SI Agent News
ConfirmedClaudeOpenAIOpen modelsChips

Nvidia chief highlights capital shortages and mobile AI racks in Berlin

Jensen Huang addressed the growing scarcity of working capital for AI infrastructure and unveiled new portable data center solutions during Nvidia's GTC event.

$7.6 trillionestimated AI capital spending 2026-2031

The short version

Why it matters

As AI demand outpaces the ability to build massive power plants and data centers, the industry is shifting toward smaller, more mobile hardware. This transition reflects a move toward efficiency to overcome the looming bottleneck of high construction costs and limited capital.

Portable AI infrastructure

During the GTC conference in Berlin, Nvidia CEO Jensen Huang introduced a new hardware solution designed to accelerate the deployment of AI capabilities. The product packs seven of the company's Vera Rubin AI racks into a single shipping container that is built to be transported on standard roads.

This move is intended to help companies get AI compute online much faster than traditional construction methods allow. Huang noted that businesses can lose billions of dollars in revenue if there are significant delays in getting their data centers operational.

By integrating these portable units into the supply chain, Nvidia aims to increase the speed at which customers can activate their systems. This approach addresses the immediate need for high-performance computing while bypassing some of the logistical hurdles of permanent site construction.

The capital bottleneck

Despite the rapid growth of the industry, Huang warned that the availability of capital is becoming a significant hurdle for the future of AI. He stated that the demand for capital is currently much greater than what is available to fuel the growth of various AI sectors.

The scale of the required investment is massive, with Goldman Sachs estimating that AI-related spending on compute, data centers, and power could hit $7.6 trillion by 2031. Huang specifically pointed out that a single 1 gigawatt data center can cost around $50 billion to build.

While Huang expressed confidence that these issues will eventually be resolved, industry insiders suggest that capital will soon become the primary bottleneck. This scarcity of working capital could slow down the expansion of infrastructure even as the demand for AI services remains high.

Neocloud market distress

The conference took place during a period of volatility for the 'neocloud' sector, which consists of companies that lease out AI compute power. Several public neocloud stocks saw significant selloffs during the final day of the event, with CoreWeave dropping 10% and Nebius falling 7%.

Investors are increasingly concerned about the heavy borrowing required to sustain these infrastructure businesses. One executive at an AI infrastructure company told reporters that the sheer magnitude of businesses operating without a credit rating is causing market anxiety.

Huang remained optimistic about the sector's fundamentals, telling CNBC that the demand for neocloud services is currently off the charts. However, the market remains wary of the financial stability of these providers as they attempt to scale up to meet global demand.

Strategic shifts in deployment

The move toward smaller and more modular data centers is not just a Nvidia initiative but a broader trend among major AI consumers. Reports indicate that prominent companies like OpenAI and Anthropic have been actively seeking out smaller data center deals to expand their reach.

These smaller deployments allow for more flexible scaling and can be deployed more rapidly than massive, centralized hubs. This shift helps companies manage the high costs of infrastructure while still providing the necessary power for advanced models.

As the industry navigates these logistical and financial challenges, the focus is shifting toward how to maximize existing resources. The goal is to maintain the momentum of AI development while overcoming the physical and economic limits of current infrastructure.

Global leaders and tech news

The report originated from this week's edition of The Tech Download newsletter. During the event in Berlin, Jensen Huang engaged with several world leaders, including French President Emmanuel Macron, German Chancellor Friedrich Merz, and U.K. Prime Minister Andy Burnham. The tour also featured unique moments on stage, such as Huang spending time interacting with small robot ducks.

Beyond the hardware announcements, other significant developments in the tech landscape were noted. Elon Musk shared that Tesla and SpaceX have established special committees to explore a potential merger. Additionally, the Trump administration recently prohibited the Chinese AI startup DeepSeek from being used by civilian U.S. government agencies due to security concerns.

The tech landscape is also seeing a shift in satellite internet demand. Following issues faced by Ukraine with SpaceX's Starlink and Starshield systems, there is an accelerating interest in alternative satellite services. However, current options remain limited as few alternatives are currently comparable to the existing systems provided by SpaceX.

What to watch

Try it

Look up the current stock performance of 'neocloud' providers like CoreWeave or Nebius to see how the market is reacting to infrastructure funding concerns.

Sources and links

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

Listen to the AI Wire headlines

Listen to the AI Wire

Listen to our daily AI shows every morning: on your commute, in the shower, wherever you are

Listen to the daily shows

First reported by CNBC · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.