens 5 · Earnings Result

Two prints to read: the FY2025 full year (Jun-2025, reported Aug-2025) and the 1H-FY2026 interim (Dec-2025, reported 25 Feb 2026).

FY2025 (yr-end 30 Jun 2025):

• Revenue HK$2,938M, +9.9%. Beat/miss vs consensus: n/a — pre-print consensus not sourced.
• EBITDA HK$2,128M, +15% (company) — EBITDA growing faster than revenue = operating leverage as fixed DC costs got spread over higher fill.
• Profit attributable HK$979M, +8%; EPS HK$0.24.
• Operating cash flow HK$2,063M, +23% — the standout line. Cash generation running well ahead of accounting earnings (heavy D&A on the asset base), the healthy signature of an infrastructure annuity.
• Final dividend HK$0.12/share; full-year DPS ~HK$0.24 est. → ~4.9% yield at HK$4.85.

1H-FY2026 (six months to 31 Dec 2025):

• Net profit HK$531.0M, +9.7% (vs HK$484.0M prior-year H1). Basic EPS HK$0.1301.
• Prior-year comparator context: H1-FY25 (to Dec-2024) revenue HK$1,469.9M, DC-business EBITDA HK$1,059.0M.
• Interpretation: steady high-single-digit profit growth, not the acceleration a fully-let MEGA IDC would produce — because Phase 1 was still ramping (see below). The soft top-line was explicitly attributed to "delayed order confirmations, as the company maintains a yield-focused approach". Management is choosing margin over speed.

The operational fact that dominates the whole dossier (July-2026 EGM Q&A):

• MEGA IDC Phase 1 is currently ~30% occupied. A single large anchor customer has committed such that occupancy rises to ~70% once fully implemented, with customer move-in commencing toward year-end 2026.
• Management: "robust and sustained demand," "several sizable orders underway," "a major order imminent," supported by improved GPU supply in the region.
• Phases 2 & 3 are being fast-tracked to commence operations by 2028.

This is the de-risking event the stock has been waiting for, and it is dated. Phase-1 fill from 30%→70% is a step-change in the highest-incremental-margin capacity the company owns; because DC EBITDA drops through at very high incremental margins once the shell is built and powered, a Phase-1 fill of that magnitude should show up as an EBITDA acceleration through FY2027.

Balance-sheet flags: third-party screens show current ratio ~0.79 and Debt/Equity ~3.38 with ROE 19.5% / ROIC ~4.1%, and one outlet headlined a "somewhat strained balance sheet". Caveat strongly: a D/E of 3.38 is implausibly high for an EBITDA-rich colo landlord and almost certainly conflates lease liabilities and/or is a data-vendor artifact — treat as unverified; the HK$2.06B operating cash flow and investment-grade-parent backing argue against genuine balance-sheet stress. The real balance-sheet question is funding the HK$15B+ MEGA IDC + fast-tracked Phases 2/3 — near-term FCF is negative-to-thin because capex is front-loaded against rent that arrives later. Precise FY25 net debt/gearing: n/a (HKEX PDF not fetched; EDGAR N/A).

Market reaction / what's priced in: the stock is down ~39% over the trailing year and roughly halved from its Feb-2025 all-time high (~HK$8.4–10.4). The market has been pricing the delay, not the demand — exactly the setup where a dated fill catalyst matters.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts/ on the shelf; HK small/mid-caps rarely post clean English transcripts. Sentiment is reconstructed from results releases and the EGM Q&A ``.

Tone arc over the last ~18 months:

• FY2024 (Aug-2024): confident, growth-mode — MEGA IDC on schedule, double-digit revenue growth, "leading position."
• 1H-FY2026 (Feb-2026): defensive / patient. The message shifted to explaining slower growth as a choice — "yield-focused approach," "delayed order confirmations." Management stopped implying imminent fill and started managing expectations on timing. This is the trough of sentiment and it maps to the trough in the stock.
• July-2026 EGM: re-acceleration / vindication. The vocabulary flips back to "robust and sustained demand," "major order imminent," "fast-tracking Phases 2 & 3," and — crucially — a specific, quantified occupancy path (30%→70%) with a dated move-in (year-end 2026). Management moved from narrative reassurance to numeric commitment.

Recurring phrases: "carrier-neutral," "premium capacity," "yield-focused," "AI and cloud adoption," "largest data centre in Hong Kong." What they stopped saying: the confident near-term fill language of FY24 disappeared in early-FY26 and only returned, in quantified form, at the July-2026 EGM. The single most important sentiment signal in the file is that management chose to put a number (30→70%) and a date (YE-2026) on the anchor deal at the EGM — infrastructure managers don't hand shareholders a specific occupancy jump unless the contract is signed.

Lens 7 · Comps

Peer set = global/regional carrier-neutral colo + hyperscale DC landlords. Multiples are `` with source/date, or n/a. No multiple is fabricated.

SUNeVision's own multiple — arithmetic shown:

• Mkt cap = 4.08B sh × HK$4.85 = HK$19.8B.
• Net debt not precisely sourced; sensitivity: at net debt HK$2–6B, EV ≈ HK$21.8–25.8B.
• On company-reported FY25 EBITDA HK$2,128M → EV/EBITDA ≈ 10.2x–12.1x across that net-debt range.
• Trailing P/E ≈ 20.2x (19.8B / 0.979B); forward P/E ≈ 18–19x on annualised 1H-FY26.

The comp verdict — this is the whole thesis in one line: SUNeVision trades at ~10–12x EV/EBITDA versus Equinix ~24x, DLR ~23x, and its own house-broker DBS target multiple of 22x. Even against the cheapest growth peer (GDS ~15x) it sits at a discount. A ~50% discount to global colo is partly justified (single-market concentration, related-party governance, HK geopolitical discount, and the utilization overhang) and partly the opportunity (if MEGA IDC fills, the same asset re-rates toward the Keppel-DC-REIT / GDS band). The multiple gap is the trade.

Lens 8 · Stock-Price Catalysts (moves >5%, ~last 5 yrs)

Precise dated intraday move history not fully sourceable web-only; the pattern is clear ``:

• 2020–2021 — GEM→Main-Board re-rating + DC secular bid. The stock compounded as the market re-priced HK DC scarcity; low double-digit revenue growth rewarded.
• Feb 2025 — all-time high HK$10.36 (2025-02-27). Peak AI-datacentre euphoria; SUNeVision bid as the listed HK AI-infra proxy.
• Mar 2025 — MEGA IDC Phase 2 initiation announced. Signalled confidence + HK$15B+ commitment.
• Mar 2025 → mid-2026 — the ~50%+ derate. From ~HK$10.4 to ~HK$4.24 low. Drivers: (a) 1H-FY26 revenue deceleration / "delayed order confirmations" — the market punished the timing gap between capex-out and rent-in; (b) broad AI-infra multiple compression through late-2025/2026; (c) HK small-cap risk-off and rate/geopolitics.
• July 2026 — EGM anchor-order disclosure (30%→70% Phase-1 fill, YE-2026 move-in). The first hard positive catalyst in over a year.

What the market actually reacts to for this name: not the quarterly EPS beat (steady high-single-digit, rarely surprising) but the leasing/fill narrative on new capacity — the timing of hyperscale commitments against a heavy capex programme. This is a pre-let vs. capex stock. It sold off on "orders delayed" and should re-rate on "orders landing." Secondary reactions: AI-infra sector beta and HK macro.

---

Phase C — Judge people & books

Lens 9 · Management

A parent-controlled board with a two-year-vacant CEO seat — the central people risk.

• Chairman: Kwok Ping-luen, Raymond — Chairman & MD of Sun Hung Kai Properties, Hong Kong's largest developer; Chairman of SUNeVision. Deep property/capital-allocation pedigree; the SHKP flywheel (land → recurring income) is exactly the strategy SUNeVision executes. Interests aligned with a >40% family-controlled parent.
• CEO: VACANT since 22 Sept 2023. Raymond Tong (CEO from 2018; ex-COO of Maxim's) resigned to relocate to Singapore; the company said it was "identifying a suitable candidate". As of the 2025 disclosures the seat was still being covered by committee — Vice-Chairman Fung Yuk-lun, Allen + then-COO Chan Man-yuen, Martin sharing the duties. A ~3-year CEO vacancy at a company deploying HK$15B+ into a bet-the-franchise campus is a real governance flag — it signals either that the parent runs the company as a division (plausible, given SHKP control) or genuine difficulty attracting a standalone chief. Either reading dents the "independent operator" case.
• Executive Directors include Kwok (Chairman), Fung (Vice-Chairman), Tung Chi-ho, Eric, and Chan (COO).
• COO succession: Martin Chan retires 27 Feb 2026 — a second senior-operator departure inside the capex-heavy window. Who runs day-to-day operations through the MEGA IDC ramp is a live question.

1. Track record: the franchise delivered — 15 straight years-ish of revenue growth to HK$2.9B, #1 HK share, a HK$15B campus funded without distress. But it is hard to separate management skill from the SHKP land-and-capital tailwind; this is a company that succeeds substantially because of its parent.
2. Tenure & skin in the game: board average tenure ~17.3 years — deeply entrenched, pro-stability, arguably not independent. Insider ownership is effectively SHKP's controlling stake (SHKP is the "substantial shareholder" per SFO Part XV); precise % and free float n/a, but free float is the well-understood minority slice under a >40%-family-controlled parent.
3. Capital allocation: the defining decision is MEGA IDC — HK$15B+ into the single most supply-constrained major DC market, funded off parent strength. Bold and on-strategy; the "yield-focused, won't-chase-orders-on-price" stance shows discipline. ROE ~19.5% is healthy; ROIC ~4.1% is depressed by the in-build, not-yet-earning capex — the number that should climb as MEGA IDC fills, and the single cleanest KPI to track the thesis.
4. Red flags: (a) the CEO vacancy; (b) continuing connected transactions with the parent (land/construction) — governance-sensitive by construction, though independently approved; (c) senior-operator churn (Tong out, Chan retiring).
5. Archetype: subsidiary-of-a-controlling-family-conglomerate, run for long-duration recurring income. Implication: extremely patient capital and staying power (great for a decade-long infra build), but minority holders ride behind SHKP's priorities and should not expect aggressive shareholder-return financial engineering.

Lens 10 · Forensic Red Flags

Regulatory findings (from regulatory/regulatory-findings.md, generated 2026-07-07): SUNeVision has no SEC CIK — not a US filer, no EDGAR enforcement search possible; total SEC findings: 0 (SEC_LR, SEC_AAER). Non-SEC web search ("SUNeVision Holdings" (FTC OR DOJ OR consent decree OR settlement OR fine OR penalty) enforcement) returned no material enforcement hits across FTC/DOJ/HK-SFC in the searches run. 10-K Item 3 equivalent is n/a (HK filer; HKEX annual-report "Legal Proceedings" not fetched web-only). Conclusion: no material regulatory or legal findings surfaced — verified via SEC EDGAR EFTS (LR, AAER, = 0) and web search as of 2026-07-07; the HK-listing legal-proceedings disclosure was not machine-read and is an open item.

Accounting-risk scan (web-only — every item flagged as unverified against primary filings):

• Related-party / connected transactions (HIGHEST-priority item). Land, buildings and construction services transact with the controlling parent SHKP under continuing connected transactions, re-approved by independent shareholders at the 30-Jun-2026 EGM. HK Listing Rules force independent-shareholder approval and annual caps — a genuine control — but the economics (are intra-group land/build prices arm's-length?) can only be judged from the circular, not sourced here. This is the item a forensic analyst must read the connected-transaction circular for. Open item.
• Revenue recognition: colo rent + power pass-through is straightforward; the ELV/systems-integration line is project-based (percentage-of-completion risk), but at ~8% of revenue it's immaterial to group quality.
• Cash flow vs. earnings — a positive tell: operating cash flow (HK$2,063M) runs well above net profit (HK$979M), the normal signature of a heavily-depreciated asset base, not an aggressive-accrual red flag. This is the reassuring direction.
• Capitalised interest / capex during build: with HK$15B+ MEGA IDC in construction, watch for interest capitalisation and the D&A start-date as Phase-1/2 assets are placed in service — it will swing reported operating income. Not quantifiable web-only.
• Balance-sheet screens (current ratio ~0.79, D/E ~3.38): flagged, but treated as unverified / likely lease-liability-inflated or a data artifact — inconsistent with the cash generation and parent backing. Do not underwrite on these numbers without the audited balance sheet.
• SBC flattering non-GAAP: no evidence HK-mid-cap-style; n/a.

Net forensic read: the accounting risk surface is low-to-moderate and dominated by one structural issue — the parent related-party nexus — which is disclosed and shareholder-gated rather than hidden. No enforcement history. The honest caveat: this scan is web-only; the connected-transaction circular and audited balance sheet are the two documents that would move a forensic opinion, and neither was machine-read.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection

Built bottom-up off the FY2025 actuals + the 1H-FY26 run-rate + the EGM occupancy path. All outputs ``; no consensus EPS line was sourceable, so n/a where I can't ground an input. Fiscal years end 30 June. No forecast.ts create (unattended --watchlist rule).

Anchors: FY25 revenue HK$2,938M, EPS HK$0.24, net profit HK$979M; 1H-FY26 net profit HK$531M (+9.7%), EPS HK$0.1301 [all web, sourced above]. Shares ~4.08B.

FY2026 (in progress, ends Jun-2026) — BASE:

• H1 done at HK$531M NP; H2 broadly similar (MEGA-IDC anchor fill only begins YE-2026, so it barely helps FY26). Full-year NP ≈ HK$1,050–1,080M; EPS ≈ HK$0.26–0.27. Revenue ≈ HK$3.1–3.2B (~+7–9%).

FY2027 (ends Jun-2027) — the inflection year:

• Base: Phase-1 fill 30%→70% earns for most of FY27 (move-in from YE-2026); high-incremental-margin capacity drops through. Revenue HK$3.5–3.8B (~+15–20%); with operating leverage, NP HK$1.25–1.40B; EPS ≈ HK$0.31–0.34.
• Bull: anchor fully in plus the "several sizable orders" and the remaining two Phase-1 floors let → revenue >HK$4.0B, EPS HK$0.36–0.40.
• Bear: move-in slips into FY28, orders "delayed" recurs → revenue ~HK$3.3B (+high-single-digit), EPS ~HK$0.28, and the derate persists.

FY2028 (ends Jun-2028):

• Phases 2 & 3 begin operations (per EGM fast-track) — but new capex re-loads D&A/interest, so early-Phase-2 is a drag before it earns. Base EPS ≈ HK$0.35–0.40; the shape depends entirely on Phase-1 stabilisation funding Phase-2 dilution.

Base-case Brier statement (logged conceptually, NOT written to forecast.ts): "1686.HK FY2027 (Jun-2027) EPS ≥ HK$0.31, p≈0.60" — hinges on the anchor moving in on schedule (YE-2026) and no fresh order-confirmation slippage.

Valuation bridge: at ~HK$4.85 / ~18–19x forward P/E / ~10–12x EV/EBITDA, the market is pricing the bear-to-base path. DBS's HK$10.0 target = 22x EV/EBITDA on FY6/27F, i.e. the base-case fill plus a re-rating toward global-colo multiples — roughly a double from spot. The gap between HK$4.85 spot and ~HK$9.5–10.5 consensus is the clearest expression of the "show-me" discount.

Lens 12 · Bull vs Bear

Bull case. SUNeVision is the only listed pure-play on the scarcest real estate in the AI build-out: powered, permitted Hong Kong data-centre capacity, in a market where a rival needs 4–5 years just to get grid power. It owns ~20% share, sits under a parent that can fund HK$15B+ without dilution, and has a two-decade interconnection moat that locks tenants in. The stock has halved from its high on a timing problem — capex out ahead of rent in — that the July-2026 EGM just put a date on: Phase-1 30%→70% with move-in from year-end 2026, "several sizable orders underway," Phases 2 & 3 fast-tracked to 2028. As MEGA IDC fills, ROIC climbs off its ~4% in-build trough, EBITDA re-accelerates on high incremental margins, and a ~10–12x EV/EBITDA asset re-rates toward the GDS-to-Keppel-DC band (15–20x) — DBS's HK$10.0 (22x, ~2× spot) prices exactly that. You are buying a scarce infrastructure annuity at half the global multiple at the moment the fill catalyst becomes visible and dated.

Bear case (2–3 permanent-impairment risks).

1. The moat doesn't travel and HK's gateway role is a geopolitical variable. A single-city land-and-power moat is worth less in a world where AI compute migrates to cheaper, faster-to-power Johor / Malaysia / Japan, and where HK's China-facing cross-border-data role can be re-priced by policy overnight. Concentration = 100% one city.
2. The capex is a bet-the-franchise duration mismatch. HK$15B+ (plus fast-tracked Phases 2/3) is deployed now against rent that arrives over years; if hyperscale demand for HK specifically softens or "delayed order confirmations" recur, the company owns a very expensive, under-let campus and the FCF hole widens. The 1H-FY26 miss showed this risk is live, not theoretical.
3. Governance / control discount is structural. ~3-year CEO vacancy, senior-operator churn, and a >40%-family-controlled parent with continuing connected land/construction transactions mean minorities ride behind SHKP — a permanent valuation haircut vs. cleaner-governed global peers.

Pre-mortem (18 months out, thesis broke — what happened?): The anchor tenant's move-in slipped again (GPU/fit-out delays), the "imminent" second order never signed, and 1H-FY27 printed another single-digit deceleration. The market concluded MEGA IDC is a decade-long fill, not a two-year one, re-based EPS estimates down, and the stock sat at ~HK$4 on a "value trap with heavy capex" narrative while Johor soaked up the marginal APAC hyperscale demand.

Are multiples too high? No — at ~10–12x EV/EBITDA / ~18–19x forward P/E, expectations are low, not high. The risk here is fundamental (does HK capacity fill?), not valuation (you are not overpaying for growth).

Contrarian view — what the market is refusing to see: the tape is still pricing the delay (down ~39% on the year) while management has already flipped to a quantified, dated fill. Markets consistently misprice the gap between "orders delayed" and "orders landing" for pre-let infrastructure — the derate treated a timing slip as a demand failure. If YE-2026 move-in happens on schedule, the "show-me" discount unwinds fast.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case:

• What structurally breaks the model: loss of Hong Kong's privileged position. The entire premium rests on HK being the scarce, low-latency, China-facing gateway. If (a) hyperscalers decide Johor/Malaysia/Japan is "good enough" for the China-adjacent workload at half the cost and a fraction of the power-lead-time, or (b) cross-border-data policy tightens, the scarcity moat evaporates and SUNeVision is a single-city landlord with a huge capex bill. There is no geographic Plan B — unlike GDS/DayOne (pan-Asia) or Equinix/DLR (global).
• Revenue concentration: the near-term story now hinges on one anchor tenant taking Phase-1 from 30%→70%. What if that single customer renegotiates, delays, or (at renewal) leaves? Hyperscalers are ruthless counterparties with optionality across the region. Concentration risk sits at both the market (100% HK) and the tenant (one anchor) level.
• Why the moat is weaker than bulls think: it's a land-and-power moat, not a technology or brand moat. Land+power can be replicated by any deep-pocketed operator willing to wait out the grid queue — and DayOne (US$4.5B Series C, pan-Asia, HK sites), AirTrunk, Vantage, Digital Realty and Equinix are all building in HK/TKO. Incumbency buys time, not permanence.
• Most dangerous competitor bulls underestimate: DayOne (ex-GDS). US$4.5B raised, ~480MW in service/under-construction + ~590MW reserved pan-Asia, heading toward a ~US$20B IPO, with HK assets (HK1/HK6). It has capital, scale, and geographic diversification SUNeVision lacks and can cross-subsidise HK pricing from a pan-Asia base. If DayOne decides to buy HK share on price, SUNeVision's "yield-focused" discipline gets tested hard.
• Worst capital-allocation / governance concerns: the continuing connected transactions with the controlling parent (land, construction) — related-party by construction; even with independent approval, minorities can't verify arm's-length pricing web-only. Add the ~3-year CEO vacancy and senior-operator churn: who is actually accountable for HK$15B of capital?
• Assumptions that must hold for HK$4.85: (i) MEGA-IDC Phase-1 fills on schedule (YE-2026 move-in, no further slip); (ii) HK retains its gateway premium; (iii) the parent keeps funding without dilution; (iv) HK power/grid allocations hold. Break any one and the base case cracks.
• If growth disappoints 20–30%: an operator carrying HK$15B+ of in-build capex with a stalled fill would see FCF stay negative longer, ROIC stuck near ~4%, EPS estimates cut, and the multiple compress further — a plausible path to ~HK$3.5–4.0 and a multi-year "capex value trap." The Feb-2025→2026 derate is the template.
• Single scenario that permanently impairs the business: a structural loss of HK's cross-border/gateway role (policy or hyperscaler-preference driven) that strands a purpose-built HK$15B+ HK campus at low occupancy. Plausibility: low-to-moderate — HK's grid-scarcity moat and existing cloud presence are real — but it is the tail that turns the whole thesis into an impairment, and it is not zero given the geopolitical overlay.

Lens 14 · Management Questions (15, ordered by information value)

1. The anchor customer taking MEGA-IDC Phase 1 from ~30% to ~70% — what is the contracted term, the committed MW/floor, and the earliest break/renewal date, and how much of that ~70% is legally committed vs. in the pipeline?
2. On what date does that anchor's rent actually start hitting the P&L, and what is the ramp schedule (MW online by quarter through FY2027)?
3. What is the per-kW / per-MW effective monthly rent you are achieving on new MEGA-IDC leases vs. your legacy campuses — is the "yield-focused" stance holding pricing, or just deferring volume?
4. How is the HK$15B+ MEGA-IDC programme (plus fast-tracked Phases 2/3) funded — group debt, parent support, or operating cash — and what is the peak net-debt / net-debt-to-EBITDA you're willing to run through the build?
5. What caused the 1H-FY26 "delayed order confirmations," and what specifically changed to give you confidence in the "several sizable orders underway / major order imminent" language at the EGM?
6. When will the CEO seat be filled, what profile are you seeking, and how are the CEO's responsibilities allocated between the Vice-Chairman, the Board and (post-Feb-2026) the COO function today?
7. What share of your prospective pipeline is AI/GPU-dense workloads vs. traditional cloud/enterprise, and how does that change your power-density and cooling capex per rack?
8. How exposed is demand to GPU supply — you cited "improved GPU supply in the region" as a demand driver; what happens to your leasing velocity if GPU allocation to HK tightens again?
9. Walk through the arm's-length basis and independent valuation underpinning the continuing connected transactions with SHKP for land and construction — what would these cost from a third party?
10. What is your secured grid-power allocation (MW) from CLP across the portfolio, and how much of Phases 2/3's >280MW target is already power-secured vs. still in the utility queue?
11. How do you assess the risk that hyperscalers substitute HK demand to Johor/Malaysia/Japan on cost and power-lead-time, and what keeps the China-facing workload anchored in Hong Kong?
12. What is DayOne / AirTrunk / Vantage's live HK capacity and pricing doing to your win-rates on new hyperscale mandates?
13. What is the expected stabilised ROIC on MEGA IDC at full fill, and by which fiscal year do you expect group ROIC to recover from its current ~4% in-build level?
14. What is your capital-return policy through the build — will the dividend be held/grown, or subordinated to funding Phases 2/3?
15. What contingency exists if a major anchor delays or exits mid-build — how re-lettable is purpose-built hyperscale space to a different tenant, and over what timeframe?

---