SOS Limited signs non-binding memo for Indonesia data center

SOS Limited signed a non-binding memorandum to explore a potential 500MW AI and cloud data-center campus in Indonesia, beginning with an approximately 50MW Phase I. Prospective tenant expressions of interest total roughly 180MW, while the local partner would seek to secure at least 60MW of power capacity before Phase I operations. The project remains contingent on a six-month due-diligence process, definitive agreements, financing and regulatory approvals, with no assurance of completion; SOS plans an equity and project-finance structure if it proceeds.

Analysis

The investable issue is financing optionality, not data-center demand. A 50MW AI-ready buildout would likely require capital broadly comparable with, or exceeding, SOS's stated equity base once land, grid interconnection, cooling, networking and contingency are included; absent contracted tenants and committed debt, the most probable equity-market transmission is dilution rather than NAV creation. The large gap between an aspirational campus headline and a financeable first phase makes any near-term rally vulnerable to a reversal when capital terms emerge.

The proposed power structure creates a less appreciated margin risk: coal-linked electricity costs can leave a wholesale-colocation operator exposed if customer pricing is fixed or escalation clauses are weaker than fuel pass-through. This is especially problematic for AI workloads, where customers require high utilization and predictable power economics but may delay commitments if regional capacity, grid reliability, or cross-border connectivity is unproven. Established Asian operators with existing customers, financing access and operating track records—not SOS—are better positioned to capture any genuine Indonesia capacity shortage.

Over the next days, this is primarily a liquidity-driven microcap event rather than a fundamental rerating catalyst. Over 1-3 months, definitive tenant commitments, disclosed capital expenditure per MW, power tariff pass-through, and binding financing are the gating evidence; without them, a six-month diligence process is more likely to sustain volatility than establish valuation support. Over 6-18 months, successful execution would require project-level leverage and customer prepayments that sharply reduce parent-company dilution.

Contrarian view: the market may overvalue the 500MW headline while underweighting the probability that the realized economic project remains materially smaller, delayed, or transferred into a financing vehicle in which SOS retains limited economics. Conversely, a binding take-or-pay lease with a creditworthy hyperscaler plus non-recourse senior debt would invalidate the bearish financing thesis and could justify a step-change in valuation.

Key Decisions for Investors

• Do not establish a strategic long in SOS on the memorandum alone. Reassess only after binding customer contracts and committed project financing are disclosed; require evidence that Phase I capex is substantially non-recourse and that SOS's required equity contribution is fundable without material dilution.
• For event-driven mandates, treat a headline-driven spike in SOS as a short/watch opportunity only after confirming borrow availability and daily liquidity. Use a tight stop on disclosure of a binding hyperscaler lease or fully committed senior debt; the thesis is a 1-6 month compression of speculative project value rather than an operating short.
• Monitor DCI Indonesia (DCII.JK) and regional data-center infrastructure proxies for confirmation of real capacity demand. If credible contracted capacity emerges, prefer established operators or infrastructure vehicles with operating assets and financing access over SOS; this captures the same thematic upside with lower execution risk.
• Set disclosure alerts for capex per MW, tenant lease duration and take-or-pay terms, electricity pass-through provisions, debt cost/tenor, and SOS's equity commitment. A project economics profile with customer-backed returns above the cost of capital and limited parent recourse would falsify the dilution-led downside case.