One thought on “Intel to spin off its Networking & Edge Group into a stand-alone business”

1. “We plan to establish key elements of our networking and communications business as a standalone company and we have begun the process of identifying strategic investors,” said an Intel spokesperson in a statement emailed to Light Reading. “Like Altera, we will remain an anchor investor enabling us to benefit from future upside as we position the business for future growth.”

   Rarely has such a big company shrunk its workforce so rapidly. Intel employed nearly 125,000 people just 18 months ago, according to the full-year report for 2023 it lodged with the US Securities and Exchange Commission (SEC). A year later, the number had plummeted to less than 109,000. But the drop in 2025 is expected to be much, much steeper. By the end of this year, Intel CEO Lip-Bu Tan expects to employ no more than 75,000 people.

   Most of the cuts have already happened, Tan told analysts on Intel’s call this week about second-quarter results. So-called “management layers” have been slashed by 50%. But massive cuts will make it harder for Intel to innovate. Having pumped $23.9 billion into capital expenditure last year, it is guiding for an outlay of only $18 billion in 2025. On the call with Tan, one analyst voiced concern about Intel’s ability to “fix” x86, the architecture at the heart of its chip technology.

   What Intel calls 18A, its process for building 1.8-nanometer transistors into its silicon, has already suffered delays and Intel will not reach “peak volumes” until the early 2030s, said David Zinsner, Intel’s chief financial officer. “I would say I wouldn’t write off 18A as potentially getting external customers at some point,” he added.

   External customers, in this context, likely means other chipmakers using Intel’s foundries to produce 18A chips. But in the telecom market, Ericsson had flagged a deal with Intel exactly two years ago that involved future reliance on 18A. “These technologies will help Intel reclaim leadership position by 2025 and enhance the future offerings for their customers,” said the Swedish network equipment vendor in a statement at the time.

   That has clearly not happened, and concern has grown about Intel’s fate and what it would mean for telecom vendors such as Ericsson. “The problem is that Intel really only has two choices, which are to invest like crazy and compete on technology or to split itself up into pieces and sell them off to the highest bidder,” said Richard Windsor, an analyst with Radio Free Mobile, in a blog written before news of the NEX divestment appeared. “Mr Tan appears to have chosen neither and is trying to navigate a course somewhere in the middle, which I think will end in the slow, painful decline of an industry titan.”

   This would obviously have ramifications that go far beyond networks. Yet Intel’s commitment to that area of its business was slipping before Windsor penned his latest words. Last year, Intel had three big overarching product groups – CCG (client computing group); DCAI (data center and AI); and NEX (network and edge). The last in that list was the smallest, generating only 12% of Intel’s $49 billion in product revenues, and it covered Intel’s activities in telecom through a sub-unit called CSG (communications solutions group). But NEX had vanished by the time Intel submitted its first-quarter report for 2025.

   Ericsson looks just as dependent on Intel as it ever did. The Swedish company claims to design its own Layer 1 application-specific integrated circuits (ASICs), relying on Intel there purely for manufacturing expertise (and, ultimately, 18A). But it previously acknowledged its use of Snow Ridge in RAN compute for tasks including “the L3 control plane, user plane, the OAM [operations, administration and maintenance] handling, and the IP [Internet Protocol] interface (backhaul interface).”

   Could Ericsson, then, buy NEX to protect an essential source of components? Acquiring network assets from Intel has always looked like it would be complicated because of the hard-to-sever umbilical link to the x86 technology at Intel’s core. The other issue is the conflict of interest a takeover would create by putting Ericsson in charge of chip technologies used by its rivals. Samsung, the world’s fifth-biggest RAN vendor, has built its entire virtual RAN strategy on Intel’s chips. Numerous smaller players are in a similar position.

   Ericsson will be cognizant of the risks if NEX is acquired by someone else. Ericsson has long had a close partnership with Intel, but another chipmaker might have different ideas about future technology development. It might also try to raise prices. Broadcom, active in other parts of the RAN, became notorious for that after its takeover of VMware, a cloud-computing developer. A preferable outcome for Ericsson, conceivably, would be a repeat of what happened with Altera – the sale of a majority stake in NEX to a private equity firm, with Intel retaining a significant share and technology role.

   Before Intel confirmed the NEX spinoff, Ericsson had pushed the idea that full virtualization of its RAN compute would give it independence from any one silicon platform. But there are gaps that are hard to fill. Intel, for instance, contributes a hardware accelerator for a very resource-hungry L1 task called forward error correction.

   https://www.lightreading.com/5g/ailing-intel-to-sell-network-biz-in-frightener-for-ericsson

   Reply