GDS Holdings Limited
China's largest independent data-centre landlord, renting power and cooled space to cloud and AI companies, and owner of about a fifth of DayOne, a private operator abroad.
The scores
Growth3/4 ↑
Recurring EBITDA grows about 7% in 2026 with a 757MW backlog doubling move-in in 2027; Growing, not yet Compounding.
Business quality3/4 →
Churn under 1% a quarter and long contracts, but legacy prices fall about 3% a year and customers can build their own.
Financial position2/4 ↓
Net debt 4.7 times EBITDA rising toward 6; RMB10B of 2026 capex against RMB1.9B of first-half operating cash; a $620M put in March 2027.
Execution record3/4 ↑
Raised guidance and bookings targets in 2026 and spun out DayOne well, but impaired RMB1.6B of sites in 2025.
Per share capture2/4 →
Shares sold at $24.50 in 2025, convertibles at $24.50 and $33.08 and 40% equity funding dilute each ADS.
Potential2/4 ↑
MIIT's compute target is huge, but bookings convert only as chips arrive and DayOne's value is unproven in public.
Risk profile3/4 →
Two customers are 41% of revenue, chips are export-controlled and the VIE can be challenged; survivable but Concentrated.
Valuation3/4 ↑
About 9.6 times recurring EBITDA for the China business after DayOne at $2.2B, near the 9.1 times new capacity costs.
Evidence quality3/4 →
Figures are filed, but bookings, EBITDA per MW and DayOne's value are management's or press reports.
If it works 1.1× · if it fails you keep about 51%.
The Verdict
We rate GDS neutral. It rents data-centre halls to China's cloud giants, and the business is mixed: customers stay, AI orders hit a record, but old contracts reprice down and borrowing pays for the building. The price is fair: the China business costs about what new halls would cost to build, and the DayOne stake sits near the value GDS puts on it. As a bet it is fragile: even a strong DayOne listing adds little to today's price, and if the listing fails about half is left. The orders are real; what is missing is a discount.
The long caseJohor's data-centre boom has to last long enough for DayOne to fill what it has booked and list at a price above GDS's own mark.
The evidence under this - 8 claims
- Bookings are not a filed metric; the 20-F counts area committed, which includes non-binding letters of intent.
- The ramp period varies by contract, and management assumes four quarters on a straight line.
Move-in depends on customers obtaining chips; an analyst asked what a GPU delay would do and the answer relied on domestic supply catching up.
The 2025 impairment was 'mainly due to lower sales price', so the repricing has already cost more than a few points of revenue on some sites.
Net debt rises from here: capex is guided at RMB10B for 2026 with most of it in the second half, against first-half operating cash flow of RMB1.9B.
- Including finance leases in debt but not lease costs in EBITDA flatters the multiple; without leases it is about 8.3 times.
- VNET, the one filed rival, trades near 7 times its 2026 EBITDA guide.
At VNET's multiple the China equity would be worth about $1.15B, so the core's value is sensitive to the multiple chosen.
A private round's price is what new money paid for preferred-ranking shares, not what the market will pay for ordinary ones.
A listing priced below the reported range, or a lock-up, would leave less; DayOne's preferred shares rank ahead of ordinary ones.
The company
A cloud company's servers need a building with steady power and cooling. GDS finds land, wins the power approvals, builds the hall to the customer's design within about a year, and then charges a monthly fee for the space and the power used.
Its customers are China's biggest cloud and internet companies. They could build for themselves, and some do, but GDS sells speed and sites they cannot easily get: a land bank of nearly 4GW in the cities and western hubs they want.
The evidence under this - 5 claims
- Bookings are not a filed metric; the 20-F counts area committed, which includes non-binding letters of intent.
- The ramp period varies by contract, and management assumes four quarters on a straight line.
Power quota needs municipal, provincial and central approval, so the land bank is not all buildable on management's timetable.
GDS does not name its largest customer; tying the 37.9% to Alibaba is the market's reading, not a filing's.
How it makes money
Rent arrives every month once a customer has moved its servers in, and most of it is left over after running costs: about 45 cents of each yuan of revenue before interest and depreciation. A new hall costs about RMB20M per megawatt to build and management expects about RMB2.2M a year of that profit from each megawatt, so each new booking pays for itself in about nine years.
The evidence under this - 4 claims
The vendor reports 2025 operating income of RMB1,536M because it leaves out the impairment; both are correct under their own definition.
RMB2.2M per MW is management's estimate from contract prices and its own cost benchmarks; no filing carries it.
The yield is a target on assets not yet built; the 2025 impairment shows older sites earning less than planned.
Where the money comes from, stream by stream
| Stream | Share of sales (FY2025) | How it earns it |
|---|---|---|
| Colocation services | 90.7% | Monthly fees for space, power and cooling in its data centres, on contracts of three to ten years, plus power used. |
| Managed services and others | 9.2% | Hosting, cloud and operating services on top of the space, shrinking since 2023. |
| Equipment sales | 0.04% | Occasional hardware sold to customers. |
Who runs it
William Huang founded GDS and has run it since 2002; Daniel Newman has run its finances since 2011. Insiders own only about 3% of the company, but Huang's special shares give him well over half the votes, which some Chinese customers now require. Pay is mostly shares that vest with time and with share-price or other goals; the filing does not say which goals they must hit.
The evidence under this - 3 claims
Class B votes were raised from twenty to fifty per share on March 10, 2026 because some customers will only contract with providers controlled by Chinese nationals, so the control is a commercial requirement as well as a founder's protection.
The grants are valued with a Monte Carlo model, which implies share-price hurdles, but the filing names no hurdle, metric or payout, so what the pay rewards cannot be read from outside.
Why it wins, and who could beat it
Why customers come back
Moving thousands of live servers to another building is slow and risky, so customers rarely leave: under 1% of the rented space ended each quarter in 2025. Even a rival, VNET, says relocating servers is operationally difficult.
But staying is not the same as paying more. As old contracts renew, the rent per square metre is falling about 3% a year, and in 2025 GDS wrote down RMB1.6B of sites because prices came in lower than planned.
The evidence under this - 4 claims
The 2025 impairment was 'mainly due to lower sales price', so the repricing has already cost more than a few points of revenue on some sites.
The vendor reports 2025 operating income of RMB1,536M because it leaves out the impairment; both are correct under their own definition.
The threat
The biggest rival is the customer itself. Alibaba has committed at least RMB380B over three years to its own cloud and AI buildings, and two customers account for about half of the space GDS has rented out. The three state telecom companies also run more than half of the market's data centres, and VNET competes for the same buyers.
The evidence under this - 4 claims
GDS does not name its largest customer; tying the 37.9% to Alibaba is the market's reading, not a filing's.
A consultant's forecast commissioned for a listing document; it publishes no capacity in MW.
VNET's debt here excludes leases; its retail business carries different margins.
The rivals: VNET, BABA, and 4 more that are private, in-house or listed abroad.
Who else does this, rival by rival
| Ticker | Rival | Where it competes | How far along | Priced against it |
|---|---|---|---|---|
| China Telecom, China Unicom and China Mobile data centresPriced against it: Listed abroad | China data centres | On sale | Listed abroad | |
| VNET | VNET's wholesale data centresPriced against it: Same product | China data centres | On sale | Same product |
| Sinnet and Range Intelligent ComputingPriced against it: Listed abroad | China data centres | On sale | Listed abroad | |
| Private | ChindataPriced against it: Private | China data centres | On sale | Private |
| Private | Private data-centre platforms building in Singapore, Malaysia and IndonesiaPriced against it: Private | DayOne stake | On sale | Private |
| BABA | Alibaba's own data centresPriced against it: Different product | China data centres | On sale | Different product |
How big it could get
How big this could get
Frost & Sullivan expects China's data-centre services market to reach RMB339.0B by 2030, growing 13.2% a year, with independent landlords like GDS growing faster than the telecom companies. The government's own target is bolder: it wants AI computing power to grow about sixfold by 2030. Nobody publishes how many megawatts of buildings that needs, so GDS's share cannot be measured.
The evidence under this - 2 claims
A consultant's forecast commissioned for a listing document; it publishes no capacity in MW.
Who has signed
In the first half of 2026 customers signed for 470MW on take-or-pay terms, meaning they pay whether or not they fill the space, and reserved another 600MW beside it. GDS does not name them, beyond saying each of the three largest cloud companies signed.
The evidence under this - 2 claims
- Bookings are not a filed metric; the 20-F counts area committed, which includes non-binding letters of intent.
- The ramp period varies by contract, and management assumes four quarters on a straight line.
Q1 bookings of about 200MW were also a record, and the shares fell 13% in two days because revenue from them starts only in late 2027.
The results
The numbers
Sales grew 11% in the year to Dec 2025, faster than the 5.5% in the year before.
Operating margin is −0.5%, down from 7.7% in the year to Dec 2022.
Running costs take 23¢ of each sales dollar, up from 15¢ in the year to Dec 2021.
Shareholders' money earned 3.9% a year, up from −4.7% in the year to Dec 2021.
It used CNY 1.33B of cash in the latest year, so the CNY 14.3B it holds lasts about 10.8 years at that pace.
It owes CNY 12.0B more than the cash it holds, about 3.6 years of the cash its operations bring in.
The share count rose 13% in 4.0 years, and pay in shares costs 2.5% of sales.
All the numbers
Filed figures from the company's 10-K and 10-Q, read 2026-10-04. Fiscal years end in December. In CNY.
| Dec 2021 | Dec 2022 | Dec 2023 | Dec 2024 | Dec 2025 | |
|---|---|---|---|---|---|
| Size and growth | |||||
| Sales | CNY 7.82B | CNY 9.27B | CNY 9.78B | CNY 10.3B | CNY 11.4B |
| Sales growth | +36% | +19% | +5.5% | +5.5% | +11% |
| Profit | |||||
| Gross margin | 23% | 21% | 20% | 22% | 23% |
| Operating margin | 7.3% | 7.7% | −23% | 11% | −0.5% |
| Net margin | −15% | −14% | −44% | 33% | 8.3% |
| What growth costs, per sales dollar | |||||
| Research and development | 0.5% | 0.3% | 0.4% | 0.4% | 0.3% |
| Sales and marketing | 1.9% | 1.6% | 1.4% | 1.1% | 1.3% |
| General and admin | 13% | 11% | 9.9% | 8.9% | 7.9% |
| Pay in shares | 5.0% | 3.1% | 3.4% | 2.9% | 2.5% |
| Capital spending | 124% | 63% | 33% | 31% | 41% |
| Cash | |||||
| Cash from operations | CNY 1.20B | CNY 2.81B | CNY 2.07B | CNY 1.94B | CNY 3.37B |
| Free cash flow | −CNY 8.50B | −CNY 3.06B | −CNY 1.13B | −CNY 1.23B | −CNY 1.33B |
| Balance sheet | |||||
| Cash and short-term investments | - | - | - | CNY 7.87B | CNY 14.3B |
| Debt | CNY 24.2B | CNY 27.1B | CNY 25.7B | CNY 26.2B | CNY 26.3B |
| Net cash | - | - | - | −CNY 18.4B | −CNY 12.0B |
| Returns and shares | |||||
| Return on capital | 1.0% | 1.0% | - | - | - |
| Shares (diluted) | 1.45B | 1.46B | 1.47B | 1.48B | 1.64B |
What growth costs
Growing does not make GDS much more profitable. Office and sales costs shrink as a share of revenue, but they were small to begin with; power, rent and the wear on each new building rise with it. What growth really costs is the building itself, paid for up front with borrowing and new shares.
The evidence under this - 2 claims
- The cost that grows with sales is cost of revenue, power, depreciation and rent, about 77 to 80 fen per yuan in every year, so the overhead savings barely move the margin.
- Capital spending is the real cost of growth: the cash-flow line for property and equipment was RMB4,691.1M in 2025 against management's capex of about RMB9B (C17), which counts more than that one line.
The yield is a target on assets not yet built; the 2025 impairment shows older sites earning less than planned.
The last report (13 Aug)
The second quarter of 2026 brought steady numbers and a big change in orders. Revenue rose 6.5% and profit before interest and depreciation 2.5%, as older contracts renewed at lower prices.
Orders were the story: 260MW booked in the quarter and 470MW in the half, a record, so management doubled its target for the year to 1GW and raised its revenue and profit guidance. The catch is timing. Customers move in over about a year after delivery, so most of this lands in the second half of 2027.
What it did to the thesisIt proved the demand turn is real and binding, and left open the part the view hangs on: whether the new space fills on time in 2027 while older contracts keep repricing down.
The quarter in detail
Earnings per share $0.52 against $0.01 expected.
The evidence under this - 6 claims
- Bookings are not a filed metric; the 20-F counts area committed, which includes non-binding letters of intent.
- The ramp period varies by contract, and management assumes four quarters on a straight line.
Q1 bookings of about 200MW were also a record, and the shares fell 13% in two days because revenue from them starts only in late 2027.
Move-in depends on customers obtaining chips; an analyst asked what a GPU delay would do and the answer relied on domestic supply catching up.
The guide includes one-time items worth about RMB429M of revenue and about RMB518M of EBITDA.
The 2025 impairment was 'mainly due to lower sales price', so the repricing has already cost more than a few points of revenue on some sites.
What management did with the money
Insiders in 2026Q3: 16 purchases and 0 sales, from their Form 4 filings.
The price, and where we disagree
In short. Fair. Take out the DayOne stake at the $2.2B GDS puts on it, and the rest is priced at about what new data centres earning the same profit would cost to build.
The shares are near their low for the year, down by about a third while revenue per share rose. They fell 11.7% in the two sessions to 2 October with no news we could find.
Two tests are close. DayOne plans to make its listing papers public in mid-October and list in November, which will put a public price on the stake. GDS reports its third quarter on 18 November, the first look at whether orders are on track for 1GW.
The evidence under this - 4 claims
- The people cautioned that plans and timing may change; SB Energy delayed its own listing the same week, and Switch, Vantage Data Centers and CyrusOne are preparing listings that compete for the same investors.
- DayOne's Form F-1 was already public on 5 October 2026 (C53), earlier than the mid-October the report gave; the price range and the listing date are still blank in it.
- 2026-10-19editedcounter-evidence added: the public F-1 arrived 5 October, ahead of the reported mid-October; claim text is the report as made
Move-in depends on customers obtaining chips; an analyst asked what a GPU delay would do and the answer relied on domestic supply catching up.
How we read the multiple
GDS's market value is small next to its debt, so the price has to be read as the whole business, debts included:
- Add RMB31.1B of debt and other claims, net of cash, to the RMB36.5B market value.
- Take out DayOne at $2.2B, and about RMB52.6B is left for the China business.
- That is about 9.6 times this year's expected profit before interest and depreciation, without one-time items.
- Building the same profit new costs about 9.1 times, so the price is close to replacement cost.
Against its own history. Cheaper than its own past, because the market no longer pays extra for growth that arrives late; the business has not become cheaper to run.
The evidence under this - 6 claims
- Including finance leases in debt but not lease costs in EBITDA flatters the multiple; without leases it is about 8.3 times.
- VNET, the one filed rival, trades near 7 times its 2026 EBITDA guide.
At VNET's multiple the China equity would be worth about $1.15B, so the core's value is sensitive to the multiple chosen.
The time-deposit figure is derived from first-half purchases less maturities, not read from a balance-sheet line.
The yield is a target on assets not yet built; the 2025 impairment shows older sites earning less than planned.
Management's own pro forma puts 2026 adjusted EBITDA growth at 6.5% and first-half growth at 12.7%, on a slightly different bridge.
A private round's price is what new money paid for preferred-ranking shares, not what the market will pay for ordinary ones.
Priced against its rivals
Only VNET Group sells the same thing to the same buyers as a large part of what it does. A price comparison needs at least two, so the price is judged against its own history instead.
Private or foreign rivals that cannot be priced: China Telecom, China Unicom, China Mobile, Beijing Sinnet Technology (300383.SZ), Range Intelligent Computing (300442.SZ), Chindata Group, DayOne Data Centers.
The bet: what it pays if it works
This bet is on one thing: the DayOne stake. The China business is counted separately, at what it would cost to build again.
If DayOne lists at the value reported in the press, GDS's stake is worth a little more than the value GDS already puts on it, so today's price rises only slightly. If the listing fails, the China business still covers about half of today's price. That is an expensive bet: the good outcome is already mostly paid for.
| DayOne stake book value in two to three years | Billion dollars of DayOne's value needed | At the stated multiple |
|---|---|---|
| $2.20B | about 11 at 20% kept | 0.9× |
| $2.98B (the big win) | about 15 at 20% kept | 1.1× |
| $3.98B | about 20 at 20% kept | 1.3× |
To pay 2× (a Fair bet) it would need about $7.98B of DayOne stake book value (about 40 billion dollars of DayOne's value at 20% kept); to pay 5× (a Cheap bet), about $24.11B of DayOne stake book value (about 121 billion dollars of DayOne's value at 20% kept), at the same multiple.
Each row is DayOne stake alone, plus the rest of the company (China data centres) counted at about $2.76B.
The steps it has to take
- Done DayOne's Series C closes, upsized to $4.5B, at about $10.9B
- Not started DayOne makes its SEC registration public, showing revenue and debt due 2026-10-31
- Not started DayOne lists in the US at or above its Series C valuation due 2026-12-31
- Not started Billable capacity passes 1GW, half of DayOne's 2.1GW of bookings due 2027-12-31
The detail behind the label
The evidence under this - 6 claims
A private round's price is what new money paid for preferred-ranking shares, not what the market will pay for ordinary ones.
Anonymous sources; DayOne declined to comment, and two other data-centre listings were delayed in September 2026.
- The people cautioned that plans and timing may change; SB Energy delayed its own listing the same week, and Switch, Vantage Data Centers and CyrusOne are preparing listings that compete for the same investors.
- DayOne's Form F-1 was already public on 5 October 2026 (C53), earlier than the mid-October the report gave; the price range and the listing date are still blank in it.
- 2026-10-19editedcounter-evidence added: the public F-1 arrived 5 October, ahead of the reported mid-October; claim text is the report as made
A listing priced below the reported range, or a lock-up, would leave less; DayOne's preferred shares rank ahead of ordinary ones.
At VNET's multiple the China equity would be worth about $1.15B, so the core's value is sensitive to the multiple chosen.
Management's own pro forma puts 2026 adjusted EBITDA growth at 6.5% and first-half growth at 12.7%, on a slightly different bridge.
Where we disagree with the market
Most analysts see a cheap AI play with a valuable DayOne stake the price ignores.
The China business is worth about what it costs to build and DayOne about its last price, which together make today's price.
The cheap look comes from counting DayOne on top of a business whose debts and preferred shares come before the common shares. Once those are counted, little is left over.
We would be wrong if 2027 move-ins more than double as promised and each megawatt earns what management says, or if DayOne lists well above its reported target.
What analysts say, and the evidence
The evidence under this - 6 claims
A private round's price is what new money paid for preferred-ranking shares, not what the market will pay for ordinary ones.
- Including finance leases in debt but not lease costs in EBITDA flatters the multiple; without leases it is about 8.3 times.
- VNET, the one filed rival, trades near 7 times its 2026 EBITDA guide.
At VNET's multiple the China equity would be worth about $1.15B, so the core's value is sensitive to the multiple chosen.
The time-deposit figure is derived from first-half purchases less maturities, not read from a balance-sheet line.
Move-in depends on customers obtaining chips; an analyst asked what a GPU delay would do and the answer relied on domestic supply catching up.
What could go wrong, and how we would know
The risks
- Two customers The two largest customers hold about half the space rented out; either building more for itself would leave halls empty.
- Borrowed growth This year's building costs about RMB10B against under RMB2B of cash earned in the first half, so debt keeps rising.
- Chips may not arrive US export rules limit China's access to AI chips; customers without chips move in late, and rent starts late.
- Prices keep falling Old contracts renew lower, and the 2025 write-down shows some sites earning less than planned.
- DayOne prices low A listing below the last private price, or a delay, would cut the stake GDS values at about two-fifths of the market value.
- Structure and politics GDS runs through contracts Chinese regulators could challenge, and US-China tension touches both its chips and its investors.
The evidence under this - 7 claims
Net debt rises from here: capex is guided at RMB10B for 2026 with most of it in the second half, against first-half operating cash flow of RMB1.9B.
The 20-F's risk factors describe US export controls and a pending Remote Access Security Act limiting China's access to advanced chips.
The 2025 impairment was 'mainly due to lower sales price', so the repricing has already cost more than a few points of revenue on some sites.
A private round's price is what new money paid for preferred-ranking shares, not what the market will pay for ordinary ones.
What we still do not know
- What DayOne earns and owes: GDS's filings carry only its balance sheet and capacity, and its own listing papers are not public yet.
- Who the two largest customers are, and how much of this year's orders they signed.
- What a megawatt rents for on new contracts compared with old ones.
The evidence under this - 3 claims
- Reuters reports DayOne now has about 2.1GW of bookings, so it has grown since the 20-F's figures.
- DayOne's own Form F-1 (C53) now gives its revenue ($484.3M in 2025) and billed capacity (666MW at 30 June 2026), so the figures this claim says GDS's filings lack are filed by DayOne itself.
- 2026-10-19editedcounter-evidence added from DayOne's public F-1 of 5 October; the claim about GDS's own filings stands
RMB2.2M per MW is management's estimate from contract prices and its own cost benchmarks; no filing carries it.
What would change our view
- Neutral→AttractiveThe bookings arrive and DayOne lists well
It needs all of:
- ○Customers move into at least 470MW of new space in 2027, double this year.checked at the next results, due by 18 Nov
- ○DayOne lists at or above its last private valuation of about $10.9B.watched as it happens
- Neutral→UnattractiveThe build outruns the bookings
It needs any of:
- ○Orders in any quarter fall below 120MW, under half of the second quarter's.checked at the next results, due by 18 Nov
- ○Debt, net of cash, rises above 6 times a year's profit before interest and depreciation, from 4.7 times.checked at the next results, due by 18 Nov
We would also be wrong if
- DayOne's listing slips past 2027.
- Rent per square metre stops falling by the end of 2027, which would help.
What is coming
- October 2026 DayOne makes its US listing filing public: DayOne's own revenue, losses and debt for the first time, testing whether the stake is worth the private mark already in the price.
Every dated event ahead
- October 2026 DayOne makes its US listing filing public: DayOne's own revenue, losses and debt for the first time, testing whether the stake is worth the private mark already in the price.
- November 2026 DayOne US listing: A public price for the stake: near $20B would lift it above the Series C mark, a delay or cut would leave the price leaning on a private valuation.
- 18 November 2026 Q3 2026 results: Whether nine-month bookings reach 700MW toward the 1GW target and recurring adjusted EBITDA holds about RMB1,400M a quarter.
- March 2027 Holder put on the $620M convertible notes: Whether the notes are refinanced with debt, cash or the DayOne proceeds rather than new shares at today's price.
- March 2027 Q4 and full-year 2026 results: Whether 2026 guidance was met without one-time items, how far legacy prices fell, and the 2027 move-in and capex plan.
Since we wrote this
- 19 OctThesis reviewDayOne's public F-1 gives the stake real numbersThe listing filing shows DayOne's revenue, loss and 666MW billed, and GDS at 19.4%. It names no price, so the view held.
- 5 OctThesis reviewThe bet now names what it ridesThe owner asked that every bet name the macro trend behind it, after BWXT's bet was missing its chief executive's case for nuclear over gas.
- 4 OctThesis writtenNeutral, medium convictionThe China business is priced near what it costs to build and DayOne near its private mark, so the stake is not a free option.
- 4 OctThesis reviewOwner's checklist: revenue streams, cost of growth, management, catalystsThe owner's checklist found the method silent on who runs the company, what growth costs, where the money comes from and what is dated ahead; this pass fills those four blocks from the 20-F.
Also checked 1 time with nothing new to act on.
Sources 20
Sources - 20
Every claim above cites one of these, or states the reasoning that derived it. The quote is frozen at the date it was read: a link can rot, and an agent fetch returns a reading of a page rather than the page. Backs counts the whole claim ledger, so it can name a claim this page does not surface.
- S01primarycompany filing2026-04-29Item 4 business description, the Competition section naming carrier-neutral classes and the three state carriers; 989 customers; the two largest at 37.9% and 11.8% of area committed and 29.0% and 12.0% of revenue; churn averaging 0.9% a quarter in 2025; contract terms of three to ten years; the DayOne stake's history (35.6%, 30.1%, about 19.9%), its US$2.2B Series C value, the share of DayOne's 2025 loss; the 2025 impairment of RMB1,561.2M; convertible notes and preferred shares; risk factors on pricing, customer concentration, export controls and the VIE structure; Note 21 revenue (colocation RMB10,373.4M, managed service and others RMB1,054.7M of RMB11,432.3M in 2025); Item 6 directors and officers, their compensation (US$19.4M, US$6.5M in cash), the 2016 share incentive plan and the March 31, 2026 beneficial ownership table; Item 16E (no repurchases) and the dividend policybacksC01C04C10C11C16C20C21C22C31C42C44C45C46C47C48C49C50
- S02primarycompany filing2026-08-13Q2 2026 revenue RMB3,088.0M and adjusted EBITDA RMB1,406.0M; area committed 784,802 sqm, utilization 79.2%, 170,355 sqm under construction 89.2% pre-committed; the June 30, 2026 balance sheet; first-half cash flows; raised 2026 guidance; the DayOne dilution gain of RMB959.9MbacksC02C09C14C17C19C32C40
- S03primarycompany filing2026-05-20Q1 2026 revenue RMB3,367.1M including about RMB429.1M of one-time revenue; adjusted EBITDA RMB1,948.7M and RMB1,430.3M without one-time items; the DayOne dilution and sale gain of RMB2,136.5MbacksC03C35
- S04primarycompany filing2026-03-172025 revenue RMB11,432.3M up 10.8% and adjusted EBITDA RMB5,403.5M; Q4 net loss after impairment; RMB13.9B of debt financing in 2025; the C-REIT and ABS monetisations; the US$385M DayOne share sale and the Huatai preferred placementbacksC04C16C39
- S05primarytranscript2026-08-13Second quarter 2026 earnings call transcript (FMP)data/cache/transcripts/GDS-2026Q2.json470MW of first-half bookings, the 1GW target, take-or-pay terms, 600MW of reservations, backlog of 757MW at RMB2.2M of adjusted EBITDA per MW, move-in of 235MW in 2026 and more than double in 2027, unit capex of RMB20M per MW, 60/40 funding, net debt at 4.7 times, monthly service revenue down about 3% a year, the GPU/CPU mix. Analysts asked about GPU supply and take-or-pay protection; none asked about DayOne's value
All of our sales agreements, including -- include a binding take-or-pay commitment.
backsC05C06C07C08C12C13C15C17C39C40C41 - S06primarytranscript2026-05-20First quarter 2026 earnings call transcript (FMP)data/cache/transcripts/GDS-2026Q1.jsonrecord bookings of about 200MW; a three-year plan of 500MW to 800MW of bookings a year and RMB30B to RMB50B of investment; a land bank of nearly 4GW; how power quota is approved
Anticipating the demand trend, we increased our secured landbank to nearly 4 gigawatts.
backsC43 - S07primarycompany filing2026-04-16a rival naming GDS as a carrier-neutral competitor; 889MW of wholesale capacity in service; relocation of customer servers called operationally difficult; Frost & Sullivan's market figures: carriers 55.7% of the 2025 market, RMB339.0B by 2030 at 13.2% a year, carrier-neutral growth of 18.5% a yearbacksC26C27
- S08primarycompany filing2026-08-18VNET's 2026 adjusted EBITDA guidance of RMB3,550M to RMB3,750M, up 19% to 26%; its June 30, 2026 debt and cashbacksC38
- S09primaryregulator2026-09-07MIIT, 15th Five-Year Plan for the information and communications industry (信息通信行业发展“十五五”规划) Ministry of Industry and Information Technology (MIIT)the plan's indicator table: intelligent computing capacity of 1,590 EFLOPS in 2025 and a 9,800 target for 2030; advanced storage from 540EB to 1,700EB; a PUE target below 1.2 for new large data centres
9 智能算力规模（每秒百亿亿次浮点运算） 1590 9800 〔8210〕 10 先进存力规模（EB） 540 1700 〔1160〕
backsC28 - S10secondarynews2026-07-20MIIT's mid-2026 figures: 2,185 EFLOPS of intelligent computing capacity at the end of June 2026 and a 71.4% utilization rate of computing facilities nationwide
China's intelligent computing capacity reached 2,185 EFLOPS by the end of June this year
backsC29 - S11primarycompany filing2025-02-24Alibaba Group Form 6-K Exhibit 99.1: Alibaba to Invest RMB380 billion in AI and Cloud Infrastructure BABAthe largest cloud buyer in China committing to build its own computing infrastructure: at least RMB380B over three years
announced plans to invest at least RMB380 billion (US$53 billion) over the next three years to advance its cloud computing and AI infrastructure
backsC30 - S12secondaryvendor data2026-10-04GDS research pack (FMP, FINRA, SEC Form 4), frozencontent/GDS/research-pack.jsonprice and return series, the price-move split, the 52-week range, analyst rating history and targets, short interest, insider statistics, headcount, VNET's multiplesbacksC32C33C35C38
- S13primarycompany filing2026-06-12two open-market sales in June 2026, both with the Rule 10b5-1 box unchecked; no open-market purchasesbacksC34
- S14tertiarynews2026-01-23people familiar putting DayOne's target IPO valuation as high as $20B, a step up from its funding rounds
DayOne Data Centers Ltd. is targeting a valuation as high as $20 billion in its planned US initial public offering
backsC23 - S15tertiarynews2026-09-28Reuters via The Star: Data centre IPO hopefuls brave tougher market as investor scrutiny grows ReutersDayOne's plan to make its SEC filing public in mid-October and list in November; a raise of up to $5B at about $20B; the $4.5B Series C closed in June; about 2.1GW of bookings
It plans to make its filing with the US Securities and Exchange Commission public in mid-October and list in November
backsC24 - S16tertiarynews2026-02-23the Series C expected to value DayOne at about $10B; banks named for the IPO
DayOne closed a US$2 billion Series C funding round which was expected to value the company at about US$10 billion
backsC23 - S17primarycompany filing2023-04-04the 2021 and 2022 income statements and cash flow statements as filed, read through the frozen numbers card (content/GDS/financials.json)backsC46C50
- S18primarycompany release2026-08-28Tenaga Nasional Berhad, Analyst Briefing 2QFY2026 (28 August 2026), slides on electricity demand and data centres Tenaga Nasional BerhadMalaysian data-centre load by quarter (603MW June 2025 to 1,256MW June 2026), data-centre energy use (1.9TWh in 1H2025, 4.0TWh in 1H2026), 8.35GW of secured maximum demand across 61 projects, 5.6GW of it in Johor, and 5 agreements of about 0.9GW signed in the half; downloaded and read with pdftotext because ir_fetch returned the raw PDF
Data centre growth underpins future electricity demand, with Johor positioned as Malaysia’s regional hub
backsC51C52 - S19primarycompany release2026-05-21Tenaga Nasional Berhad, Analyst Briefing 1QFY2026 (21 May 2026), slides on data centres Tenaga Nasional Berhadabout 8.3GW of secured data-centre maximum demand at March 2026, 1,054MW of load in March 2026 against 485MW a year earlier, and 3 agreements of about 0.8GW signed in the quarter
YTD Progress: 3 ESAs signed with a total capacity of ~0.8GW
backsC51C52 - S20primarycompany filing2026-10-05DayOne Data Centers Limited, Form F-1 registration statement filed 5 October 2026 DayOne Data Centers LimitedDayOne's public Form F-1: revenue of US$484.3M in 2025 and US$512.0M in the first half of 2026; net loss of US$367.1M in 2025 and US$77.2M in the first half of 2026; Billings of 444MW at the end of 2025 and 666MW at 30 June 2026; the top two customers at 84.3% of first-half revenue; GDS holding 64,000,000 shares, 19.4%; ADS price range and offering size left blank; Nasdaq symbol DODC applied for
Our revenue increased by 238.0% to US$512.0 million in the six months ended June 30, 2026, from US$151.5 million in the same period in 2025.
backsC53