There is a major shakeup happening in the industry, and it impacts two companies we cover at STH. Pat Gelsinger who spent 30 years at Intel, then became the President and COO of EMC, and has been the CEO of VMware since 2012, will be replacing Bob Swan as the CEO of Intel. This is a big deal.
Pat Gelsinger Next CEO of Intel
On February 15, 2021, Pat Gelsinger will become the new CEO of Intel. The Intel board saw the need to change an industry narrative where customers were turning on Intel. On the consumer side, Apple issued its declaration of independence in terms of its M1 Arm-based CPU. AWS, Google, Microsoft, Oracle, and others are deploying AMD and Arm CPUs in their data centers. While insatiable market demand for compute as the new currency of industrial might has led to great financial performance, at some point, Intel needs to change the narrative from being the beneficiary of a rising tide lifting all ships.
One of the biggest challenges at Intel is its past success. Intel became so dominant in desktop and server markets with market share in the high 90% range, that it had limited competition. We focus on the server segments but there IBM Power is very cool, but IBM could not ignite a spark that would take it out of that single-digit server market share realm. AMD was effectively one failed product launch away from having to re-evaluate its role in the CPU market before Zen. AMD’s product concept was wise given that it was launching a solid product at a time when Intel’s process technology went from leading at 14nm to behind just trying to get 10nm out in quantity. Without 10nm, Intel could not deliver PCIe Gen4 servers. Without PCIe Gen4 servers, it did not have a market for its PCIe Gen4 SSDs, network adapters, AI accelerators, and other products.
At Intel, it had fallen into a common trap. As a company gets larger and sees success, more of the revenue comes from traditional lines of business. If you have a $1B product line in a non-competitive market, such as the server market had been from 2013-2019 or so, raising prices to existing customers by 10% is an easier way to get $100M of incremental (mostly margin) than building a new $100M product line from scratch. In consulting, we used to call the teams that worked at the same big customers “farmers” because they would work almost exclusively on the same account for years.
Hunters on the other hand were folks that went out and sold new types of projects and to new clients. In my consulting career, I spent much of my senior associate days as a farmer, but by the time I made director, all I wanted to do was to be a hunter and sell new products to clients that often used other consulting firms. Hunting is hard because one expends significantly more effort chasing smaller dollars which means to hit my sales targets I had to work on new offerings and find new customers often pitching against incumbent providers or to startups that had never purchased management consulting services. I still would hit my annual sales targets within 3-4 months, but there is always a sense of insecurity having to wake up and hunt in the market versus knowing you had a stable client. Selling even just $5M in consulting work often meant a large number of pitches and constant innovation to win relatively smaller deals that one would hopefully lead to a client becoming a longer-term client. Selling Core and Xeon x86 is not hunting, it is farming. XPU is hunting. As an example, Intel Xe, the company’s GPU efforts, is going after a market segment dominated by NVIDIA and AMD.
Perhaps the biggest question is how Pat will demonstrate that he is still a hunter, rather than a farmer. Over the past seven or so years, Intel became so dominant in x86 that when its process leadership failed the product side, and the product leadership did not adapt quickly, it started a cycle of disruption that has allowed AMD on the desktop and server-side, and Arm architectures to get footholds in the market. Intel became complacent as it effectively had a monopoly in the server and desktop space for years.
The 10nm delays, pushing the Ice Lake Xeon series to the point where it is news that they are now in production in 2021 instead of launching in 2019 has forced Intel into an unfamiliar space, the company needs to transform back from farming a customer base to again proving that it can deliver industry-leading innovation. Intel has set the path of innovation away from the CPU and to the vision of an “XPU”. With XPUs, Intel can move into new spaces such as GPUs, AI accelerators, and customized solutions with modular IP blocks. This is the right model for Intel’s future. The XPU focus acknowledges that moving to new segments increases its silicon TAM while also offering a wider range of competitors.
Challenges at Intel today mirror what happened at VMware. VMware was a clear pioneer in virtualization. These days, the large virtualization players most likely do not use VMware. Public cloud providers are not running atop of ESXi. Instead, they are using KVM virtualization. VMware-focused professionals still extol the virtues of VMware, but the cloud providers and even the Open Stack community has changed the market, similar to what Intel is experiencing.
Losing in its traditional market of virtualization, Pat and the VMware team recognized this was inevitable and moved to other market segments. For storage, we got vSAN even though that, in many ways, put the company in competition with EMC. While it would compete with EMC, vSAN allows VMware to compete with Nutanix, Microsoft, and others. For networking, the 2012 Nicira acquisition allowed VMware to add a network virtualization component and NSX. This led to VMware competing with many of the networking hardware and software vendors by changing where the center of networking and control logic resides.
VMware was not early to Kubernetes, Project Pacific/ VMware Tanzu is the company’s offering that arrives after much of the industry had already deployed the new standard. VMware needed to build an offering here in order to ensure its lucrative enterprise customers did not need to reduce their VMware server counts as workloads shifted from virtual machines.
While VMware lost an enormous amount of virtualization market share, the company has innovated either internally or through acquisition to expand to adjacent markets. VMware’s new paying customer sign-ups are not as numerous as some of the big cloud players, but by offering a more complete stack, it can remain relevant for its enterprise customers.
That strategy at VMware aligns nicely with Intel’s XPU efforts. With XPU, the company may lose x86 market share. It may lose share in desktops and servers, but the company’s fortunes lie in embracing the disruptors and offering high-value silicon solutions going after product segments that are new or that were previous adjacencies.
Intel as a company and the industry context dictates that there are only a handful of people that could assume the role of Intel’s chief and have a reasonable chance at success. Pat Gelsinger with his long history at Intel, along with the relationships he has built with Intel’s major customers at VMware mean he is on that shortlist of folks, and perhaps the top outside candidate.
We are seeing Intel move to its XPU strategy as an example of the company moving from farming its existing customer base to hunting new markets. The question now is what changes Pat will make to transition this farming culture to a hunting culture. Intel is large enough that it needs a mix of farmers and hunters, but it is clearly in need of changing that mix.
Congratulations to Pat on this personal milestone. There are very few people who start working at a company like Intel as a teenager, spend 30 years there, and do not at some point have dreams of becoming its CEO. As great as EMC, Dell Technologies, and VMware has been, we need to take a moment and recognize on a personal level this is someone achieving a dream.