Cerebras surges over 10%! Oracle announces first integration with Cerebras and Google, expands Nvidia partnership to "multi-gigawatt level"

Oracle announces the introduction of Cerebras Systems' AI chips into its cloud services, marking another major client for this emerging challenger to Nvidia and accelerating the evolution of the competitive landscape in the AI inference chip market.

Clay Magouyrk, Oracle's co-CEO, announced this partnership at the company's annual customer conference on Tuesday, stating that Cerebras systems are expected to be officially deployed on Oracle Cloud Infrastructure in the first half of 2027.

At the same time, Oracle also announced a series of intensive partnership agreements with chip suppliers at the conference, including not only Cerebras, but also Nvidia, Google, and SambaNova Systems supported by Intel.

Following the announcement, Cerebras' stock price surged by over 10% intraday. However, since the company's IPO in May this year, its cumulative share price decline remains as high as 41% as of Monday's close.

This collaboration is the latest move in Oracle's large-scale data center expansion strategy. At the conference, Oracle announced a series of intensive partnership agreements with chip suppliers, covering not only Cerebras but also Nvidia, Google, and Intel-backed SambaNova Systems.

In the partnership with Nvidia, Oracle stated that it will scale deployments to the "multi-gigawatt level," adopting Nvidia's latest generation "Vera Rubin" AI servers.

Nvidia CEO Jensen Huang previously stated that the Vera Rubin system could generate up to $40 billion in revenue per gigawatt for Nvidia, but neither company disclosed the specific amount involved in this agreement.

It is worth noting that even in systems deploying Nvidia hardware, Oracle has chosen to replace some of Nvidia’s network products with its self-developed "Acceleron" networking chips, further strengthening its self-sufficiency in infrastructure.

Additionally, Oracle announced its first introduction of Google's TPU (Tensor Processing Unit) chips.

Mahesh Thiagarajan, President of Oracle Cloud Infrastructure, stated: "Different AI workloads have different computational requirements. Diverse options give customers flexibility, enabling them to choose the most suitable infrastructure for each workload based on performance, throughput, and cost efficiency."

The entry of Oracle is of great significance for Cerebras. Cerebras’ core technology uses an ultra-large chip that occupies an entire silicon wafer, rather than the conventional method of slicing wafers into multiple processors. The company claims this architecture can significantly reduce response latency for AI inference requests.

"This is a major expansion of our tier-one customer base," said Cerebras CEO Andrew Feldman. "We are being deployed at large scale in other companies’ data centers."

The company has already signed well-known clients such as OpenAI and Amazon Web Services. According to the average analyst forecasts compiled by Bloomberg, Cerebras’ revenue is expected to reach about $3 billion next year—a roughly threefold increase from this year—while full-year revenue for this year is expected to remain below $1 billion.

Nevertheless, investor concerns persist. Nvidia currently approaches $100 billion in AI chip sales per quarter, giving it overwhelming market dominance over Cerebras.

With Nvidia dominating the market for training chips, both Cerebras and SambaNova are targeting the more fragmented inference market, but whether they can gain a firm foothold remains a key issue closely watched by investors.

On capital expenditures, Oracle has yet to disclose the scale of Cerebras product purchases planned or the specific financing arrangements for chip acquisitions.

At last month's earnings call, Oracle noted that customers are increasingly bearing the costs of chip purchases themselves and maintained its capital expenditure target of approximately $90-95 billion for the 2027 fiscal year.

Previously, Oracle stated it would invest $70 billion in capital spending in the fiscal year ending May 2027, aiming to become a leading supplier of top-tier AI computing power.

Oracle also plans to host Investor Day events this week, where it may disclose more details regarding its data center expansion strategy and capital allocation plans.