Oracle Adds Cerebras, Google TPUs to Cloud Lineup; Nvidia Deal Goes Multi-Gigawatt
TREE NEWS reports: Oracle announced Tuesday at its annual customer conference that it will integrate Cerebras Systems’ AI chips into its cloud infrastructure, marking a major win for one of Nvidia’s most credible challengers in the AI inference market. The Cerebras systems are expected to be deployed on Oracle Cloud Infrastructure in the first half of 2027. Oracle also unveiled a slew of other chip-supplier agreements, including a deal with Nvidia to scale deployments to “multi-gigawatt” levels using Nvidia’s latest Vera Rubin AI servers, and its first-ever adoption of Google’s TPU chips. Shares of Cerebras jumped more than 10% intraday on the news, though the stock remains down 41% since its May IPO.
Market Implications
The announcement underscores Oracle’s aggressive push to become a top-tier AI infrastructure provider and highlights the accelerating diversification of the AI chip supply chain. For Nvidia, the multi-gigawatt expansion reinforces its dominant position in AI training, but Oracle’s decision to use its own Acceleron networking chips in Nvidia-based systems signals a subtle shift toward infrastructure independence. The inclusion of Google TPUs and Cerebras chips suggests cloud providers are increasingly optimizing for inference workloads, where cost and latency matter more than raw training throughput. This could pressure Nvidia’s pricing power in the inference segment over time, though its scale—nearly $100 billion in quarterly AI chip sales—remains unchallenged.
For Cerebras, the Oracle deal is a significant validation, adding to existing customers like OpenAI and AWS. Analysts expect Cerebras revenue to triple to about $3 billion next year, but the company still faces steep competition and investor skepticism. The news may lift sentiment for other AI chip startups and semiconductor suppliers, while also benefiting Oracle’s stock by reinforcing its cloud growth narrative. Bond investors will watch Oracle’s capital expenditure plans closely; the company maintained its fiscal 2027 capex target of $90–95 billion, with details expected at its investor day this week.
Key Takeaways
- Oracle’s multi-vendor AI chip strategy reduces reliance on Nvidia and could reshape cloud infrastructure competition.
- Cerebras gains a marquee customer but must prove it can scale and win share in the inference market.
- Nvidia’s dominance remains intact, but the multi-gigawatt deal and networking substitution bear watching.
- Investors should monitor Oracle’s capex funding and customer adoption trends for AI infrastructure.
