Oracle expands chip supply deal with Nvidia, adds Google and Cerebras offerings

What happened

Oracle said it is expanding its cloud AI chip deals with Nvidia to a multi-gigawatt scale using Nvidia's newest Vera Rubin AI servers. Oracle will also begin offering Google TPU chips, plus chips from SambaNova and Cerebras, and will use its own Acceleron networking chips in some Nvidia-based systems.

Confirmed

Global impact / market context

Oracle is widening its supplier list while deepening its Nvidia tie-up, so cloud customers can pick chips by workload and cost. That could lift Oracle's AI hosting revenue but also raise capital spending on data centers and reduce Nvidia's share of networking content inside those systems.

Analyst inference

Nvidia dominates AI chips, so Oracle adding Google, SambaNova and Cerebras offerings signals buyers want alternatives for inference, which means running trained AI models to answer user requests. Nvidia's own chief executive has said Vera Rubin systems can generate up to $40 billion in revenue per gigawatt, showing the scale at stake.

Analyst inference

What to watch

1. Oracle and Nvidia did not disclose the value of the expanded deal, so watch for any later figures on contract size or gigawatt capacity added. Confirmed
2. Oracle plans to offer Google TPU chips for the first time, plus SambaNova and Cerebras chips aimed at inference, so watch how quickly these options reach customers. Proposed
3. Oracle skipping some Nvidia networking chips for its own Acceleron chips could shift revenue within the deal, so watch whether Nvidia's networking content per server falls. Analyst inference

Evidence

• Original source link unavailable