Neocloud at a Crossroads: CoreWeave, NBIS, and the High-Stakes Race to Power AI’s Future
The Battle for AI’s Backbone Is On. Two Very Different Contenders Are Leading the Charge.
The future of AI infrastructure, once a background element of technology, has moved into the spotlight. In early March, discussions among financial analysts centered on this evolving sector and the key companies shaping its direction. Two distinct organizations are particularly prominent in this arena, each representing a different approach to the burgeoning field of artificial intelligence.

CoreWeave, a rapidly growing cloud company from New Jersey with origins in cryptocurrency operations, and NBIS, a more reserved European entity describing itself as an "underdog," are indicative of the significant shifts occurring in the infrastructure supporting AI's rapid expansion. One entity has secured substantial contracts and high-profile clients, while the other operates with greater flexibility and a strategy focused on diversification and specialized markets.
Together, these companies highlight the emergence of specialized "neocloud" platforms. These platforms are designed specifically for AI infrastructure, posing a challenge to established cloud providers like AWS, Azure, and Google Cloud.
A Rapidly Scaling Entity Facing Structural Questions
CoreWeave's growth trajectory has been exceptionally fast.
Did you know? CoreWeave has experienced explosive revenue growth, skyrocketing from $15.83 million in 2022 to a staggering $1.92 billion in 2024. That's a 1346% increase in just one year, followed by another 737% jump the next!
Within two years, annual revenue increased dramatically from $16 million in 2022 to almost $1.9 billion by 2024. This growth has been partly driven by significant agreements, notably an $11.9 billion deal with OpenAI finalized recently. This agreement involves OpenAI committing to purchasing computing resources through 2030 in exchange for stock in CoreWeave at IPO price, a strategic exchange of equity for infrastructure that has attracted scrutiny.
To fulfill this contract, CoreWeave has established a dedicated entity and is working quickly to finalize facility certifications. This agreement, seen by some as a landmark and by others as potentially overambitious, has solidified CoreWeave's position as a crucial infrastructure partner for OpenAI.
However, despite the impressive revenue figures, underlying vulnerabilities exist. In 2024, a significant portion of CoreWeave’s revenue, 62%, was attributed to Microsoft alone, an increase from 35% the previous year. The company’s top two clients, likely Microsoft and OpenAI, now account for a substantial 77% of its total revenue.
This degree of customer concentration is considered a potential risk factor.
It is suggested that if these key relationships were to change or weaken, CoreWeave's overall value proposition could be significantly impacted, presenting a classic scenario of over-reliance.
Financially, the company faces increasing pressures. CoreWeave reported substantial losses in both 2023 and 2024, escalating from $593.7 million to $863.4 million respectively. These losses are largely attributed to significant capital investments in new data centers, advanced GPU technology, and workforce expansion, all aimed at supporting the large-scale demands of a few major clients.
Did you know? Despite massive revenue growth, CoreWeave's net losses have also ballooned, from $31 million in 2022 to a staggering $863.4 million in 2024. These increasing losses highlight the significant investments the company is making to fuel its rapid expansion in the AI infrastructure space.
Despite these financial losses, CoreWeave is proceeding with its IPO plans, aiming to capitalize on the current interest in AI. However, skepticism persists.
It has been observed that the equity-for-access arrangement is more typical of early-stage, venture-backed companies rather than firms operating at a larger scale, implying a shift in the competitive landscape.
The Potential of a Diversified Approach
In contrast to CoreWeave, NBIS, based in a renovated warehouse in Amsterdam, presents a different model.

NBIS operates without major headline-grabbing contracts or inflated valuations, instead focusing on securing a position within the AI mid-market. They target sectors such as healthcare, logistics, and industrial automation, which require adaptable and tailored computing environments rather than simply large volumes of GPU processing power.
A key aspect of their strategy is a lower-profile approach.
The company emphasizes solving complex problems within specific industries and identifies this as a significant market opportunity, diverging from strategies focused on rapid scaling and high valuations.
Unlike CoreWeave, NBIS does not depend on a few major clients. Its customer base is more varied, potentially less glamorous but also more stable. This diversification is argued to be crucial as AI adoption becomes more nuanced and industry-specific.
Comparison: CoreWeave vs NBIS
Feature | CoreWeave | NBIS (Nebius) |
---|---|---|
Customer Base | Highly concentrated, includes Microsoft, OpenAI (implicitly), Cohere, Meta, NVIDIA, and IBM. | More diversified, includes over 40 managed clients (Q3 2024), Fortune 500 companies, and diverse Toloka clients. |
Customer Concentration | High. Microsoft accounted for 62% of 2024 revenue. Top 2 clients: 77% of revenue. Increasing concentration from 2022. | Low. No single customer dominates revenue. |
Contract Structure | 96% of revenue from multi-year, locked-in contracts (2-5 years). | Not specified, likely more varied and flexible due to niche market focus. |
Key Takeaway | High revenue visibility but significant customer concentration risk, especially reliance on Microsoft. Potential risk if AI funding cools or key customer relationships change. | Lower customer concentration reduces risk and may offer greater long-term stability and growth potential through diversification. |
However, NBIS also faces its own challenges, primarily whether the demand for bespoke infrastructure from smaller and medium-sized businesses will scale sufficiently in the near term.
It is noted that various sectors are currently in transition regarding AI adoption. Some industries are embracing AI extensively, while others are still exploring its applications.
Despite these uncertainties, early indications suggest NBIS could be well-positioned in a market beyond the dominance of hyperscale providers. Its more capital-efficient model, without heavy investment burdens or dependence on a single chip supplier, is appealing to investors who are cautious about highly leveraged infrastructure ventures.
The Rise of Specialized Cloud Infrastructure
The emergence of both CoreWeave and NBIS reflects a broader trend: the growth of neocloud infrastructure providers.
These providers differ from traditional cloud platforms by focusing on infrastructure optimized for AI workloads, emphasizing low latency, high data throughput, and close integration with advanced chip technologies.
It is observed that the focus is evolving beyond just GPUs to include broader business development, proximity to user needs, and problem-solving at the edge, areas where traditional cloud service providers may lack agility.
NVIDIA is engaging with both CoreWeave and NBIS, supplying advanced hardware and early access programs. However, the broader strategy may involve building a wide network of AI-focused cloud partners, potentially bypassing traditional hyperscale providers when beneficial.
The aim may be to support a diverse neocloud ecosystem rather than favoring any single company.
Investor Perspectives: Opportunity or Overvaluation?
For investors, the central question is not the profitability of AI infrastructure but rather market leadership and structure.
Investor Sentiment towards AI Infrastructure (Bullish vs. Cautious)
Sentiment Category | Key Points |
---|---|
Bullish | AI is widely recognized as a significant investment opportunity. |
Investors are actively seeking to integrate AI across various businesses. | |
AI and ML are considered the most transformative technologies in the short term. | |
Private markets show optimism regarding AI-related investments. | |
Cautious | Concerns exist about decreasing returns from large language models. |
There is a potential risk of overvaluation in AI stocks and assets. | |
Limitations in chip and power supply for AI growth are a concern. | |
Rapid AI development and potential for market bubbles are causing caution. | |
Key Data Points | Microsoft is planning substantial investments in AI data centers for fiscal year 2025. |
Alibaba has committed significant investment to AI and cloud computing over several years. | |
A significant portion of investors are maintaining their investment pace, while some are slightly reducing it. | |
Market indicators may suggest underlying weaknesses despite overall high valuations. |
It is suggested that if CoreWeave can diversify its customer base and achieve even modest profitability, it could reach a substantial valuation within a few years. However, shifts in demand or margin compression could destabilize this potential.
NBIS, while possibly lacking the same level of excitement, is attracting investors seeking more stable, less volatile exposure to the AI market growth.
This indicates a potential division within the industry between large-scale, capital-intensive players and niche providers focused on specialized solutions.
This divergence represents contrasting strategies of rapid scaling versus specialized expertise. Both approaches have potential, but market conditions are demanding, particularly given current valuations.
The Future Landscape
As generative AI expands across various aspects of life, from information retrieval to industry applications, competition in infrastructure will likely intensify.
The success of CoreWeave’s large contracts relative to its financial structure remains to be seen. Similarly, NBIS's ability to attract enough enterprise clients to support its specialized model is uncertain. The degree to which NVIDIA's ecosystem strategy will establish neocloud providers as permanent elements of the technology landscape is also an open question.
Ultimately, the success of these strategies will be determined over time.
For now, it is clear that the competition to power the future of AI is actively underway, and the outcomes will likely shape not only markets but also the underlying framework of artificial intelligence itself.