AI Equipment Finance Can Match Cash Timing—Not Demand

DigitalOcean has real demand evidence.

Second-quarter revenue rose 29% year on year to $281 million. AI Customer ARR rose 212% to $234 million. Remaining performance obligations reached $894 million, including $366 million that the company expects to recognise over the next 12 months. DigitalOcean also reported its first nine-figure annual customer commitments and said the weighted-average contract life moved from 1.6 years to more than three years.

That is stronger evidence than an order announcement with no customer commitment behind it. The company also generated $110 million of operating cash flow in the quarter and reported $767 million of cash at 30 June.

But the financing claim asks a more exact question than whether demand exists. DigitalOcean says its new equipment facility will more closely align cash outflows with revenue. The public filings show how the cash outflows will be fixed. They do not yet provide the same precision for the AI cash flows expected to meet them.

What the facility actually fixes

The committed facility is $725 million. A further $300 million accordion would take the total to $1.025 billion, but that increase remains conditional on additional lender commitments and other requirements.

Until 10 September 2027, DigitalOcean can request advances funding up to 90% of eligible equipment cost. It supplies the remaining 10% as prepaid rent. Each advance is leased back to the company, paid monthly in arrears and fully amortised by 10 September 2030.

The interest rate is fixed when each advance is made at the applicable term-SOFR swap rate plus 2.75 percentage points. Undrawn commitments also carry fees, and early repayment can attract a premium. The facility is guaranteed by DigitalOcean and certain subsidiaries and secured by the equipment and related collateral.

This structure changes timing. It replaces a large upfront equipment payment with a scheduled claim extending through 2030. It does not remove utilisation risk, customer risk or the possibility that revenue arrives more slowly than the lease amortises.

The headline AI metric is broader than AI product revenue

DigitalOcean’s AI Customer ARR is not a disclosed run rate for AI products alone. The company annualises its most recent quarter of revenue from customers who use one or more AI or machine-learning offerings. That amount includes all of those customers’ infrastructure, platform and software-as-a-service revenue during the period.

The metric is useful for identifying the growth of AI-using customers. It cannot be used as a like-for-like measure of cash generated by the specific equipment financed under the new facility.

RPO is firmer because it represents contracted future services, but it spans the company’s service portfolio rather than only the financed AI equipment. DigitalOcean also cautions that RPO does not capture the timing of customer consumption and can increase when a customer moves from usage-based to commitment-based terms without the change necessarily representing incremental revenue.

The result is not an absence of demand. It is an attribution gap: the fixed equipment obligations are visible more precisely than the revenue stream assigned to the assets.

The obligations already extend beyond the new facility

At 30 June, DigitalOcean carried $577.7 million of finance-lease and equipment-financing obligations on a present-value basis. It expected another $281.6 million of undiscounted payments from server leases commencing in July, with a weighted-average term of 4.9 years.

The company also disclosed $2.76 billion of estimated undiscounted fixed payment obligations, primarily for colocation space not yet commenced. Those leases were scheduled to begin between July 2026 and June 2028 and had a weighted-average term of 11.2 years.

These amounts cannot be added mechanically and compared with AI Customer ARR or RPO. They cover different assets, service periods and business activity, and some were not yet recognised on the balance sheet. They do show why financing availability and contract quality must be assessed together: the infrastructure footprint creates payment commitments across several layers before all of the associated revenue is recognised.

The countercase is substantial

This is not a distress-financing story. DigitalOcean raised $887.9 million in a follow-on equity offering during the first half and repaid $500 million of term-loan principal. It ended June with positive net income, operating cash flow and adjusted free cash flow.

The customer evidence is also meaningful. The company has signed large annual commitments, expanded RPO and reported strong growth in its highest-spending cohorts. Equipment finance can be a rational way to align the useful life of servers with multi-year customer revenue while preserving cash for the rest of the platform.

There is, however, an important measurement boundary. DigitalOcean’s adjusted free cash flow excludes equipment acquired under financing arrangements, finance leases and future contractual commitments. The company itself says the measure does not represent residual cash available after debt obligations. Positive adjusted free cash flow therefore does not settle whether the financed capacity will earn an adequate return after all fixed claims.

What BoomRisk will watch

• Facility drawdown. How much of the committed $725 million is used, whether the accordion is exercised and the weighted financing cost when advances are fixed.
• Contract conversion. Whether the $366 million of expected 12-month RPO recognition arrives alongside the facility’s monthly amortisation.
• AI attribution. Whether DigitalOcean begins separating AI product revenue from the broader revenue of customers who use any AI offering.
• Customer concentration. Whether the top 25 customers, which supplied about 20% of second-quarter revenue, remain durable as the capacity base expands.
• Cash returns. Whether depreciation, interest, colocation costs and lease principal allow operating cash generation to keep pace with the larger infrastructure footprint.
• Full agreements. The 8-K summarises the facility, but DigitalOcean said the complete equipment-finance agreements would be filed with its September-quarter Form 10-Q.

This paper does not change the authoritative BoomRisk score. At publication, Capex & Cash Returns and AI Monetisation remain at 4/5, Credit Stress reads 3/5, and strong semiconductor demand remains the clearest counterweight to the elevated-risk reading.

The conclusion

DigitalOcean has shown that customers are committing to more cloud and AI capacity. It has also found lenders willing to finance the equipment required to serve them.

Those are genuine strengths. They answer whether the buildout can continue.

They do not yet answer whether the financed assets are matched by independently attributable AI revenue, utilisation and cash returns through 2030. Equipment finance can align the schedule of cash payments. The investment case still depends on whether demand aligns with the obligations.

BoomRisk is a financial-risk monitoring framework, not an investment recommendation or market forecast. AI Customer ARR, adjusted EBITDA and adjusted free cash flow are company-defined non-GAAP or operating metrics. Facility capacity, customer commitments and reported RPO do not establish future utilisation, margins or returns.