Nebius, a leading European AI infrastructure provider, stated it needs $250 billion in funding to execute its planned AI data centre and compute expansion through 2030.
AI infrastructure companies face an increasingly complex investment equation. As demand for computing power required to run artificial intelligence models accelerates, financing needs and capital expenditures are rising just as rapidly. Competition has evolved beyond a technological race into a broader battle for capital, power, and the speed at which data centers can be built and brought online.
Nebius estimates that deploying 5 gigawatts of contracted AI infrastructure capacity by 2030 will require nearly $250 billion in financing. The company has financed approximately 20% of this total requirement and still needs to raise roughly $200 billion. Speaking at the Citi Global Technology, Media and Telecommunications Conference, Nebius CEO Arkady Volozh identified financing and construction as the two main challenges facing the company's expansion. His estimate implies a cost of approximately $50 million per megawatt, covering data centers, graphics processing units (GPUs), and supporting infrastructure.
Nebius is pursuing a multi-channel financing strategy combining corporate financing and asset-backed funding to limit pressure on its balance sheet. In August, the company priced $5 billion of convertible notes, including $3 billion of senior unsecured notes due in 2030 at 0.50% interest and $2 billion of notes due in 2034 at 4.50%. This offering followed a roughly $775 million asset-backed loan secured by deployed GPUs and customer cash flows, carrying interest at one-month Term SOFR plus 250 basis points and maturing in October 2030.
Nebius ended the second quarter of 2026 with approximately $8 billion in cash and cash equivalents after raising nearly $2.8 billion through its at-the-market equity program. The company expects customer prepayments to exceed $9 billion this year. Upfront customer payments covered between 50% and 60% of capital expenditures associated with contracts during the second quarter, reducing reliance on debt or additional equity issuance.
Customer prepayments have emerged as a critical financing source for AI infrastructure companies, enabling them to begin building projects before collecting full future revenues. Approximately 70% of Nebius' second-quarter deals included upfront payments, expected to provide more than $9 billion in funding during 2026. The company generated $2.3 billion in operating cash flow during the second quarter, while its committed backlog stood at approximately $40 billion. Management believes this contracted backlog could support additional opportunities for asset-backed financing.
Nebius has also expanded its strategy beyond directly financing and owning data centers. The company opened its cloud infrastructure to partner-owned facilities under a model in which partners finance and own the data centers and hardware while Nebius provides the platform, software, and architecture required to operate them. This approach allows the company to expand without bearing the full capital expenditure burden on its own balance sheet.
Volozh said contracts ranging from one to three years generate approximately $20 million to $25 million in revenue per megawatt. Shorter-term agreements signed by customers seeking immediate computing capacity can generate between $40 million and $50 million per megawatt. At current pricing levels, these rates support Nebius' estimated capital expenditure payback period of around 22 months. However, Volozh warned that prices could decline as additional capacity enters the market and supply and demand move toward balance.
Nebius reported second-quarter revenue of $582 million, representing a 454% increase from the same period a year earlier. The company maintained its 2026 capital expenditure guidance of between $20 billion and $25 billion. Nebius expects connected power capacity to reach between 800 megawatts and 1 gigawatt by year-end, with plans to add more than 1 gigawatt of new capacity in 2027. However, these additional power resources will not immediately translate into revenue. Data centers must first be equipped with GPU clusters and integrated into Nebius' cloud platform—processes requiring time, capital, and operational expertise.
Nebius faces aggressive expansion from specialized AI infrastructure providers such as CoreWeave, as well as technology giants including Microsoft and Amazon. CoreWeave expects its 2026 capital expenditures to reach between $35 billion and $39 billion, up from previous guidance of $31 billion to $35 billion. The company's capital expenditures reached $9.4 billion in the second quarter, while construction in progress increased to $11.9 billion. Its expansion model relies heavily on long-term contracts, with capital expenditures front-loaded at the contract's beginning, requiring a combination of debt, customer prepayments, and corporate-level capital.
Microsoft possesses significantly greater financing and operational capacity than its competitors. The company spent approximately $41 billion on capital expenditures during the fourth quarter of fiscal 2026, with roughly two-thirds allocated to short-lived assets, particularly CPUs and GPUs, while the remainder went toward long-lived assets. Microsoft expects quarterly capital expenditures to exceed $50 billion in the first quarter of fiscal 2027 and has raised its estimated 2026 capital spending to approximately $175 billion. During the same quarter, Microsoft added 31 data centers across five continents, along with another gigawatt of capacity, and expects to double its overall capacity within two years.
Although customer prepayments, operating cash flow, and contracted backlog provide significant support for Nebius' investment program, the company's ability to generate sustainable returns depends on several factors, particularly execution speed, continued demand for AI services, and its ability to scale profitably. Risks increase as more competitors enter the market, while computing capacity prices could decline as supply expands. Energy, equipment, and financing costs remain significant factors affecting the economics of new AI infrastructure projects.
The AI infrastructure market has undergone a broader transformation. It is no longer simply an exceptional growth opportunity; it has become a highly capital-intensive industry requiring massive investments well ahead of revenue generation. Companies face the challenge of striking a delicate balance between securing customer commitments and upfront payments, obtaining sufficient financing, and ensuring that the computing capacity being built is actually utilized.
Nebius shares have fallen approximately 12% over the past month, compared with a 4.2% decline in the broader Internet Software and Services industry. The stock is trading at a forward price-to-sales multiple of approximately 5.95 times, compared with an industry average of 4.13 times. Despite the share price decline, analysts have raised their 2026 earnings estimates for Nebius over the past 60 days. The stock currently carries a "Hold" rating from Zacks Investment Research.
These indicators highlight the conflicting forces shaping the sector. Strong operating growth and rapidly rising revenues continue to attract investors, but the enormous scale of capital expenditure, reliance on external financing, and the potential for pricing pressure are creating challenges for valuations and profit margins.