Nvidia Earnings Spotlight AI Data Center Demand Shift

Nvidia’s latest earnings underscored just how aggressively the AI infrastructure market continues to expand, reports Bisnow. The chipmaker reported a record $82B in quarterly revenue on May 21, up 85% year over year, while projecting another jump to $92B next quarter. Nearly all of that growth came from AI-related data center demand, highlighting how the sector has become the backbone of the broader AI economy.

CEO Jensen Huang described demand as having “gone parabolic” during the company’s earnings call, reflecting the pace at which cloud providers, enterprises, and AI startups are deploying new computing infrastructure.

AI Data Center Demand Reaches New Scale

Nvidia’s data center segment generated $75B in quarterly revenue. That marked a 92% increase from a year earlier. Executives said more than 80 data centers now deploy over 10 megawatts of Nvidia hardware. That figure doubled over the past year.

The growth reflects a broader AI infrastructure spending wave. Amazon, Microsoft, Google, Meta, and Oracle plan to spend more than $700B in 2026. Much of that spending targets AI data centers and accelerated computing systems.

Nvidia CFO Colette Kress cited forecasts showing hyperscaler capex could surpass $1T in 2027. Forecasts also project spending could hit $3T to $4T by 2030. Kress said companies across industries continue adopting AI tools at a rapid pace. She added that demand now spans chips, infrastructure, energy, and applications.

The Rise of Neocloud Providers

One of the more important shifts in Nvidia’s earnings was the growing role of so-called neocloud operators. While hyperscalers still accounted for roughly half of Nvidia’s data center revenue, the remainder increasingly came from AI-focused cloud firms, enterprises, and industrial users.

Nvidia said revenue from neocloud customers tripled year over year. The category includes operators like CoreWeave, Lambda, Crusoe, Nebius, Core Scientific, and Nscale, many of which are rapidly leasing and developing data center capacity to support AI workloads.

Unlike hyperscalers that often design proprietary AI systems internally, neoclouds typically rely on turnkey infrastructure that can be deployed quickly. Nvidia executives argued that demand favors integrated systems combining chips, networking, and software into a single package.

The trend also reflects growing demand for inference computing. These systems power real-time AI interactions after model training. Those workloads often sit closer to population centers. As a result, smaller regional AI deployments continue gaining traction. That demand has also boosted investor interest in data center REITs tied to inference-heavy workloads and edge deployments.

Inference Computing Reshapes the Market

Nvidia executives called the quarter an “inflection” point for inference demand. Enterprise and neocloud customers drove a larger share of revenue. Those customers increasingly rely on real-time AI applications instead of model training workloads.

That shift carries major implications for commercial real estate. Inference computing often requires a different data center footprint than hyperscale AI clusters. Smaller deployments near end users can increase demand for edge facilities and secondary-market capacity near large metros.

The trend could also expand the tenant pool competing for powered shell inventory and utility access. For years, developers focused mainly on hyperscalers. Now, AI-native operators want fast deployments and flexible infrastructure options.

Why It Matters

Nvidia’s earnings reinforced that AI infrastructure demand is no longer concentrated among a handful of Big Tech companies. The rapid expansion of neocloud providers signals a more diversified tenant ecosystem for data center owners, developers, and capital providers.

That diversification could reshape investment strategies across the sector. According to CBRE’s 2026 North America Data Center Trends report, vacancy rates in primary markets remain near historic lows as AI deployments tighten power availability and accelerate preleasing activity.

The growth in inference computing may also increase pressure on urban power grids and push more AI infrastructure into edge and regional markets where latency matters.

What’s Next

Investors and developers will be watching whether neocloud growth remains sustainable as hyperscalers continue building proprietary AI infrastructure and custom chips. Competition from Amazon, Google, and other semiconductor players remains a key risk for Nvidia.

Still, Nvidia executives signaled confidence that demand for AI computing systems will continue climbing over the next several years. If capital expenditure forecasts hold, the next phase of the AI boom could extend well beyond hyperscale campuses and reshape data center development patterns across both primary and secondary markets.