Oracle Chairman Larry Ellison said on Wednesday that the company sold its stake in chip design firm Ampere Computing "because we no longer see strategic importance in continuing to design, manufacture, and use our own chips in our cloud data centers."

The Acquisition about Ampere

Japanese investment giant SoftBank acquired Ampere last month in a $6.5 billion all-cash deal to bolster its artificial intelligence capabilities as the investor seeks major AI infrastructure projects, such as the Stargate project in the United States.

Oracle said it gained $2.7 billion pre-tax from the sale of its stake in Ampere—which, according to previous regulatory filings, will give it a 29% stake by May 2024.

Oracle's approach to chip design differs from hyperscale competitors Microsoft, Amazon Web Services, and Google. It holds a minority stake in Ampere, which, in addition to manufacturing processors for Oracle, produces Arm-compatible CPUs for other companies, rather than employing its own teams to develop processors specifically for internal use.

However, Oracle's sale of its Ampere architecture stake contrasts sharply with the practices of rival hyperscale data center operators, who have been developing their own chips to meet the high demands of AI and reduce computing costs.

While Oracle is a major buyer of Ampere CPUs, the chip design company has been trying to build its sales business through channel partners for the past few years.

In a May interview with CRN, Ampere's Chief Product Officer, Jeff Wittich, said the company wanted to grow its channel business "very quickly," but acknowledged that it didn't represent a large percentage of sales at the time.

The executive was then discussing the new Ampere Systems Builder program, designed to accelerate server platform development by bringing together multiple IT infrastructure vendors, thereby creating greater flexibility and reducing costs for AI and cloud computing infrastructure.

"Our next step is to generate significant interest and engagement from these channel partners and start incorporating them into the program," Wittich said.

The New Ampere System Builders Program

The executive was discussing a new Ampere System Builder program aimed at accelerating server platform development by bringing together multiple IT infrastructure vendors, thereby creating greater flexibility and reducing costs for AI and cloud computing infrastructure.

"We hope to take the next step by generating broad interest and engagement from these channel partners and starting to incorporate them into the program," Vititch said.

Ellison explained the rationale for selling the Ampere architecture stake in Oracle's second-quarter earnings report. He stated that the move resulted in the company "now committed to a chip-neutral policy, working closely with all CPU and GPU vendors."

"Of course, we will continue to source the latest GPUs from Nvidia, but we need to be prepared to deploy any chips our customers want to buy," Oracle CTO Ellison said in a statement. "AI technology will change a lot in the coming years, and we must remain agile to adapt to those changes."

While Oracle has been a significant buyer of Nvidia GPUs and related components, the company announced an agreement in October with AMD to launch the "first publicly accessible AI supercluster" powered by AMD Instinct MI450 GPUs. The cluster is expected to be deployed for the first time in the third quarter of next year, initially using 50,000 GPUs.

The competitive landscape of the semiconductor industry is shifting in other areas. Qualcomm plans to re-enter the server CPU market and launch two AI acceleration chips within the next two years. Google may also shake up the market, as it is reportedly exploring deploying its TPUs (traditionally used for its own infrastructure) in customers' data centers, including Oracle's major customer, Meta.

Source: Compiled from CRN

View more at EASELINK