VNET Group
Carrier-neutral colocation and wholesale hyperscale data-center leasing (power + space, recurring), plus managed hosting/cloud/VPN services; licensed operator of Microsoft Azure and Microsoft 365 in mainland China. Capital-intensive build-and-lease economics on a Cayman holdco with China VIE operations, US-listed as an ADR on Nasdaq.
Earnings, margins, COGS & capex
Two-engine model: a mature but low-growth retail colocation/managed-hosting base (Q1 2026 retail IDC RMB1.02B, +5.4%) and a fast-scaling wholesale hyperscale segment that is now the growth story. Wholesale revenue +77.4% in FY2025 to RMB3.46B and +58.1% in Q1 2026 to RMB1.06B, lifting group growth to ~20% while GAAP margins compress as new capacity depreciates ahead of full utilization. The economics are capital-intensive and levered; the bet is that pre-leased AI/cloud capacity converts today's capex into durable EBITDA.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~77¢ is cost of goods and ~0¢ operating expense, leaving ~23¢ of operating profit.
Revenue trend
Margins
down from 25.2% YoY (new-capacity depreciation drag)
up from 43.1% YoY
up from 30.4% YoY
FY2026 guide RMB3.55-3.75B, +19.2% to +25.9%
COGS structure
Dominated by depreciation & amortization on data-center assets, electricity/power, rack/rental and bandwidth costs; power and D&A are the swing items and the reason GAAP gross margin compresses during a capacity ramp while cash gross margin rises.
Capex
RMB1.75B in Q1 2026 (US$254.1M), majority to wholesale IDC; FY2026 guide RMB10-12B to deliver 450-500MW of new wholesale capacity. Wholesale capacity in service reached 907MW (committed 869MW) with 516MW under construction at 31 Mar 2026. Funded via debt and operating cash flow; the CATL-affiliate transaction is a secondary share purchase from Shandong Hi-Speed and does not directly inject capital into VNET.
Latest earnings
Revenue +19.8% YoY and adjusted EBITDA +30.6% YoY, both solid; GAAP net loss widened to RMB531.8M, driven largely by RMB486.2M of income tax expense on capital transactions (REIT listings + the CATL-affiliate deal) plus ramp-stage depreciation — an operationally strong, optically noisy print
FY2026 reaffirmed: revenue RMB11.5-11.8B, adjusted EBITDA RMB3.55-3.75B, capex RMB10-12B, 450-500MW delivery
- Wholesale IDC revenue
- RMB1.06B, +58.1% YoY
- Retail IDC revenue
- RMB1.02B, +5.4% YoY
- Wholesale capacity in service / committed
- 907MW / 869MW (31 Mar 2026)
- Wholesale utilization
- 75.7% (31 Mar 2026)
- Capacity under construction
- 516MW
- New wholesale orders YTD 2026
- 517MW (incl. 510MW Greater Beijing)
- Adjusted EBITDA
- RMB891.5M, +30.6% YoY
Growth drivers
- Wholesale hyperscale IDC — 517MW of new orders YTD 2026, incl. a 510MW order from a leading internet customer for the Greater Beijing Area, delivering 2026-2028
- China domestic AI/cloud capex cycle (draft ~2 trillion yuan / ~US$295B five-year national data-center plan) lifting demand for compliant in-country capacity, though the plan is state-carrier-led and mandates ~80% domestic chips
- Rising wholesale utilization (75.7% at 31 Mar 2026 vs 70.1% at 31 Dec 2025) converting built capacity to revenue
- Microsoft Azure / Microsoft 365-in-China operation as a sticky, differentiated managed-services annuity
- Incoming CATL-affiliate anchor shareholder (via secondary purchase) bringing strategic credibility and a power/energy-storage angle for AI-era data centers
Bull & bear
A cheap, domestic-champion AI-infrastructure play: VNET has the land, licenses, footprint and now pre-committed hyperscale orders plus an incoming CATL-affiliate anchor shareholder to ride China's AI/cloud capex wave, at a wholesale-driven ~20% growth rate and an EV/EBITDA discount to peer GDS.
- Wholesale IDC is inflecting — +58% in Q1 2026 and 517MW of new orders YTD, with a ~500MW pipeline delivering through 2028 that underwrites future EBITDA
- Adjusted EBITDA growing faster than revenue (+30.6% vs +19.8%) with utilization climbing to 75.7% — operating leverage is showing
- CATL-affiliate ~38.1% stake (~$942M, closing expected Q4 2026) is strategic validation and a potential power/energy edge for AI-era DCs
- Exposure to China's draft ~US$295B national data-center/AI buildout as a demand tailwind for compliant in-country capacity
- Microsoft-in-China franchise is a durable, hard-to-replicate annuity underappreciated in the multiple
- Street consensus Buy / Strong Buy with a ~$15 average price target vs ~$7.8 ADS
A levered, cash-burning Chinese ADR whose growth is bought with debt-funded capex, exposed to delisting, chip-export and overbuild risk, with GAAP losses and single-customer concentration that make the pre-leased pipeline less de-risked than it looks.
- FY2026 capex (RMB10-12B) roughly equals revenue — free cash flow is deeply negative and growth depends on continued debt/equity funding
- GAAP gross margin fell to 22.9% and net losses persist (Q1 2026 net loss RMB531.8M); the profitability is adjusted-EBITDA, not GAAP or FCF
- Concentration risk: a single leading internet customer accounts for 510MW of the 517MW order book — a demand air-pocket or renegotiation would sting
- Geopolitical tail: HFCAA/ADR-delisting risk and US chip-export controls could starve China AI data centers of accelerators; the national plan's ~80% domestic-chip mandate is itself a supply constraint
- Levered balance sheet into a higher-rate world; net debt several turns of EBITDA
- The CATL deal is a secondary transfer from Shandong Hi-Speed — it validates but does not directly recapitalize VNET, closes only in Q4 2026, and the founder retains control; the strategic story must still execute
What it is worth
EV / forward adjusted EBITDA vs peer GDS, cross-checked against the FY2026 guide. EV ~ $2.2B equity + ~$2B net debt (total debt not disclosed in the quarterly release) ~= ~$4.2B against FY2026 adjusted-EBITDA guide of RMB3.55-3.75B (~$515-545M) ~= ~8x forward EV/EBITDA — a discount to GDS, appropriate for higher leverage, GAAP losses and single-customer concentration.
~$5-6 / ADS
capex burn + leverage bite, a hyperscaler order slips or export controls cap AI demand, and ADR/geopolitical risk re-widens the discount.
~$9-11 / ADS
guidance delivered, utilization grinds higher, modest multiple expansion as EBITDA compounds ~20%+ but FCF stays negative.
~$15-18 / ADS
wholesale pipeline fills at 80%+ utilization, the CATL close catalyzes a re-rate toward GDS-like EV/EBITDA, and China AI-DC demand stays strong (in line with the Street ~$15 average target).
At ~$7.8/ADS the market is paying a below-peer multiple for ~20% top-line growth and a rising-utilization wholesale ramp, effectively pricing in execution, funding and geopolitical risk. Re-rating hinges on the CATL close, the ~500MW pipeline landing on schedule, and free-cash-flow inflection as capacity fills. Net-debt figure is an estimate; total debt not broken out in the Q1 release. Not financial advice; a mainland-China-exposed ADR named here for analytical context only.
SWOT
Strengths
- Largest carrier-neutral IDC operator in China with a nationwide footprint and 20+ year operating history
- Wholesale segment scaling fast (+77% FY2025, +58% Q1 2026) with large pre-committed hyperscale orders de-risking the capex
- Licensed operator of Microsoft Azure and Microsoft 365 in mainland China — a rare, sticky, high-barrier franchise
- Rising utilization (75.7%) and adjusted EBITDA margin (33.1%) show ramp discipline
- Incoming CATL-affiliate anchor shareholder adds credibility and an AI-power/energy angle
Weaknesses
- Capital-intensive and levered — FY2026 capex roughly equals revenue, negative free cash flow, net-debt-financed growth
- GAAP gross margin compressing and persistent GAAP net losses during the build phase (Q1 2026 net loss RMB531.8M)
- Customer concentration in a handful of hyperscalers/internet giants — one 510MW order is a single leading customer
- Legacy retail colocation is low-growth (+5.4% in Q1) and margin-diluting relative to wholesale
- History of governance/leverage overhangs that kept the ADR a laggard for years
Opportunities
- China's multi-year AI/cloud data-center buildout (draft ~US$295B plan) as a demand tailwind for in-country compliant capacity
- Pre-leased ~500MW pipeline (2026-2028) converting to durable recurring EBITDA
- Potential re-rating toward peer GDS if AI demand and the CATL tie-up execute
- Power/renewables + storage optimization with a CATL-affiliate shareholder lowering TCO for AI data centers
- Possible C-REIT / asset monetization to recycle capital (VNET listed two REIT projects in Q1 2026)
Threats
- US-China tension: ADR delisting/HFCAA risk, chip-export controls limiting AI-accelerator supply into China DCs
- China's national AI-DC plan skews facility operation to state carriers (China Mobile/Telecom) and mandates ~80% domestic chips, which may cap the addressable share for carrier-neutral operators
- Regulatory and geopolitical constraints on Chinese tech capex and cross-border data
- Power availability, PUE/energy regulation, and grid constraints on siting new capacity
- Rising rates / financing cost on a levered balance sheet
- Overbuild risk if China DC supply outpaces AI monetization, pressuring wholesale pricing/utilization
Moats, dependencies & bottlenecks
Moats
permitted, power-secured DC land in Tier-1 China metros Power quotas, siting permits and interconnection in Beijing/Shanghai/Greater Bay are genuinely scarce; hard to replicate quickly
Licensed operator of Microsoft Azure / Microsoft 365 in mainland China Regulatory + contractual barrier; a distinctive franchise that anchors enterprise cloud demand
20+ year carrier-neutral interconnection density and enterprise relationships Network-effect of interconnection, but wholesale hyperscale is more commoditized than retail colocation
CATL-affiliate strategic backing (credibility + potential power/energy-storage integration) Secondary-purchase deal closes Q4 2026; strategic value real but unproven until integrated, and it injects no direct capital into VNET
Dependencies
Tencent, ByteDance, Baidu and the 'leading internet customer') Customer / revenue concentration Wholesale order book is dominated by a few large tenants; 510MW from one customer
AI accelerators / chips (NVIDIA and China-compliant alternatives such as Huawei) Supply / demand enabler US export controls on advanced GPUs can throttle the AI compute that fills DCs; China's domestic-chip push is not yet at the leading edge
Capex ~= revenue; growth requires ongoing debt/equity funding, and the CATL deal is a secondary purchase, not a primary raise
Input / regulatory Electricity is the largest variable cost and a siting gate for new capacity
Azure/M365-in-China revenue depends on continuation of the licensing relationship
Regulatory / structural VIE/Cayman ADR structure carries delisting and enforcement tail risk
Advantages
- First-mover scale and scarce permitted power/land in Tier-1 China metros
- Distinctive Microsoft-in-China operating franchise
- Large pre-committed wholesale order book de-risking near-term fill
- Incoming CATL-affiliate strategic shareholder + power/energy angle
- Rising utilization and adjusted-EBITDA operating leverage
Weaknesses
- Negative free cash flow and persistent GAAP net losses
- High leverage into a capital-intensive build cycle
- Customer concentration in a few hyperscalers
- Legacy retail colocation drag on growth and mix
- Geopolitical/ADR structural overhang depressing the multiple
Bottlenecks
- Power availability and energy-efficiency (PUE) quotas gating new-capacity siting
- Access to AI accelerators for tenants — export controls and the domestic-chip transition limit how fast Chinese AI compute (and thus DC demand) scales
- Balance-sheet capacity / cost of capital to fund RMB10-12B annual capex
- Construction and delivery cadence — the ~500MW pipeline must land on schedule (2026-2028) to convert orders to EBITDA
- Utilization ramp lag — capacity depreciates before it fills, compressing near-term GAAP margins
Top signals & trends
Top signals
Strategic anchor + power/energy angle; note it is a secondary transfer (no direct capital to VNET) — watch for close, board seats, and any separate primary injection
Key to converting capex into EBITDA; keep watching quarterly
Macro demand tailwind, but state-carrier-led and domestic-chip-mandated — not uniquely VNET's
Free-cash-flow burn and funding dependence
Can cap AI DC demand and add structural ADR risk
Trends
High positive · Structural multi-year demand for in-country compliant hyperscale capacity
Positive (growth) / mixed (margin) · Drives revenue growth but compresses GAAP gross margin during ramp
Raises capex/complexity but favors operators with power-secured sites and energy partners
Negative / structural · Chip supply and ADR-listing risk; China's domestic-chip mandate is both a hedge and a constraint
Positive optionality · VNET listed two REIT projects in Q1 2026 — a lever to fund growth without pure leverage
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.
AI accelerators that drive tenant compute demand; China-compliant supply is export-control-gated
Data-center power, cooling and thermal-management systems for high-density AI racks
Incoming ~38.1% strategic shareholder (via affiliate) plus potential energy-storage/power supplier for AI-era data centers
Server and IT hardware OEMs supplying the underlying compute infrastructure
Hyperscale cloud tenant and demand driver for wholesale capacity
Cloud/internet hyperscale tenant
AI/cloud tenant
Private; among the largest China AI/internet capacity buyers — the type of 'leading internet customer' likely behind the 510MW Beijing order (not company-confirmed)
Azure/M365-in-China run on VNET — both partner and demand source
Closest China carrier-neutral hyperscale peer; larger, higher-multiple, also a China AI-DC re-rating beneficiary with international (DayOne) expansion
State-owned carriers with vast in-house IDC capacity — the dominant competing supply, gatekeepers of interconnection, and designated operators of the national AI-compute grid
Bain-backed hyperscale operator (taken private from Nasdaq in 2023); direct wholesale competitor for hyperscale tenants
Global carrier-neutral leader; competes at the edges/APAC and sets the interconnection standard, limited direct mainland-China overlap
Global wholesale/hyperscale leader; benchmark for the build-and-lease model, minimal direct mainland overlap