What are the best neoclouds to invest in?

What counts as a neocloud?

A neocloud is a cloud provider built primarily around GPU-as-a-service and AI/HPC workloads rather than a broad, general-purpose cloud catalogue. The category is differentiated by specialized GPU clusters, high-performance networking, data services and orchestration software. The market is still geographically concentrated: one industry forecast assigns 88% of 2026 neocloud GPUaaS revenue to one region, declining to 72% by 2030 as sovereign-cloud capacity expands. [1]

The investment case is straightforward: AI training and inference require scarce accelerated-computing capacity, while many customers prefer an independent provider for speed, flexibility or access to capacity unavailable from the largest hyperscalers. The counter-case is equally important: neoclouds must finance expensive GPUs and data centers before utilization and pricing are proven.

Public-market ranking

1. Nebius — best risk/reward starting point

Nebius reported Q2 2026 revenue of $582.3 million, up 454% year over year, and adjusted EBITDA of $236.2 million, versus a $21.0 million loss in Q2 2025. Cash and cash equivalents were $8.042 billion at June 30, 2026. [2] It describes its offering as a full-stack AI platform spanning data, model training and production deployment. [2]

That combination—very rapid growth, positive adjusted EBITDA and a large liquidity cushion—makes Nebius the strongest balanced candidate in this comparison. The judgment is not the same as saying the stock is cheap: the research set does not provide a current share price, market capitalization or valuation multiple, so valuation must be checked separately before investing.

The principal risks are execution and capital allocation. Nebius is expanding data-center capacity and requires sites, power and financing; its own release warns that forward-looking expectations are not guarantees of future performance. [2] Investors should therefore monitor GPU deployment, utilization, pricing, cash burn and dilution—not just revenue growth.

2. CoreWeave — best scale exposure, but leverage matters

CoreWeave reported Q2 2026 revenue of $2.575 billion, making it the largest disclosed quarterly revenue base among the directly comparable public candidates in the evidence reviewed. [3] Its business is a concentrated way to express the view that AI infrastructure demand will remain strong.

CoreWeave's attraction is scale and operating focus. Its customer disclosures include major enterprises and AI pioneers, including Bentley Systems, Caterpillar, Grammarly, Isomorphic Labs and Sunday Robotics, as well as expanded relationships with Cognition, Databricks, Hudson River Trading, Periodic Labs, Rescale and Runway ML. [4]

The trade-off is financial and operational intensity. A neocloud's GPUs, networking and facilities require substantial upfront capital, and the business can be vulnerable to utilization shortfalls, refinancing pressure, falling GPU rental prices or customer concentration. CoreWeave is therefore better suited to an investor who accepts higher volatility and balance-sheet risk in exchange for greater scale exposure.

3. Applied Digital — contract-backed upside with project risk

Applied Digital signed a 15-year take-or-pay lease with a high-investment-grade U.S. hyperscaler for 300 MW of critical IT load at its Delta Forge 1 AI Factory campus. The company says the lease represents approximately $7.5 billion of base-term contracted revenue, with initial operations expected in calendar year 2027. [5]

This is attractive because contracted revenue can reduce demand uncertainty relative to a purely on-demand GPU cloud. However, the contract is not the same as current revenue or current cash flow: delivery depends on construction, power, equipment, financing and commissioning. Applied Digital should consequently be evaluated as an infrastructure-development and execution story, not as a mature software company.

4. IREN — speculative AI conversion

IREN's FY26 AI Cloud Services revenue increased approximately eightfold to $128.8 million, while the company described FY26 as an ongoing transition from Bitcoin mining to AI Cloud Services. [6]

The upside is that IREN may repurpose renewable-powered data-center infrastructure toward a higher-growth AI market. The risk is that the AI business is still a transition rather than a demonstrated, mature earnings engine. Investors must underwrite both AI-cloud execution and residual Bitcoin economics, making IREN the most speculative of the four public candidates.

Private-company watchlist

Crusoe is the strongest private-company candidate in the evidence reviewed from a disclosed financing and valuation standpoint: it announced a $1.375 billion Series E at a $10 billion valuation and describes itself as an AI-factory company. [7] Fluidstack reportedly raised an $830 million Series A at a $7.5 billion valuation, led by Situational Awareness. [8] Nscale is developing vertically integrated AI infrastructure spanning GPU compute, networking, data services and orchestration software across Europe, North America and Asia. [9] Runpod announced a $20 million round led by Intel Capital and Dell Technologies Capital. [10]

These are not direct substitutes for the public stocks. Private valuation marks may not be independently comparable to public-market capitalization, and private shares may be unavailable, illiquid or restricted. They belong on a venture, secondary-market or future-IPO watchlist unless an investor has legitimate access and can evaluate the actual terms.

How to choose among them

For a diversified public portfolio: start with Nebius, subject to valuation discipline. Its reported Q2 profitability and cash position provide a better cushion than a pure growth narrative, although adjusted EBITDA is not GAAP net income.

For maximum AI-infrastructure operating exposure: consider CoreWeave only if the portfolio can tolerate leverage, dilution and large drawdowns. Scale does not eliminate financing or utilization risk.

For contract visibility: Applied Digital is the clearest candidate, but the key underwriting question is whether the company can deliver the contracted campus on schedule and within an acceptable financing structure.

For high-risk optionality: IREN offers AI-conversion upside, but its legacy Bitcoin exposure makes the thesis more complex and less pure.

Risks that apply to the whole group

1. Capital intensity and financing risk. An on-demand neocloud without contracted revenue must convince lenders that future cash flow will cover loan payments without guaranteed income. [11]

2. Technology and pricing risk. GPU generations can become obsolete, supply can normalize, and customers may shift workloads to hyperscalers or internal infrastructure.

3. Power and construction risk. The scarce input is not only GPUs; sites, electricity, grid interconnection, cooling and networking can delay deployments.

4. Operational concentration. AI data-center upgrades can create multiday cloud outages, while AI costs, data lock-in and operational risk remain material concerns. [12]

5. Valuation risk. The strongest operating company is not automatically the strongest investment if its share price already discounts years of growth. Current valuation data was not established in this research, so no price-based buy recommendation is warranted.