Intel is leaving the Ethernet switching industry just a few years after a splashy acquisition.
At one point, Intel (INTC -6.41%) was determined to expand its reach in the data center. Despite rival AMD’s recent product lineup, the company has long been the market leader in server CPUs. However, Intel saw this as an opportunity because a data center contains a lot more than just CPUs.
In an effort to expand its presence in the networking sector of the data center, Intel purchased Barefoot Networks in 2019. Software and Ethernet switch chips were created by Barefoot Networks with a focus on programmability and adaptability. “The addition of Barefoot Networks will support our focus on end-to-end cloud networking and infrastructure leadership,” the deal was announced in an Intel news release.
The Network and Edge Group (NEX) of Intel eventually housed the Ethernet switch chip efforts of the company, which brought in $8.9 billion in revenue in 2022. Intel still considers this segment to be an essential component of its transformation efforts because it encompasses a wide range of products.
Getting out of Ethernet switches Despite the fact that NEX will continue to be a significant part of Intel,
the company said in its fourth-quarter report that it would stop investing in its network switching product line in the future. As a result, Barefoot Networks-acquired chips in the Tofino series are no longer being developed. Intel will continue to provide customer and product support.
Since CEO Pat Gelsinger took over at the beginning of 2021, Intel has closed seven businesses. The switching business is the seventh. These exits have resulted in savings of more than $1.5 billion, which will be crucial as Intel navigates an unprecedented PC market downturn.
The memory industry as a whole is another notable exit. In a complicated deal that was announced in 2020, Intel sold off its flash memory and solid-state drive businesses to SK Hynix. Intel has already transferred its SSD business and NAND memory manufacturing facility in China, despite the fact that that transaction will not be completed completely until 2025. It seems like a good idea to exit memory. The memory chip industry is experiencing an even more significant decline than the demand for CPUs for PCs and servers.
Optane, the company’s brand name for products based on 3D XPoint,
was the other part of Intel’s memory business. Intel and Micron collaborated on the development of a novel kind of memory called 3D XPoint, and it looked very promising. It was faster than NAND, was nonvolatile, and could be used in superfast SSDs or even replace DRAM in some cases. However, the technology never achieved sufficient traction. Last year, Intel wiped out the business by selling off $559 million worth of inventory.
Under Gelsinger, Intel is focusing on the best opportunities in areas where it already has or could potentially gain a long-term competitive advantage. CPUs found in personal computers and servers meet the requirements. The company’s attempts to enter the discrete graphics card market with its Arc graphics cards are also relevant. Additionally, Intel’s move into the foundry industry makes sense due to the fact that few businesses are able to produce chips using cutting-edge technology due to high capital and technical know-how requirements.
Memory was never very helpful. Memory chips are generally considered to be commonplace goods whose prices are determined by supply and demand. In theory, 3D XPoint made sense, but in practice, it just didn’t work as a business. In 2019, it seemed like a good idea to enter the Ethernet switching chip market, but Intel has decided that investing in that market is not worth it.
By the end of 2025, Intel expects to have cut costs by as much as $10 billion, with a goal of cutting costs by $3 billion this year. One way to cut costs is to exit non-core businesses. Intel’s aggressive efforts to control its costs appear to have cost the Ethernet switch chip business a victim.