Intel Bets Chips on Network Platforms Consolidation
Chipmaker Intel is anticipating a round of consolidation in the "fragmented" market for network platforms, where it competes against a host of companies including Broadcom and Nvidia.
While the semiconductor giant is dominant in the data center business, with a market share of about 97%, it is one of numerous players vying for a bigger piece of the pie in networks.
Sandra Rivera, the general manager of Intel's network platforms group, says fragmentation makes the market ripe for consolidation, raising the prospect that some companies will be swallowed up during takeover activity.
It is unclear whether Intel Corp. (Nasdaq: INTC) could play an active role in this consolidation, but the company is certainly looking to its network platforms business for a sales boost.
Brian Krzanich, Intel's CEO, told analysts at the start of the year that he was counting on networking and storage, "where we have very low market share today," for growth at its data center business over the rest of the decade.
Rivera's network platforms business currently forms a small but rapidly growing part of the data center group, which -- in turn -- has been a star performer for Intel in recent quarters.
Sales at that division rose by nearly 8% in 2016, to $17.2 billion, accounting for about 29% of Intel's total. By contrast, revenues from the client computing group, Intel's biggest, were up just 2%, to $32.9 billion.
Unfortunately, Intel does not break out details of revenues from network platforms and Rivera is coy when pressed for a figure.
But during a previous conversation with Light Reading she revealed that her bit of Intel's business made between $1 billion and $1.8 billion in 2015. (See A Few Minutes With Intel's Sandra Rivera.)
Light Reading understands that revenues grew by roughly 40% last year, which would put them between $1.4 billion and $2.52 billion, and that Intel is aiming for sales growth of about 10% in 2017.
In this particular market, broader developments appear to have strengthened Intel. For one thing, the transition from purpose-built "black boxes" to cloud technologies is spurring the replacement of more specialized application-specific integrated circuits (ASICs) and digital signal processors (DSPs) with the general-purpose central processing units (CPUs) off which Intel has grown fat.
Intel will also welcome growing telco interest in "edge computing" with the forthcoming transition to 5G. If they are to support more advanced applications, operators may need to move computing resources out of centralized facilities and much closer to end users (at the network's "edge") to minimize "latency" -- the delay that occurs when signals travel over data networks. Any such investments in edge-computing facilities and new processing power should clearly benefit Intel. (See Intel Eyes Fab 5G Future, Not Another Mobile Mess.)
While Intel appears to be on the attack in networks, its strategy for the rest of its data center group is basically a defensive one. Yet the challenges that Intel faces in both areas are related.
ARM Holdings plc (Nasdaq: ARMHY; London: ARM) has been eyeing a much bigger role in the server market, partly through tie-ups with the makers of white boxes, and it is attracting interest from both telcos and web-scale giants. (See Qualcomm Challenges Intel With New ARM Chip.)
Last year, for instance, France's Orange (NYSE: FTE) flagged white box experiments in which it replaced x86 chips, based on Intel architecture, with ARM-based smartphone processors.
Far more troubling for Intel was the recent news that Microsoft Corp. (Nasdaq: MSFT) is planning to use processors from ARM licensee Qualcomm in its cloud servers. (See Microsoft, Qualcomm Bring ARM to Azure.)
With developments in artificial intelligence (AI), Nvidia Corp. (Nasdaq: NVDA) has emerged as another worry for Intel.
A maker of so-called graphics processing units (GPUs), Nvidia is seen as a potential threat to Intel's data center business in some corners of the industry. The basic rationale is that Nvidia's more specialized GPUs are better than Intel's general-purpose CPUs at supporting virtual reality, AI and machine learning applications.
The likes of Google and Microsoft have recently been drawn to these more specialized chips, and even investing in their own processor designs. (See Hypercloud Guys Make Big AI Push.)
— Iain Morris, , News Editor, Light Reading