As the world's largest semiconductor company, Intel sees significant technology changes before most, however its enormous size makes it challenging to disentangle micro-trends from macro-performance. Doing so requires looking past the raw earnings numbers into the colorful comments provided during management's discussion of its report, which is why last week's Intel Q4 2016 earnings call was so revealing. Although the headline numbers, with both revenue and earnings beating analyst expectations by 4 to 5%, weren't earthshaking, the subsurface shifts within business units, particularly the growth engine that is the Data Center Group (DCG), are tectonic.
In the post-PC era, smartphones have become people's default device for information and communication and PC performance outstrips user needs, resulting in replacement cycles that have lengthened to a half decade or more. This has left Intel's thriving business of server processors responsible for top-line growth at the company. Revenue from Intel's Data Center Group (DCG) has increased faster than the declines at its Client Computing Group's (CCG). However, the heady days of mid-teens growth rates are over, with Q4 results showing an 8% increase. While DCG is still healthy, the contributors to its results indicate that it is going through a significant transformation with enterprise business slumping as cloud provider sales explode.
Picks and shovels for the cloud
The extent of Intel's changing data center business became apparent during the earnings conference call where CEO Brian Krzanich revealed that sales to cloud service providers constitute the majority of DCG's business, growing 24% last quarter even as sales to enterprise suppliers dropped 3%. Furthermore, Krzanich noted that sales to carriers and other telcos rose 19%. For the year, Intel CFO Bob Swan said that revenue to cloud and communications providers grew 30% even as sales to enterprises and government agencies fell 7%.
Intel has long promised revenue growth in the mid-teens for its data center products. When asked whether the slowdown was related to increasing share of cloud provider revenue, Krzanich had this to say:
[The] enterprise continues to decline. I think that certainly some of that is, that it's moving to the public cloud, it's moving to those areas at a faster rate than I think we expected …
Remember, for us, enterprise is now less than 50% of our overall Data Center business and the areas that are growing even faster or as fast as cloud in most cases are the networking and storage space, which we have very low market share in still, and it's a great opportunity for us.
The 37% shift in sales from enterprise to cloud customers illustrates how rapidly Intel has become tethered to the cloud ecosystem. This changing sales profile is a mixed blessing for Intel. While cloud services like AWS, Facebook, Google and Microsoft buy processors and motherboards by truck-full, they also operate at much higher efficiency than most enterprises, meaning they need far less hardware per unit of work. The counterbalance according to Swan is that cloud-builders "value performance" and prefer higher-end processors, shifting the mix to more expensive products.
Cloud operators are not only responsible for a greater share of Intel's server CPUs. They are also driving development and sales of other infrastructure products like flash and high-performance memory (3D XPoint), Ethernet switches and high-speed communications fabric (Omni-Path) controllers. Instead of feeling threatened as a growing share of business goes to a few big customers, Krzanich sees cloud services as a catalyst for growth:
We believe as the world becomes connected, cloud will grow at a much, much faster rate. And I made a point in the prepared remarks, where if you look at the cloud of today being mostly based on people, the average person will generate about 1.5 gig of data a day. An autonomous car, when those things start hitting the road – and we've started to build these data centers for some of the trials we're working with – I mean you're talking about petabytes of data that you're having to deal with and 4,000 gigabytes a day off is the average autonomous car. You put a couple of those on the road and you need petabytes of storage to handle that. So, we do believe that the cloud will move at a faster rate as these connected devices become basically more available. That said, the cloud is becoming bigger than the enterprise.
Intel's results reveal that enterprise IT's migration from private data centers to public service providers is happening faster than many people (including Intel executives) expected. Gartner, which predicts a $1 trillion shift in IT spending to the cloud by 2020, figures that cloud spending will still amount to less than a quarter of the IT budget by that time. I believe these numbers are vastly understated.
Although the outlook for cloud service providers couldn't be better, with AWS and Azure doubling in size every year or two, the implications for other IT suppliers are much less rosy. While Intel is benefitting, for now, it will face increased competition as cloud builders with their deep bench of engineering talent are much more open to alternative processor architectures like GPUs, ARM and POWER. Indeed, one of the beneficiaries has been NVIDIA, which has seen its stock price triple over the past year largely on the back of the increased use of GPUs for machine learning applications and the introduction of new GPU-based cloud services (see my coverage here and here). As CFO Colette Kress said on NVIDIA's recent earnings call,
GPU deep learning is revolutionizing AI, and is poised to impact every industry worldwide. Hyperscale companies like Facebook, Microsoft and Baidu are using it to solve problems for their billions of consumers.
Cloud GPU computing has shown explosive growth. Amazon Web Services, Microsoft Azure and AliCloud are deploying NVIDIA GPUs for AI, data analytics and HPC. AWS has recently announced its new EC2 P2 instance, which scales up to 16 GPUs to accelerate a wide range of AI applications, including image and video recognition, unstructured data analytics and video transcoding.
Traditional enterprise suppliers face even greater challenges as carriers and cloud providers increasingly eliminate the middleman and work directly with contract manufacturers. Indeed, IDC's latest server market estimate found that the fastest growing 'vendor', and one of the few categories with any growth, was 'ODM Direct' – sales by vanilla original device manufacturers. The customer exodus and sales deterioration is already evident at places like HP Enterprise, where server, storage and networking revenue dropped 9% last quarter, and Cisco, which saw most recent sales of its switches and servers fall 7% and 3% respectively.
Enterprise flight to the cloud is also ominous for emergent storage and hypercoverged infrastructure firms like Nimble, which continues to lose more money than analysts expected, SimpliVity, which HPE acquired for a fraction of its $1 billion-plus private valuation, and even HCI-pioneer Nutanix. As organizations become more familiar and comfortable with running critical workloads in the public cloud, widely-assumed growth of hybrid cloud infrastructure could vanish under a barrage of public cloud price cuts, compelling new services and corporate strategies emphasizing rapid IT innovation over operational control.
As I discussed in this column on conversational software interfaces, next-generation software is increasingly reliant on sophisticated cloud infrastructure and application services for data analytics, machine learning and cognitive computing. These require massive, distributed infrastructure and technical expertise to build and operate and it's unlikely many organizations will undertake the time and expense to replicate them with private infrastructure when renting is so easy and effective. As Tom Friedman notes in his recent book, Thank you for being late,:
Twenty years ago you needed to be a government to access this kind of computing power in the cloud … Now you only need a Visa card.
What Visa once did to layaway plans and Airbnb is now doing to timeshares is a phenomenon we can see happening next to dedicated enterprise infrastructure as cloud services take over.