The Oracle datacenter delay is becoming a major obstacle for OpenAI, as several large AI campuses originally scheduled for 2027 now appear likely to arrive in 2028. This shift highlights how shortages in skilled labor and essential materials continue to slow hyperscale infrastructure across the industry. As delays grow, the impact stretches far beyond a single project.
Why the Oracle datacenter delay emerged
Australia’s Stargate AI program, announced earlier this year by Oracle, OpenAI and SoftBank, set aggressive expansion goals. The companies planned to build massive U.S.-based facilities supporting up to 2 million AI accelerators and more than 5 GW of power. Although Oracle still intends to meet those milestones, labor scarcity and material constraints continue to interfere with construction timelines.
Bloomberg reports that Oracle cannot hire specialized workers quickly enough. In addition, the company struggles to secure crucial materials, whether for power systems, cooling hardware or core building components. As a result, Oracle pushed back several OpenAI campuses by at least one year.
How shortages magnified the Oracle datacenter delay
These new delays are not surprising, considering the scale of the Stargate initiative. Oracle must coordinate large construction crews, advanced electrical systems and high-density infrastructure — all during a global supply crunch. Therefore, bottlenecks were almost inevitable.
Even with these challenges, Oracle continues to expand. CEO Clay Magouyrk recently pointed to the company’s strong momentum:
- 147 active regions worldwide
- 64 new regions in development
- Nearly 400 MW of new data center capacity delivered in a single quarter
He also highlighted the Abilene, Texas SuperCluster, which hosts nearly 200,000 Nvidia GPUs. Oracle delivered the facility in only a few months, demonstrating that the company can still move quickly when resources are available.
Oracle ships new AI hardware despite delays
Magouyrk also confirmed progress on hardware fulfillment. Oracle has already delivered more than 96,000 Nvidia Grace Blackwell GB200 accelerators and began shipping AMD MI355 units this quarter. He stressed that Oracle accepts new orders only when it can guarantee delivery. This approach helps the company stay aligned with its margins and maintain customer trust.
What the Oracle datacenter delay means for OpenAI
Because the Oracle datacenter delay affects multiple U.S. campuses, OpenAI may need to adjust parts of its future infrastructure roadmap. Training and inference capacity remains in high demand, and shifting timelines could slow the rollout of next-generation AI systems.
Industry observers note that the delay also underscores a larger issue: even the most resource-rich cloud providers cannot escape global labor and supply shortages. Consequently, the entire AI infrastructure sector may experience longer build cycles than expected.
The road ahead
Although the datacenter delay pushes several projects into 2028, Oracle says the overall expansion plan remains intact. The company expects demand for AI compute to keep rising, and it continues to build capacity aggressively. Whether the revised schedule holds will depend on how quickly the labor market and supply chain stabilize.
For now, one thing is clear: the Oracle datacenter delay highlights the growing tension between AI acceleration and the physical limits of global infrastructure.
Read also
Join the discussion in our Facebook community.