What would an investment committee make of a data-centre operator deriving 69.2% of its revenue from one customer?
Edition 26, The Credit Behind the Compute, closed by asking how a customer-concentrated operator turns contracted growth into resilient cash flow for investors. DayOne Data Centers provides a revealing case: its capacity under construction was 99.3% pre-committed as of 20 September 2026, yet one unnamed customer accounted for 69.2% of its unaudited H1 (first half) 2026 revenue.
These facts are not contradictory. A large anchor customer can make a campus easier to finance while leaving the operator dependent on that same relationship. The underwriting question is what survives between the contract and the eventual return to equity.
1 The deal
DayOne filed Form F-1 (a registration statement for certain foreign private issuers) with the SEC (Securities and Exchange Commission) on 5 October 2026 for a proposed United States IPO (initial public offering). The filing reports that its unnamed largest customer generated 69.2% of total recognised revenue in unaudited H1 2026; the second-largest contributed 15.1% on the same basis. Together, they represented 84.3%, expressly disclosed in the filing. For FY2025 (financial year 2025), the comparable top-two figure was 81.6%; the individually rounded shares of 69.4% and 12.3% sum to 81.7%. [1]
The company is building a large hyperscale wholesale platform on the strength of long-term demand commitments. As of 20 September, it had 2,281 MW (megawatts) of Bookings — legally committed IT (information technology) power capacity — primarily from seven global hyperscale and leading technology customers. In its 14 October announcement, DayOne reported that Bookings came from 15 customers as of 20 September. That customer-count observation is not a new 14 October KPI (key performance indicator) snapshot. Seven major customers and 15 booking customers describe different populations; neither statement tells us how booked capacity is divided among those customers. [2]
The later issuer announcement reported that Billings, or committed capacity already generating income, had exceeded 1 GW (gigawatt). [2] The financing architecture behind that expansion deserves separate scrutiny alongside the operating milestones.
Concentration can be a feature of this business model. A contracted anchor tenant can improve a purpose-built campus's ability to attract financing before the first hall is energised, subject to the lender's assessment of obligations and delivery risk. But contracted megawatts, recognised revenue and cash available to shareholders are three different measurements.
2 The engineering read
From booked MW to billed MW
The Form F-1 defines these operating metrics as IT power capacity, not dollars. All figures below are as of 20 September 2026. [1]
| DayOne metric | Reported | What it actually measures |
|---|---|---|
| Bookings | 2,281 MW | Legally committed capacity; Billings + Backlog |
| Billings | 953 MW | Committed capacity already income-generating |
| Backlog | 1,328 MW | Committed capacity not yet income-generating |
| Capacity in Service | 962 MW | Ready-for-service, equipped capacity |
| Capacity Under Construction | 1,328 MW | Capacity physically under active construction, not yet ready for service |
| Utilisation | 99.1% | Billings divided by Capacity in Service |
| Pre-commitment | 99.3% | Backlog under construction, a subset of total Backlog, divided by Capacity Under Construction |
Important distinction
Backlog and Capacity Under Construction each round to 1,328 MW but are different measures. Dividing the two displayed totals gives 100%, not the issuer-reported 99.3% pre-commitment rate; the precise numerator is the under-construction subset, not the whole backlog.
Five relevant dates should not be blended: 30 June (unaudited H1 financials), 20 September (operating KPI snapshot), 5 October (F-1 filing), 13 October (seven-lender financing announcement), and 14 October (later issuer operating update). Billings above 1 GW in the 14 October announcement do not retroactively alter the September observation of 953 MW.
Power delivery is a cash-flow gate
For a developer, a power-secured site is not automatically a revenue-producing site. Grid readiness, electrical and cooling systems, commissioning, customer acceptance and contractual availability commitments all influence when booked MW begin billing. Delay can defer cash receipts while development financing costs continue. Who ultimately bears each delay cost depends on the individual agreements, not a generic project-finance assumption.
DayOne's F-1 warns that delivery delays or missed service commitments may lead to financial penalties, service credits, fee reductions and, in some cases, termination. Extended force-majeure events can permit termination without penalty, while material defaults and certain SLA (service-level agreement) breaches create separate termination risks, subject to contractual conditions and cure periods. In some contracts, customers may also terminate without financial penalty if committed power availability falls below a specified minimum for a continuous period. That is a qualified contractual right, not an automatic exit for every power shortfall. The engineering promise and the legal payment promise must be assessed together.
3 The capital allocation read
The easy response to 69.2% concentration is to apply a valuation discount automatically. The equally easy response to long-term, take-or-pay commitments is to assume that counterparty risk has been removed. Neither follows from the filings alone.
Anchor customers can reduce speculative lease-up risk and help lenders underwrite project deployment. DayOne says a significant majority of its agreements span 10–15 years, and substantially all September Bookings were covered by long-term take-or-pay arrangements. That is constructive demand visibility — conditional on the obligor, the terms and the operator's ability to perform. Importantly, DayOne also discloses that all customer contracts permit early termination subject to specified costs and penalties, which, after a certain number of years, would usually be less than the revenue DayOne otherwise expected over the remaining contractual term. Long tenor is therefore not a lifetime payment guarantee.
Six underwriting gates between contract and equity return
| Gate | The investment committee's question |
|---|---|
| 1 · Obligor | Which legal entity owes payment, and what parent or affiliate support is enforceable? |
| 2 · Contract | What fees are minimum or take-or-pay, and which carve-outs, termination rights and remedies matter? |
| 3 · Delivery | When can grid power, cooling, commissioning and customer acceptance start contractual billing? Who bears delay? |
| 4 · Collections and margin | How much revenue becomes collected service-fee cash after power pass-through and operating costs? |
| 5 · Financing | What construction draws, debt service, reserves, covenants and refinancing requirements rank ahead of equity? |
| 6 · Equity residual | After maintenance, expansion capex (capital expenditure), debt payments and dilution, what cash or return can shareholders actually receive? |
The F-1 reports that certain customer agreements are entered into with local affiliates and financial support varies. A recognised global brand is not itself proof of a parent guarantee. The underwriter must distinguish these customer-payment protections from the operator's separate performance obligations and remedies.
The financial evidence makes this more than a theoretical checklist. DayOne's unaudited H1 2026 financial statements reported US$512.0 million of revenue and approximately US$14.0 million of net cash provided by operating activities. While operating cash flow was low relative to revenue, the cash-flow statement shows significant cash absorption through increases in receivables and other operating and contract assets during the period. The key investment question is whether cash conversion strengthens as contracted capacity enters service and customer payments are collected.
Those same unaudited interim statements recorded about US$4.48 billion of net investing cash outflows and approximately US$4.49 billion of net financing inflows. Investing cash flow includes items other than physical construction capex; it should not be relabelled as an all-capex figure. [1]
At 30 June 2026, the unaudited interim balance sheet reported total borrowings of approximately US$4.43 billion, excluding finance leases. Management estimated a further US$11.4 billion of development cost to complete projects associated with its Bookings. That estimate covers the booked project portfolio, not simply the 1,328 MW Backlog figure, and is not proof of an unfunded shortfall.
The Johor financing structure makes the funding sequence more concrete. The package combines United States dollar financing with RM (Malaysian ringgit) borrowing. On 13 October, a seven-lender consortium announced an additional RM15 billion in financing (reported by The Tech Capital as approximately US$3.67 billion [4]): a US$2.4 billion offshore term-loan facility and a RM5.5 billion Murabahah (Islamic cost-plus financing) facility. [3] Together with RM15 billion arranged in 2025, the Johor financing platform totals RM30 billion (approximately US$7.2 billion) [2]. These are arranged facility amounts, not proof of amounts drawn or an additional US$7.2 billion to add to 30 June borrowings.
For equity investors, the underwriting questions are post-June borrowing and hedging positions, financing costs, amortisation and covenant headroom, and whether collections can cover debt service during commissioning.
DayOne's filed financing documents identify the Johor borrower and guarantor entities, prepayment terms and financial covenants. Certain distributions funded from EBITDA (earnings before interest, taxes, depreciation and amortisation) under these arrangements require at least 1.2× finance-service coverage and net leverage below 5.2×, among other conditions. Even when customers pay, insufficient covenant headroom may prevent cash from reaching shareholders.
Exhibit 10.8 separately records a DayOne-parent completion and cost-overrun undertaking to the Security Agent, with cost-overrun support extending up to three months after each covered building becomes ready for service. These are lender protections, not customer parent guarantees, and do not establish current debt drawings, actual covenant headroom or distributable equity cash. [5]
There is also an important denominator problem. In the same unaudited H1 2026 statements, electricity costs were US$203.1 million, against US$512.0 million in total revenue. Consumption-based electricity charges are generally passed through to customers. The filing does not disclose the largest customer's split between power pass-through and service fees. A 69.2% share of total revenue should not be assumed to equal 69.2% of operating contribution.
Contract value is visibility, not present value
Under “Visible, De-risked Growth Profile Underpinned by Customer Commitments” and “Certain Definitions,” the F-1 also reports approximately US$34 billion of remaining total contract value as of 20 September: issuer-defined future contracted service fees under executed agreements, excluding recognised revenue and specified unexercised or non-binding commitments. It is not collected cash, a discounted present value or a guaranteed equity return. Nor do we know whether the largest customer represents 69.2% of that contract-value measure. [1]
As a static arithmetic illustration, a hypothetical 10% reduction in the largest customer's revenue, with all else unchanged, would reduce group revenue by 6.92%, or US$35.4 million on the unaudited H1 2026 revenue base. A hypothetical 25% reduction gives 17.30%, or US$88.6 million. Neither scenario predicts a default or estimates EBITDA, credit losses or equity value. Power-cost offsets, contractual remedies, collection timing and re-letting could all alter outcomes.
Geography adds a separate consideration. Malaysia contributed 87.0% of unaudited H1 2026 revenue, but the filing does not quantify how much of that geographic revenue comes from the largest customer. The exposures should be tested jointly where they coincide, not presumed to be identical.
Investment Lens
4 What this means for the broader market
A brief contrast makes the point. Equinix reported that its largest customer represented approximately 2% of H1 2026 recurring revenue, and its largest 50 approximately 36%. [6] DayOne's 69.2% and 84.3% figures measure total revenue, not recurring revenue. Equinix also operates a different, more distributed colocation and interconnection model. The comparison illustrates two customer architectures; it cannot rank returns, credit quality or valuation without comparable economics and pricing.
For infrastructure investors, concentrated offtake is neither inherently a flaw nor automatically a moat. It may enable faster, less speculative development. It can also locate credit, execution and refinancing exposure in a small number of agreements. The decision is whether the contract protections, customer and site economics, delivery history and financing cushion justify the price paid.
An anchor contract can make a megawatt financeable at closing. The harder test is whether that megawatt stays cash-generative for equity through delivery, collections, debt service and reinvestment.
When a handful of agreements underpin billions of dollars of data-centre development, which of those commitments will become durable cash for the operator's investors — and what price should they pay for the uncertainty that remains?
That is the question I'm taking forward next.