In a significant strategic pivot, Oracle Corporation has officially withdrawn from the competitive arena of designing its own AI and server chips. The move, announced by Chairman Larry Ellison, comes as the company capitalizes on a major financial windfall from the sale of its stake in chip designer Ampere Computing, signaling a new direction focused on flexibility and vendor-agnostic infrastructure.
Oracle Capitalizes on Ampere Sale, Nets USD 2.7 Billion
Oracle has finalized its exit from direct investment in server chip design by selling its approximately 29% stake in Ampere Computing. The buyer was Japanese investment giant SoftBank Group, which completed a USD 65 billion all-cash acquisition of Ampere last month. From this transaction, Oracle realized a substantial pre-tax profit of roughly USD 2.7 billion (approximately CNY 190.73 billion). This sale not only provides Oracle with significant capital but also marks the definitive end of its venture into owning and developing proprietary processor architectures for its cloud data centers, a path it had explored through its minority investment in Ampere.
Key Financial Transaction: Ampere Stake Sale
- Seller: Oracle Corporation
- Buyer: SoftBank Group
- Asset Sold: ~29% stake in Ampere Computing
- Deal Context: Part of SoftBank's USD 65B all-cash acquisition of Ampere
- Oracle's Gain: ~USD 2.7B (CNY 190.73B) pre-tax profit
Ellison Outlines "Chip Neutrality" as New Core Strategy
The rationale behind this decisive shift was clearly articulated by Oracle Chairman Larry Ellison during a recent earnings call. He stated that the company no longer sees strategic value in designing, manufacturing, and using its own chips within its cloud infrastructure. Instead, Oracle is committing to a policy of "Chip Neutrality." Ellison emphasized that the rapid and unpredictable evolution of AI technology necessitates maximum agility. To remain competitive, Oracle must retain the ability to deploy whatever silicon its customers demand, whether from Nvidia, AMD, or future market entrants, rather than being locked into its own in-house architecture.
A Divergent Path from Hyperscaler Competitors
This strategy places Oracle on a distinctly different trajectory compared to its major rivals in the cloud sector. Hyperscale providers like Microsoft Azure, Amazon Web Services (AWS), and Google Cloud Platform (GCP) are heavily investing in internal teams to develop custom processors tailored to their specific workloads. The goal for these competitors is to reduce computing costs and optimize performance for demanding AI applications. Oracle's approach, by contrast, transforms it into a pure-play integrator, leveraging best-in-class hardware from multiple suppliers to build its cloud offerings.
Strategic Comparison: Oracle vs. Hyperscalers
| Aspect | Oracle's New Strategy | Typical Hyperscaler (AWS, Google, Microsoft) Strategy |
|---|---|---|
| Chip Development | "Chip Neutrality" – No in-house design. | Heavy investment in custom, in-house processor design (e.g., AWS Graviton, Google TPU). |
| Primary Goal | Flexibility and vendor choice to adapt to rapid AI changes. | Cost reduction and performance optimization for specific workloads. |
| Business Model | Integrator of third-party best-in-class hardware. | Vertical integration, controlling the full stack from silicon to service. |
Building a Multi-Vendor Hardware Ecosystem
With the "Chip Neutrality" doctrine established, Oracle is actively diversifying its hardware partnerships. While it continues to be a large-scale purchaser of Nvidia GPUs, it is expanding its portfolio. A key example is a recently announced agreement with AMD. Oracle plans to launch its first public AI supercluster powered by AMD's upcoming Instinct MI450 GPUs, with an initial scale of 50,000 units targeted for the third quarter of 2026. This move underscores Oracle's intent to construct a more resilient and cost-effective cloud infrastructure by fostering competition and choice among top-tier chip vendors, betting that flexibility will be a greater asset than proprietary control in the fast-moving AI era.