Nebius Seeks $4.5B for AI Compute
Nebius’ planned $4.5B convertible-note raise shows AI infrastructure financing becoming more debt-heavy and capex-intensive.
Nebius is not a clean private-startup funding story, but it is a strong AI infrastructure finance signal. The amount alone shows how capital-intensive the AI compute race has become.
What happened
Amsterdam-based AI infrastructure provider Nebius said it is looking to raise $4.5B through convertible notes.
The planned financing includes $2.75B due in 2030 and $1.75B due in 2034. The proceeds are intended for data-centre buildout, AI cloud investment and GPUs.
Nebius sits in the neocloud category: infrastructure providers trying to meet demand for AI compute outside the largest hyperscalers. These companies need huge upfront capital because GPUs, data-centre capacity, power contracts and networking infrastructure are expensive long before customers generate enough revenue to cover the investment.
Why it matters
AI infrastructure is increasingly being financed like industrial infrastructure. Equity alone is often not enough. Companies are turning to debt, convertible notes, lease structures and long-term customer commitments to fund buildout.
That creates opportunity, but also risk. Compute demand is high today, yet infrastructure companies must manage hardware depreciation, utilisation, customer concentration and power availability. A large financing package can accelerate growth, but it also increases execution pressure.
The bigger picture
The AI boom is moving from software economics into capital-markets economics. The question is no longer only which model is best, but who can finance, build and operate enough compute to serve demand profitably. Nebius’ planned raise shows how the AI stack is becoming more infrastructure-heavy, with financing strategy becoming just as important as technical strategy.