GDS Holdings
Build-own-lease colocation: develops data centers in Tier-1 China metros and leases capacity under long-term contracts (retail colocation + hyperscale wholesale) to Chinese cloud/internet giants; capital-intensive, debt-funded, with asset monetization (C-REIT, DayOne stake sales) as a deleveraging lever.
Earnings, margins, COGS & capex
A capital-intensive infrastructure operator that turned GAAP-profitable in FY2025 (net income RMB959.4M / US$137.2M, first profitable year) — but the profit was carried by ~RMB3.08B of non-recurring gains (RMB2,364.1M C-REIT deconsolidation gain + RMB715.9M DayOne equity-method dilution gain); ex those, continuing operations would have posted a loss, and Q4'25 alone was a RMB462.8M net loss on a RMB1,561.2M lease-asset impairment. Q1 2026 headline strength — RMB2,652.1M net income, 57.9% adj. EBITDA margin — is again heavily inflated by one-time DayOne-related gains (RMB2,136.5M equity-method dilution gain); normalized revenue growth was only +7.9%. The genuine positives are record ~200MW quarterly bookings signaling an AI demand inflection, and an off-balance-sheet ~19.9% stake in international arm DayOne now exploring a ~$20B dual IPO.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~0¢ is cost of goods and ~0¢ operating expense, leaving ~100¢ of operating profit (~2025¢ net).
Revenue trend
Margins
improving, but Q1 boost is one-time-driven
structurally high (infra economics); core ~48%
near breakeven, volatile (Q4'25 impairment-driven loss)
first profitable year FY2025 but non-core-driven; not sustainable at Q1 level
COGS structure
Dominated by depreciation of data-center assets, electricity/power costs, and facility rent/O&M — a high-fixed-cost base where utilization ramp is the primary margin lever. Power cost and PUE (energy efficiency) are the swing variables and are regulated under China's energy 'dual-control' policy.
Capex
RMB4,610.6M actual in FY2025 (~40% of revenue); ~RMB9,000M guided for FY2026 (~71% of revenue) for data-center construction and equipment in China; historically the reason for chronic negative FCF. Management is shifting toward asset-light/monetization (C-REIT, third-party capital in DayOne) to fund growth without further ballooning on-balance-sheet debt.
Latest earnings
Beat — revenue RMB3,367.1M vs. ~RMB3,010M consensus; diluted EPS RMB10.58/ADS (US$1.53) vs. low-single-digit-cents consensus (surprise driven by one-time DayOne dilution/C-REIT gains, not core ops); stock reaction muted/mixed
FY2026 reaffirmed: revenue RMB12,400M-12,900M; adjusted EBITDA RMB5,750M-6,000M; capex ~RMB9,000M
- Net new bookings
- ~200MW (record quarter)
- Utilization rate
- 77.3%
- Area in service
- 674,269 sqm
- Area utilized
- 520,929 sqm
- Net income
- RMB2,652.1M (US$384.5M), +247.1% YoY
- Adjusted EBITDA
- RMB1,948.7M (US$282.5M), +47.2% YoY
- DayOne stake sale
- US$385M realized; residual ~19.9% stake carried ~US$2.2B (Series C price)
Growth drivers
- AI-driven demand inflection in China — record ~200MW net new bookings in Q1 2026, described as highest-ever single quarter
- Scarcity of power/land in Tier-1 metros (Beijing/Shanghai/Shenzhen) supporting pricing and pre-commitment
- Utilization ramp (77.3% in Q1'26) on already-built capacity converting to revenue and EBITDA
- DayOne international arm (SE Asia, Japan, Europe) — deconsolidated, GDS holds ~19.9% stake, exploring ~$20B dual IPO in US + Singapore
- Balance-sheet deleveraging via DayOne stake sales (US$385M realized in Q1'26) and China C-REIT monetization
Bull & bear
GDS is the scarce, dominant China data-center platform hitting an AI-demand inflection (record 200MW bookings), turning profitable and deleveraging, while its ~19.9% DayOne stake — an international operator eyeing a ~$20B IPO — is a large under-the-radar value catalyst the China-discounted stock isn't pricing.
- Record ~200MW single-quarter bookings signal an AI-led demand turn after years of soft China colocation growth
- Turned GAAP-profitable in FY2025 for the first time; utilization at 77.3% with a large built-but-not-yet-monetized capacity base to ramp
- DayOne (~19.9% stake, SE Asia/Japan/Europe) exploring a ~$20B dual IPO (US + Singapore) — a sum-of-parts catalyst plus non-dilutive capital, sitting outside China regulatory constraints
- Active deleveraging: US$385M DayOne stake sale realized, C-REIT monetization, asset-light pivot to fund growth without more balance-sheet debt
- Scarce Tier-1 power/land creates a durable supply moat and pricing power as new capacity is permit-constrained
- HKEX dual listing (9698) materially de-risks a US delisting scenario
Strip out the one-time DayOne/C-REIT gains and GDS is a highly-levered (~US$4.5B net debt), FCF-negative operator growing core revenue ~8% with core operations that would still be loss-making, wrapped in a China VIE structure exposed to geopolitical, regulatory, and refinancing risk — a lot of debt and macro fragility for the price.
- ~RMB46B total debt / ~RMB31B (US$4.5B) net debt makes equity a thin, rate-sensitive residual; refinancing risk is real
- Free cash flow persistently negative (~-RMB1.25B FY2025) — the model doesn't self-fund
- FY2025 profit and Q1'26's blowout EPS are both one-time-gain-driven (~RMB3.08B of C-REIT + DayOne gains in FY2025); normalized revenue grew only +7.9% and core operations are ~breakeven-to-loss
- Concentration in a few Chinese internet giants whose capex is cyclical and policy-sensitive
- VIE/offshore ADS structure over PRC assets, plus HFCAA and US-China tension, imposes a persistent valuation discount and tail delisting risk
- China energy dual-control caps new Tier-1 supply and raises power costs; domestic non-AI demand remains soft
- DayOne value is real but uncertain in timing/dilution and outside ADS holders' direct control (GDS holds only ~19.9% and does not control it)
What it is worth
Sum-of-parts (China core DC on EV/EBITDA + DayOne stake) cross-checked vs. multiples
China demand stays soft ex-AI, leverage/refinancing pressure and rate sensitivity weigh, DayOne IPO slips or dilutes, and the VIE/geopolitical discount persists — retest toward the ~$27 52-week low or below.
Mid-to-high single-digit core revenue growth, adj. EBITDA in the guided RMB5.75-6.0B range, gradual deleveraging; stock range-bound around ~$30 with DayOne optionality as the main catalyst.
AI booking momentum sustains double-digit core growth, DayOne IPOs near ~$20B crystallizing a multi-billion stake, and deleveraging re-rates the China core toward peer multiples — upside toward/above the ~$48 52-week high.
Equity ~$5.9-6.0B; EV ~$10B+ incl. ~US$4.5B net debt. On FY2026 guided adj. EBITDA ~RMB5.9B (~US$0.8B), EV/EBITDA ~12-13x — a discount to US peers (EQIX/DLR mid-to-high teens+) reflecting China/VIE/leverage risk. Headline P/E is not meaningful (one-time-gain-inflated FY2025/Q1'26 earnings). The swing factor is the ~19.9% DayOne stake, potentially worth ~$4B at a ~$20B IPO valuation vs. ~US$2.2B carried (Series C price).
SWOT
Strengths
- Largest carrier-neutral data-center platform in China with scarce, hard-to-replicate footprint in supply-constrained Tier-1 metros
- Long-term contracts with high-quality anchor tenants (Alibaba, Tencent, ByteDance) delivering revenue visibility and high utilization
- Structurally high adjusted-EBITDA margins (~47% FY2025, ~48% core) typical of infrastructure economics
- Optionality from DayOne — a ~19.9% stake in a fast-growing international operator exploring a ~$20B IPO, largely off the market's China-discounted valuation
Weaknesses
- Very high leverage (~RMB46B total debt, ~RMB31B/US$4.5B net debt) making equity value rate- and refinancing-sensitive
- Chronically negative free cash flow (~-RMB1.25B FY2025) — growth is not self-funding
- Core GAAP profitability marginal — FY2025 net income and Q1'26 headline profit are both one-time-gain-driven (C-REIT deconsolidation + DayOne dilution), obscuring modest ~7.9% normalized growth
- Customer concentration among a handful of Chinese internet/cloud giants whose own capex cycles are volatile
Opportunities
- China AI capex super-cycle lifting hyperscale demand and bookings beyond the historically soft domestic colocation market
- DayOne IPO crystallizing a large, currently-underappreciated ~19.9% stake and providing non-dilutive growth capital
- Asset monetization (C-REIT) recycling capital to fund builds and cut leverage
- Pricing power as new Tier-1 supply is constrained by energy/permit policy
Threats
- US-China geopolitical and delisting (HFCAA) residual risk — VIE/offshore structure over China assets limits legal claim for ADS holders
- China energy 'dual-control' and data-sovereignty regulation capping new capacity and raising power/PUE costs
- Rising/refinancing interest cost on a heavily levered balance sheet
- Soft China macro / weaker enterprise-cloud demand outside the AI vertical
- FX mismatch — RMB revenue vs. USD-reported results and USD-denominated debt exposure
Moats, dependencies & bottlenecks
Moats
Scarce Tier-1 power/land footprint in supply-constrained China metros Permit/energy 'dual-control' limits new Beijing/Shanghai/Shenzhen supply, entrenching incumbents
Moderate-Strong Largest independent China platform; interconnection creates network effects for tenants
Switching costs on mission-critical long-term colocation contracts High migration cost and multi-year contracts stabilize revenue but tenants have negotiating power at scale
Concentration is both a moat and a dependency risk
Dependencies
Tencent, ByteDance) capex Revenue concentrated in a few tenants whose spending is cyclical and AI-timing-dependent
~US$4.5B net debt; growth and solvency hinge on continued access to capital at manageable rates
Power is a top cost and dual-control policy caps expansion in prime markets
Geopolitical/structural Offshore VIE over PRC assets; HKEX listing partially mitigates delisting tail
A large chunk of the bull case sits in a ~19.9% stake GDS doesn't control
RMB revenue vs. USD-reported results and USD-linked obligations
Advantages
- Largest independent, carrier-neutral data-center scale in China
- Early, dense positioning in scarce Tier-1 markets
- Established hyperscaler relationships and high utilization
- DayOne gives international growth exposure outside China's regulatory ceiling
Weaknesses
- High leverage and negative free cash flow
- Marginal core profitability masked by one-time gains
- Tenant and geography concentration
- Structural China/VIE valuation discount and geopolitical overhang
Bottlenecks
- Power availability and energy-consumption quotas in Tier-1 metros
- Balance-sheet capacity — leverage limits debt-funded growth without asset monetization
- Domestic (non-AI) enterprise-cloud demand remains soft
- Capital recycling pace (C-REIT / DayOne sales) needed to fund build pipeline
Top signals & trends
Top signals
Strongest demand signal in the company's history; AI-led inflection
Inflection, though the RMB959.4M profit is one-time-gain-aided (core ~breakeven-to-loss)
Non-dilutive capital and lower leverage trajectory
Core demand still modest despite AI narrative
Growth not self-funding; reliant on capital markets
Potential large value crystallization for the ~19.9% stake
Trends
Hyperscale demand for GPU-ready, high-density capacity lifts bookings and pricing
Caps new supply (good for incumbents' pricing) but raises cost and limits GDS's own expansion
New capital-recycling channel to deleverage and fund builds
SE Asia/Japan demand from global hyperscalers powers the international arm
Sustains a valuation discount vs. US data-center peers
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Data-center power/cooling/thermal-management equipment supplier
Indirect demand driver — GPU cycle sets AI data-center density and demand
Network connectivity and, for some, competing capacity; also power/fiber partners
Electricity supply is a top cost input and expansion constraint
Anchor hyperscale tenant (Alibaba Cloud)
Major cloud/internet tenant
Large private AI/internet tenant driving GPU-density demand
Cloud/AI tenant
US/global cloud and AI companies anchoring DayOne's SE Asia/Japan capacity
Direct China carrier-neutral data-center peer, also Nasdaq-listed and AI-demand-levered
Former China hyperscale peer, taken private by Bain Capital (2023); competes for the same hyperscale wholesale demand
Global carrier-neutral leader; direct rival to DayOne in SE Asia/Asia-Pacific interconnection
Global hyperscale/wholesale data-center REIT; competes with DayOne for global hyperscaler footprints in Asia
APAC hyperscale specialist (acquired by Blackstone, 2024); a primary DayOne competitor in SE Asia/Japan
Singapore-based pan-Asia operator competing with DayOne
Warburg-backed pan-Asia data-center platform overlapping DayOne markets
State telco data-center arms — large but less carrier-neutral; compete for domestic capacity