Monday, November 16, 2026 · Week 47 DE · EN · FR · ES Dark
AI

Gartner forecasts a wave of neocloud takeovers

Spending on AI cloud infrastructure is rising. According to Gartner, more than 80 percent of neoclouds will lose their independence by 2030.

By Alec Chizhik October 21, 2026 4 min read
Gartner forecasts a wave of neocloud takeovers

According to a new Gartner forecast, more than 80 percent of neocloud providers will cease to operate independently by 2030. The stated reasons are cost pressure and a lack of differentiation in business models. For most providers, the growing AI cloud infrastructure business will not bring in enough to amortize their chips before the next generation.

Key takeaways

  • According to Gartner’s forecast, more than 80 percent of neocloud providers will cease to operate independently by 2030. Cost pressure and a lack of differentiation in the business model are cited as the reasons.
  • Spending on AI-optimized IaaS will rise from around 59 billion euros in 2027 to around 155 billion euros in 2030. For most providers of AI compute, this growth still will not be enough to amortize the chips before the next generational change.
  • Hyperscalers may be able to carry AI investments for years from other revenue. Neocloud providers, by contrast, must generate revenue from AI-optimized IaaS quickly to keep their operations running.

Related:Reply shortens Porsche’s troubleshooting with a context graph  /  Kubermatic CEO Scheele: Running AI models like software

Growth and consolidation at the same time

The market for AI compute from the cloud is growing strongly. According to the forecast by market research firm Gartner, spending on AI-optimized IaaS will rise from around 58.74 billion euros in 2027 to around 154.86 billion euros in 2030. IaaS stands for Infrastructure as a Service, meaning rented compute, storage and network capacity. The AI-optimized variant is designed for graphics processors and other AI accelerators.

For the specialized providers, the forecast published on Wednesday is nonetheless sobering. According to the forecast, more than 80 percent of neocloud providers will cease to operate independently by 2030. The reasons are cost pressure and a lack of differentiation among business models. Neoclouds are cloud providers that have specialized in GPU capacity for AI applications.

What is a neocloud? Neoclouds are cloud providers that have specialized in compute capacity for AI applications. Their core business is renting out GPU capacity as an infrastructure service, known in industry jargon as AI-optimized IaaS. Hyperscalers, by contrast, offer AI compute as part of a broad cloud portfolio.

Hyperscalers and neoclouds compared

When it comes to financing, the paths of neoclouds and hyperscalers, the operators of the large global cloud platforms, diverge. According to Gartner’s assessment, hyperscalers have extensive and diversified sources of revenue. These may be able to carry AI investments for years.

The forecast gives neocloud providers no such leeway. They must generate revenue from AI-optimized IaaS and the associated platforms quickly to keep their operations running. The time pressure is closely tied to the hardware itself.

Chip cycle and impairment risk

By Gartner’s calculation, the forecast spending growth will not be enough for most providers of AI compute to amortize their investments before the current generation of AI semiconductors is replaced. Amortizing means recouping the purchase costs through revenue. For GPU clusters, this calculation works against the pace of new chip generations.

At the level of the overall market, the sum at stake is far larger. AI infrastructure spending equivalent to more than around 3,120 billion euros could be exposed to impairment risk by the end of 2030, according to Gartner’s forecast. That applies if AI providers keep up their investment pace. An impairment means that companies must write down the book value of assets on an unscheduled basis.

Nebius finances itself through customer contracts

Nebius shows how a specialized provider organizes its capital needs. This year, the company has closed its first senior secured credit facility of around 689.75 million euros. A credit facility is a line of credit. Senior secured means that the lenders are paid first when the collateral is liquidated. Nebius wants to use the money to accelerate the global expansion of its AI cloud.

The collateral consists of installed GPU infrastructure and contractually agreed payment streams from a customer with an investment-grade rating, meaning a credit standing assessed as solid. The financing therefore rests on revenue from the AI business that has already been agreed. In March, NVIDIA had also announced that it would invest around 1.78 billion euros in Nebius.

CoreWeave plans expansion in India

CoreWeave is relying on a local partner. Last week, the provider announced that, together with AdaniConneX, it wants to expand its AI cloud by 240 megawatts of data center capacity in Navi Mumbai. The expansion is planned. The NVIDIA Vera Rubin Platform is to be used at the site.

Nebius and CoreWeave are thus expanding during the very period in which, according to Gartner’s forecast, more than 80 percent of neocloud providers will cease to operate independently. One relies on secured loans and a chip manufacturer as an investor. The other relies on a data center partner.

Frequently asked questions

Why is NVIDIA investing in Nebius?

NVIDIA justified the announced investment of around 1.78 billion euros by citing its confidence in Nebius’s business. As a second reason, the chip manufacturer named the provider’s technical expertise across the entire AI technology stack.

When does the Nebius credit facility expire?

The credit facility runs until 31 October 2030. It therefore ends in the same year by which, according to Gartner’s forecast, more than 80 percent of neocloud providers will cease to operate independently.

What does CoreWeave plan to use the capacity in India for?

The NVIDIA Vera Rubin Platform in Navi Mumbai is intended to cover AI training, inference, reasoning processes and agentic workloads. Inference is the ongoing operation of trained models. Agentic workloads come from AI systems that execute multi-step tasks autonomously.

Image source: AI-generated (October 2026)

Translated from the German original using artificial intelligence. The German version is authoritative.

Also available in

EspañolDeutsch
MBF Media Newsletter

The monthly briefing for decision-makers

Once a month, the MBF Media Newsletter gathers what matters from cloudmagazin, MyBusinessFuture, Digital Chiefs and SecurityToday, curated by the editorial team.

Around 23,500 IT and business decision-makers read this newsletter. Read along.

Subscribe for free
MBF Media Newsletter, aktuelle Ausgabe auf dem iPhone
A magazine by Evernine Media GmbH