Nebius Q2 AI Cloud Surges 514%; Morgan Stanley Questions 5GW Expansion Feasibility

Woofun AI reports that Nebius delivered robust second-quarter earnings anchored by its AI Cloud division, yet Morgan Stanley maintains an Equal-weight rating despite raising financial forecasts, signaling deep skepticism regarding the feasibility of the company's ambitious 5GW expansion plan.

The financial results for the second quarter of 2026 reveal a dramatic acceleration in top-line performance. Total revenue reached $582.3 million, marking a 454% increase year-on-year. The primary engine of this growth was the AI Cloud segment, which generated approximately $575 million, representing a 514% year-on-year rise and accounting for around 99% of total revenue. This concentration underscores the company's pivot toward infrastructure services.

Furthermore, annualized recurring revenue (ARR) climbed to $3 billion at the end of the quarter, up about 56% from $1.92 billion recorded at the end of the first quarter, indicating strong retention and expansion within the existing customer base.

Strategic capacity targets have been aggressively upgraded to match this revenue trajectory. Nebius now plans to secure 5GW of power capacity by the end of 2026, exceeding its previous target of "over 4GW." Starting in 2027, the company aims to deploy more than 1GW of active capacity each year.

This shift signals a transformation from a rapidly growing GPU cloud service provider into a larger-scale AI infrastructure operator.

However, the distinction between contracted power and active capacity is critical; the former represents secured resources for future data centers, while the latter denotes operational hash rate capable of generating immediate revenue.

Morgan Stanley's analyst stance reflects a cautious optimism tempered by execution risks. The firm retained an Equal-weight rating and a target price of $144, even after raising forecasts for future revenue, adjusted EBITDA, and capital expenditures. The rationale is straightforward: the market has already priced in the demand surge, and the remaining uncertainty lies in whether Nebius can convert contracted power into actual hash rate that generates revenue on time. The focus has shifted from order acquisition to the physical and logistical realization of capacity.

Deal dynamics in the second quarter highlight the intensity of current market demand. Nebius secured four significant AI Cloud deals, with contract durations ranging from 1 to 3 years and an average deal size of over $1 billion. The annualized contract value of these new deals exceeded $20 million/MW.

Notably, some short-term capacity deals lasting 3 to 6 months in the third quarter surpassed $40 million/MW. The total value of new contracts in the second quarter was roughly four times that of the previous quarter, while the total value of contracts from new customers increased by over 9 times, demonstrating rapid market penetration.

Woofun AI data shows that cash flow efficiency has improved alongside deal volume, though timing remains a constraint. The estimated payback period for new deals in the second quarter was about 1 year and 10 months, significantly shorter than the previous 2 to 3 years. This acceleration is vital for an AI Cloud provider that must purchase GPUs, build data centers, and secure power supplies before recovering investments through leases. Higher contract prices, shorter payback cycles, and some customers' willingness to pay upfront payments help alleviate cash pressure.

However, most of these new contracts will not go live until later in 2026, limiting their impact on annual revenue, with significant effects expected in 2027 and beyond.

Long-term forecasts reflect the scale of upcoming capital deployment. Morgan Stanley expects active capacity to rise from 170MW in 2025 to 780MW in 2026, 1,680MW in 2027, and 2,620MW in 2028. Consequently, revenue forecasts for FY27 and FY28 were raised to $10.926 billion and $19.059 billion respectively, with adjusted EBITDA projections set at $6.294 billion and $12.108 billion. These higher revenue projections necessitate substantial investment; capital expenditures for FY27 and FY28 are expected to reach $36.980 billion and $41.066 billion respectively, representing increases of 19.0% and 13.5% from previous forecasts.

Operational intensity in the second quarter illustrates the capital-heavy nature of this growth. Nebius's adjusted EBITDA reached $236.2 million, with a profit margin of 40.6%, significantly higher than the market's expectation of around 30%. Operating cash flow stood at $2.2461 billion, but expenses for purchasing property, equipment, and intangible assets amounted to $5.657 billion. This disparity shows that while operations generate substantial cash, construction pace and spending scale outpace internal generation. A positive indicator is the pricing of Blackwell hash rate; management cited that the price in the first capacity auction was 15% higher than the highest previously charged price, with some short-term deals reaching twice the average level of new deals in the second quarter.

Risks and constraints remain significant despite strong performance. Scenario-based valuations vary widely: the bull market target price is $400, the baseline scenario valuation range is $130 to $300, and the bear market target price is $70. Under a bull case, Nebius could deploy over 5GW of power capacity by 2030.

However, three main constraints exist. First, financing via asset-backed financing, corporate debt, or equity-linked instruments is required to fund the 1GW+ annual active capacity deployment starting in 2027. Second, approval and construction delays pose risks, exemplified by the 300MW planned Vineland project in New Jersey, which requires public hearings and local approvals. Third, competition from large cloud providers such as Microsoft, Amazon, and Google threatens to compress the short-term price premium once GPU supply becomes more abundant.

The core challenge for Nebius is no longer securing orders but executing deployment. The 5GW contracted power target implies massive subsequent challenges in building, deploying, and converting capacity into cash flow on time. This execution gap is the primary reason Morgan Stanley raised forecasts but did not increase the target price, highlighting that the shift from order acquisition to physical realization is the defining risk for the company's next phase of growth.