US AI Data Center "Narrative Shift": From the Excitement of "Big Deals" to the "Blame Game" of Ongoing Delays

Market sentiment in the US artificial intelligence data center sector is undergoing a dramatic reversal. As unprecedented server cluster construction encounters real-world obstacles, the previous excitement fueled by gigawatt-level (GW) large-scale transactions and record contracts is fading, replaced by frequent project delays and ensuing blame-shifting.

In this race for computing power, the most notable development is that supply chain constraints have begun to materially impact business performance. AI cloud service provider CoreWeave has warned investors that due to delays from third-party developers, this quarter's revenue could face a writedown of $100 million to $200 million. This signal indicates that the complexity of data center construction is dragging multi-billion dollar projects into the quagmire of delay, causing involved parties to start blaming each other and seeking to clarify responsibility as schedules slip.

This friction is not limited to individual companies but also exposes systemic bottlenecks faced by the entire industry. Currently, because GPU shipments have outpaced facility construction speed, some companies are being forced to keep expensive hardware idle in warehouses, waiting for deployment. At the same time, physical limitations ranging from power access to equipment delivery are increasingly at odds with clients' aggressive demands.

As main buyers like the "Fab Five" (OpenAI, Google, Meta, Anthropic, and xAI) intensify the urgency of computing power delivery, infrastructure-level delays are reshaping market expectations. Investors and industry participants are gradually realizing that AI infrastructure development is transitioning from an early phase of capital exuberance to a turbulent period filled with execution risks and blame games.

Revenue Warning Sparks Dispute over Responsibility

CoreWeave, AI cloud service provider to giants like Microsoft and OpenAI, has become a classic example in this blame game. CoreWeave CEO Mike Intrator admitted earlier this month that temporary delays from "slow progress by third-party data center developers" would significantly impact the company’s current quarter revenue. Although Intrator didn't name anyone directly, industry speculation points to major partner Core Scientific as the developer in question.

Such speculation is well-founded. According to sources, as early as eight months ago Microsoft reduced some contracts with CoreWeave due to delays at a data center in Denton, Texas, for which Core Scientific was responsible for power supply.

In February this year, Core Scientific stated in its earnings call that the completion date for a certain data center project would be pushed from 2025 to early 2026, likely referring to the aforementioned facility. Even though OpenAI stepped in later and signed a $12 billion contract with CoreWeave to lease servers at this facility, the tension from delays remains unresolved.

Regarding the accusations of delay, Core Scientific CEO Adam Sullivan did not directly respond to any specific project, but instead gave a sharp critique of the industry’s situation. He pointed out that many AI data center construction schedules are "unrealistic," unless developers have already secured long lead-time equipment like generators and skilled contractors in advance. Sullivan told the media that when a listed company pre-announces a delay and another party waits until the last moment to disclose, it creates confusion and erodes market confidence. In addition, Core Scientific shareholders earlier voted down a $9 billion acquisition offer from CoreWeave, which may be another reason for tensions between the two sides.

High-Stakes Game under Thin Profits

While project delays are common in the construction industry, under the current AI computing power race, the stakes are very different. To meet delivery pressure from clients like OpenAI, Oracle executives earlier this year voiced strong dissatisfaction with contractors at a construction site in Abilene, Texas.

This anxiety stems from stringent financial terms. It is understood that contracts between cloud service providers and clients usually include punitive clauses: if providers miss deadlines or servers have downtime due to malfunctions, clients can reduce payments. For GPU cloud rental businesses with already thin profit margins, these operational issues can have a substantial negative impact on financial results.

This explains why, when multi-billion dollar projects miss deadlines by even a few weeks or months, all parties are eager to identify responsibility. For companies that have promised rapid delivery schedules, the race to bring Nvidia GPU clusters online remains a severe challenge.

Hardware Backlog and Strategic Adjustment

As power supply becomes increasingly scarce, lagging construction progress has led to supply-demand mismatches between hardware and facilities. Several developers have revealed that current GPU shipment speeds far exceed data center construction speeds, forcing some companies to store rows of idle GPUs in warehouses, waiting for orders to determine their destination.

Facing this reality, major tech firms are adjusting strategies to create buffers. Meta CFO Susan Li acknowledged this tension during the company's earnings call in late October, stating that Meta is now "phasing in data center builds," meaning all facilities except GPU racks are being prepared ahead of time so capacity can be quickly ramped up when needed in the future.

The fact that even a major developer like Meta must establish buffer mechanisms highlights the industry’s current predicament: the physical limits of labor, equipment, utilities, and contractor bandwidth are colliding head-on with clients’ endless demands. As power supply becomes harder to secure, more clients may opt to work with multiple data center providers to hedge their risks in the future.

Risk Warning and DisclaimerThe market involves risk, and investing requires caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situations, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article fit their particular circumstances. Investing based on this information is at your own risk.