Subscribe to The Value Anchor Topic / The Trend Catcher Topic —unlock the full historical archive and never miss a weekly pick again.

The AI boom is entering a new phase — and this time, the biggest opportunity may no longer be AI models or software, but the infrastructure powering them.

Over the past few months, GPU cloud rental prices across North America have surged sharply. According to infrastructure data provider Ornn, hourly rental pricing for Nvidia’s latest Blackwell GPUs has climbed above $5, while SemiAnalysis estimates that one-year H100 leasing contracts rebounded nearly 40% from late-2025 lows.

The stock market has reacted quickly. Shares of AI infrastructure and Neocloud operators including CoreWeave(CRWV.US), NEBIUS(NBIS.US)s Group, and IREN Limited(IREN.US) have surged this year, while legacy cloud giant Oracle Corporation(ORCL.US) has also emerged as one of Wall Street’s biggest AI winners.

The reason is simple: AI is becoming an infrastructure scarcity story.

What Is A Neocloud?

If traditional cloud providers are “general-purpose supermarkets,” Neocloud companies are specialized AI utilities built almost entirely around GPU compute.

Instead of offering broad enterprise services, Neocloud operators focus on delivering AI infrastructure as efficiently as possible through:

• High-density GPU clusters
• Liquid-cooled AI data centers
• InfiniBand networking
• Bare-metal deployment
• AI-optimized orchestration software

Their business model is essentially “doing less, but better.” Everything is designed for one goal: maximizing AI compute performance and availability.

That specialization is becoming increasingly valuable as AI workloads grow more power-intensive.

Why The Sector Is Suddenly Exploding

The rise of Neocloud is being driven by a structural mismatch between AI demand and infrastructure supply.

Traditional Clouds Are Hitting Capacity Limits

Most legacy cloud data centers were never designed for massive AI workloads. Modern AI clusters require significantly higher power density, cooling systems, and networking performance than traditional enterprise computing.

Upgrading old facilities is expensive and slow. Building entirely new AI-ready campuses can take years.

That opened the door for Neocloud players, which built AI-native infrastructure from the ground up instead of retrofitting legacy systems.

AI Token Growth Is Fueling A Compute Flywheel

AI adoption is accelerating across chatbots, enterprise copilots, AI coding tools, video generation, and autonomous agents.

Every AI interaction consumes tokens — and every token ultimately requires GPU compute.

More importantly, inference demand is becoming persistent and recurring, creating long-term infrastructure utilization rather than temporary training spikes.

The result is a powerful flywheel:

More AI adoption → more token demand → more GPU demand → tighter infrastructure supply.

That dynamic is giving infrastructure providers unusually strong pricing power.

GPU Compute Has Become A Scarce Asset

According to SemiAnalysis, Neocloud providers are increasingly negotiating larger prepayments, longer contracts, and higher pricing with customers desperate for immediate compute access.

In previous cloud cycles, infrastructure was treated as a commodity business.

Today, elite AI compute behaves more like a strategic resource.

That shift is fundamentally changing the economics of the industry.

The Key Players Leading The Neocloud Boom

CoreWeave(CRWV.US): The Pure AI Infrastructure Trade

CoreWeave has become one of the market’s most aggressive AI infrastructure companies.

Unlike traditional cloud providers, the company focuses almost entirely on GPU rentals. Its biggest advantage is specialization — particularly its emphasis on bare-metal architecture, which allows workloads to run directly on physical hardware rather than through virtualization layers.

That improves GPU utilization and reduces latency, both critical advantages in an environment where GPU supply remains constrained.

CoreWeave has also secured massive AI infrastructure agreements involving Microsoft, OpenAI, Meta, Anthropic, and Nvidia, reinforcing its position as one of the fastest-growing AI compute providers in the market.

Perhaps most importantly, the company is viewed as one of Nvidia’s closest cloud ecosystem partners — a major advantage when GPU access itself has become a competitive moat.

NEBIUS(NBIS.US): Europe’s AI Infrastructure Challenger

Nebius, which emerged from Yandex’s international asset restructuring, is positioning itself as a full-stack AI infrastructure platform spanning Europe, the U.S., and the Middle East.

Unlike pure GPU lessors, Nebius combines cloud infrastructure with AI developer tools, managed services, and software layers designed to simplify AI deployment.

That makes the company particularly attractive to enterprises and mid-sized AI teams that lack the resources to build infrastructure internally.

Nebius may also benefit from Europe’s growing “AI sovereignty” push, as governments and corporations increasingly seek regional alternatives to dominant U.S. hyperscalers.

IREN Limited(IREN.US): From Crypto Mining To AI Infrastructure

IREN highlights another major theme in the AI infrastructure boom: power access is becoming a strategic advantage.

Originally built as a renewable-powered Bitcoin mining company, IREN already possessed many of the assets needed for AI data centers, including electricity access, cooling infrastructure, land, and grid connectivity.

That allows the company to transition into AI infrastructure far faster than traditional developers building new facilities from scratch.

As AI power consumption rises globally, access to low-cost electricity may become just as important as GPU access itself.

Oracle Corporation(ORCL.US): The Legacy Giant Quietly Winning AI

While smaller Neocloud firms attract most of the attention, Oracle has quietly become one of the largest AI infrastructure beneficiaries.

The company has aggressively expanded Oracle Cloud Infrastructure (OCI) for AI workloads while leveraging its massive enterprise customer base.

That installed customer network could become increasingly valuable as corporations move from AI experimentation toward large-scale deployment.

Unlike startups, Oracle already possesses deep enterprise relationships, global infrastructure, and the balance sheet required to compete aggressively in the AI arms race.

The Bigger Picture

The Neocloud boom reflects a broader shift happening across the AI economy.

For years, software captured most of the value in technology.

But AI is changing the equation because compute itself has become the bottleneck.

The companies controlling GPUs, power infrastructure, cooling systems, and AI-ready data centers are gaining enormous leverage as global demand for AI compute continues to accelerate.

That is why firms like CoreWeave, Nebius Group, IREN Limited, and Oracle Corporation are increasingly being viewed not simply as cloud providers, but as the foundational utilities of the AI era.