Strong AI computing demand cannot mask data center delivery delay risks; Morgan Stanley maintains Nebius (NBIS.US) "Equal-weight" rating and US$144 price target.

Morgan Stanley released a research note after AI cloud computing service provider Nebius Group (NBIS.US) reported its third-quarter results, stating that although the company's revenue and profitability both beat market expectations, AI computing demand remained strong, and customer contract scale and computing leasing prices rose further, delays in third-party data center deliveries led the company to lower its 2026 revenue guidance and defer part of its revenue growth to the first half of 2027. Considering the execution risks facing data center construction and computing deployment, as well as the financing pressure from future large-scale capital expenditures, Morgan Stanley maintained its "Equal-weight" rating and US$144 price target on the stock, implying approximately 12% upside from its November 12 closing price of US$128.64. The earnings report showed that Nebius's third-quarter revenue rose 571.3% year over year to US$980.8 million, above market expectations. Among this, the company's core AI cloud computing business revenue rose 642.8% year over year to approximately US$974 million, accounting for more than 99% of total group revenue. On profitability, Nebius's third-quarter adjusted EBITDA reached US$465.8 million, with an adjusted EBITDA margin of 47.5%, up from 40.6% in the previous quarter and also better than market expectations. According to Morgan Stanley's financial model, the company's third-quarter adjusted earnings per share were US$0.31. However, Morgan Stanley noted that Nebius's third-quarter revenue included US$325 million in non-recurring revenue related to its asset-light operating model, accounting for about one-third of total revenue for the quarter. Therefore, despite the company's strong third-quarter financial performance, investors still need to pay attention to the growth of its recurring cloud computing revenue. Morgan Stanley analysts Ryan Lountzis and Adam Wood noted that the positive signals from Nebius's earnings were mainly concentrated in three areas: AI computing demand, pricing power, and financing arrangements. First, the tight supply-demand balance in the AI computing market is still driving Nebius to raise service prices. In the third quarter, the annual contract value per megawatt (MW) of computing capacity exceeded US$25 million, up from US$20 million to US$25 million in the second quarter. At the same time, the average hourly GPU rental price rose more than 20% quarter over quarter. Since the beginning of the third quarter, the total value of customer contracts won by Nebius has exceeded US$12 billion. Management said that the scale of contracts signed by customers is continuing to expand, prices for 2027 computing capacity are still rising, and some customers have even begun seeking to lock in computing resources for 2028. Morgan Stanley believes that these indicators show that, despite growing market concerns about the scale and returns of AI infrastructure investment, Nebius still faces strong customer demand and is able to secure more favorable commercial terms through higher pricing. However, data center delivery delays are becoming the main factor constraining Nebius's short-term performance growth. Nebius said that because the delivery timelines for several third-party colocation data center projects have been postponed to the first half of 2027, part of the computing capacity originally planned to enter operation this year and the related revenue will also be delayed accordingly. Among these, the data center project in Vineland, New Jersey, which is primarily intended to serve Microsoft Corporation (MSFT.US), has become a market focus. It is understood that the third-party data center supplier for the project is gradually phasing out some temporary power facilities while advancing the deployment of Bloom Energy (BE.US) fuel cell power systems. This adjustment has delayed part of the computing capacity originally scheduled for delivery to Microsoft Corporation this year. Nebius currently expects to complete delivery of more than 50% of the capacity under its computing contract with Microsoft Corporation by the end of 2026, with the remainder planned for delivery in the second quarter of 2027. Affected by the above factors, Nebius adjusted its full-year 2026 revenue guidance to US$3.0 billion to US$3.2 billion from the previous US$3.0 billion to US$3.4 billion, lowering the upper end of the guidance by US$200 million. At the same time, the company expects to achieve annualized recurring revenue (ARR) of more than US$7 billion by February 2027, compared with its previous target of US$7 billion to US$9 billion in ARR by the end of 2026. Morgan Stanley believes that although the relevant revenue is mainly deferred due to changes in the timing of computing deployment rather than weakening customer demand, the recent project delivery delays highlight the uncertainty over whether Nebius can convert signed computing capacity into actual revenue in a timely manner in an environment constrained by the supply of data centers, power, and other infrastructure. Despite the impact on short-term delivery progress, Nebius management maintained its long-term computing expansion targets unchanged. Currently, Nebius has signed power capacity of more than 4 gigawatts (GW), and the company still expects to increase this figure to 5 GW by the end of 2026, with plans to add more than 1 GW of operational capacity in 2027. To achieve this goal, Nebius recently announced several new data center projects and expansion plans, including adding 300 MW of capacity in Missouri, bringing the total signed project capacity there to 1.1 GW; advancing a multi-campus AI data center project with total capacity of 100 MW in Alberta, Canada; and adding a 50 MW third-party colocation data center project in South Carolina. Management noted that as more computing capacity in the future comes from the company's own data centers, Nebius will be better able to control project construction and delivery timelines, thereby reducing its reliance on third-party data center suppliers. In addition to computing deployment progress, Nebius's financing plan to support large-scale business expansion is another key focus of the earnings report. Nebius expects capital expenditures in 2027 to reach US$40 billion to US$45 billion. Of this, approximately 75% to 85% of funding needs are expected to be met through customer prepayments and asset-backed financing related to contracts with existing investment-grade-rated customers. The company expects customer prepayments to reach US$17 billion to US$20 billion in 2027, and this amount is not a cap. It is worth noting that customer prepayments from newly signed contracts in the third quarter are already sufficient to cover more than 100% of the capital expenditure needs of the related projects, compared with only 50% to 60% in the second quarter. Management said this trend continued into the fourth quarter. Morgan Stanley believes that customers' willingness to provide a higher proportion of prepayments reflects, on the one hand, large technology companies' urgent need to lock in future AI computing resources, and on the other hand, helps Nebius reduce its reliance on external financing during business expansion. However, even if existing financing arrangements can cover most capital expenditures, Nebius still needs to raise additional funds for future data center construction. Considering recent changes in the credit market and the company's massive investment scale over the next several years, financing costs and the stability of financing channels will remain factors investors need to monitor closely. Based on the latest earnings report and computing deployment plans, Morgan Stanley adjusted its financial forecasts for Nebius over the next several years. The bank lowered its 2026 revenue forecast for Nebius to US$3.050 billion from US$3.389 billion, a decrease of 10%; raised its 2027 revenue forecast to US$12.025 billion from US$10.926 billion, an increase of 10.1%; and raised its 2028 revenue forecast to US$21.018 billion from US$19.059 billion. On profitability, Morgan Stanley expects Nebius's adjusted EBITDA to increase from US$1.220 billion in 2026 to US$6.921 billion in 2027, and further to US$13.348 billion in 2028. Over the same period, adjusted EBITDA margins are expected to reach 40%, 57.6%, and 63.5%, respectively. However, with capital expenditures and financing costs rising, Morgan Stanley expects Nebius to continue facing significant cash flow pressure over the next several years. The bank expects the company's free cash flow to be negative US$21.141 billion, negative US$25.113 billion, and negative US$27.718 billion in 2026, 2027, and 2028, respectively. On valuation, Morgan Stanley maintained its US$144 price target for Nebius, based on the company's 2028 EBITDA forecast of approximately US$13.3 billion, applying a 6x enterprise value-to-EBITDA (EV/EBITDA) multiple, and taking into account the net debt level at the end of 2027. The bank noted that although Nebius's vertically integrated AI cloud computing platform has gained recognition from large customers, and its lower total cost of ownership and the monetization potential of its software business are expected to drive long-term margin and free cash flow improvement, its valuation multiple should be below the software industry average given the capital-intensive nature of its business. In a bull case, if Nebius can successfully bring more than 5 GW of power capacity into operation by the end of 2030, further expand its customer base, and improve profitability through cost advantages and software business monetization, Morgan Stanley believes the stock could reach US$400. Conversely, if AI computing supply and demand gradually move toward balance, competition among large cloud computing service providers intensifies, and data center delivery difficulties exceed expectations, the stock's valuation in a bear case could fall to US$70. Morgan Stanley believes that Nebius's current customer demand and pricing trends remain encouraging, but in an environment constrained by computing infrastructure supply, whether the company can complete large-scale data center construction on time, convert signed capacity into actual revenue, and obtain sustained financing at reasonable cost will be key factors determining its future performance and valuation.