Oracle’s Data-Centre Problem Highlights AI’s Next Bottleneck

Oracle's New Mexico data-centre difficulties are putting electricity, financing and construction risks at the centre of the AI infrastructure boom.

The AI boom has plenty of capital. Finding enough electricity to support it is becoming harder.

Oracle has issued a force majeure notice connected to Project Jupiter, a huge New Mexico data-centre campus being developed by a unit of Blue Owl Capital, amid concerns over potential delays in securing power infrastructure.

Oracle maintains that the project remains on schedule, while Blue Owl says the notice does not change the parties' financial commitments.

Investors are still paying attention.

Project Jupiter is designed to support OpenAI and eventually handle approximately 2.45 gigawatts of capacity. Around $18 billion of loans are tied to the development, making it a major test of how the increasingly capital-intensive AI buildout will be financed.

Building AI is getting harder

The challenge facing the project goes beyond buying GPUs.

Large AI data centres require enormous amounts of electricity, transmission infrastructure and cooling, alongside regulatory approvals and years of construction.

Those physical constraints are becoming increasingly important as technology companies commit hundreds of billions of dollars to new computing capacity.

Delays matter financially too. Interest and construction costs continue accumulating even when a data centre begins generating revenue later than expected.

That is why the situation in New Mexico matters beyond Oracle.

Investors have spent much of the AI boom watching chip sales and capital-expenditure announcements. The next phase may depend increasingly on power availability, financing costs and execution.

Demand for AI computing remains enormous. Building enough infrastructure to satisfy it may prove considerably more difficult.