How is a neocloud different from a hyperscaler?

A hyperscaler runs a general cloud with its own applications and funds capacity from a large existing business. A neocloud sells GPU hours only and funds growth with project debt and equity. Same customers, opposite balance sheets.

Why are neoclouds the first place credit stress shows?

Their interest is paid from financing rather than from operations, their assets depreciate in three to five years, and their revenue depends on a handful of large contracts. Any change in demand pacing or rental prices reaches their lenders before it reaches a hyperscaler’s.

Which neoclouds does Closelook track?

CoreWeave, Nebius, TeraWulf and Applied Digital on the credit tape; the same names plus DigitalOcean, Equinix and Zhipu in the compute group of the Agentic Ecosystem index.

Glossary term

Neocloud

A specialist cloud provider that rents GPU capacity for AI workloads rather than offering a general cloud — CoreWeave, Nebius, Lambda and similar. Neoclouds grow on contracted capacity financed by project debt, which places them on the most funding-dependent rung of the AI ladder.

AI-generated — produced automatically by Closelook’s systems under this site’s editorial policy.

What it means

A neocloud does one thing: it buys accelerators, racks them in leased or owned data centres, and rents the capacity to model labs, enterprises and sometimes to hyperscalers themselves. Its revenue is contracted — multi-year capacity agreements — and its assets are the GPUs, which depreciate fast. Because the business is capital-intensive and young, most of the build is financed with debt secured on the contracts and the hardware, plus equity raised at intervals.

The model works while GPU rental prices hold, contracts renew and lenders keep extending. It is stressed when rental prices fall faster than the hardware depreciates or when a large customer’s demand is managed rather than raced.

Why it matters for the AI trade

Neoclouds are the project-funded rung (Tier 3) of Closelook’s funding ladder: on the credit stress tape CoreWeave, Nebius, TeraWulf and Applied Digital show operating cash flow covering 0.18× of capex, new debt at 1.10× capex and interest not serviced from the business. Their stress score of 86 in September 2026 was read as normal cycle behaviour — the rung where stress is expected to show first. They are also the compute group of the Agentic Ecosystem index, which fell while its security group rose on 14 September.

How Closelook uses it

The compute-tightness and spot-rental boards track the price of the capacity neoclouds sell; the GPU depreciation entry covers the accounting question behind their asset values. A neocloud whose residual widens on the tape is the earliest credit signal the AI trade would give.

Common questions

How is a neocloud different from a hyperscaler?: A hyperscaler runs a general cloud with its own applications and funds capacity from a large existing business. A neocloud sells GPU hours only and funds growth with project debt and equity. Same customers, opposite balance sheets.
Why are neoclouds the first place credit stress shows?: Their interest is paid from financing rather than from operations, their assets depreciate in three to five years, and their revenue depends on a handful of large contracts. Any change in demand pacing or rental prices reaches their lenders before it reaches a hyperscaler’s.
Which neoclouds does Closelook track?: CoreWeave, Nebius, TeraWulf and Applied Digital on the credit tape; the same names plus DigitalOcean, Equinix and Zhipu in the compute group of the Agentic Ecosystem index.