When Intel was hunting for a new CEO in 2018, one technology pundit took to the Twitterverse to nominate then-VMware CEO Pat Gelsinger, Intel’s former CTO, as the best candidate for the job.
The ever-polite Gelsinger thanked the TV journalist for the shout-out, but said he had no plans to leave VMware for the chipmaker, closing his tweet with the exclamation, “The future is software!!!”
Now that Gelsinger finds himself three years later in that exact seat—having succeeded Bob Swan as Intel CEO this past February—his tweet remains no less prophetic.
[RELATED: Intel’s Gelsinger – AMD’s Lead ‘Over’ With New Alder Lake, Sapphire Rapids CPUs]
That’s because Gelsinger believes software has an outsized role to play in the future of computing and, consequently, at Intel. It’s driving him to rethink Intel’s software strategy and adopt a “software-first” approach that aims to make Intel the silicon platform of choice when it comes to running a vast array of applications, from the cloud to the edge.
To Gelsinger, this means engaging more independent software developers than ever before. It also means considering new software products and services, some of which could be paid offerings that are sold by channel partners. The new mindset means an Intel that looks very different from the one that Gelsinger left in 2009 after a 30-year run to join EMC.
“One of the things that I’ve learned in my 11-year ‘vacation’ [at VMware and EMC] is delivering silicon that isn’t supported by software is a bug,” Gelsinger said. “We have to deliver the software capabilities, and then we have to empower it, accelerate it, make it more secure with hardware underneath it. And to me, this is the big bit flip that I need to drive at Intel.”
Software has long been a strength of Intel’s. For the most part, applications just work on Intel’s central processing units, which has helped the Santa Clara, Calif.-based chipmaker make up for areas where it has fallen behind in the CPU market. And the company has been introducing a variety of new resources to ensure various workloads are optimized for Intel’s processors.
But Gelsinger wants Intel to do more and go faster. Applications are becoming increasingly complex, creating a greater need for heterogenous computing systems that consist of not just CPUs but other types of processors, like GPUs, FPGAs and a new category of chips, infrastructure processing units. And the field is getting more competitive, thanks to Nvidia, AMD, large Intel customers like Apple developing homegrown chips and a flood of semiconductor startups such as Ampere Computing and Fungible, both based in Santa Clara, Calif.
“For Pat, he’s driving a broad recognition across Intel that if we win with software, then we’re going to make it easier for our ecosystem of partners to deliver solutions around Intel,” said John Kalvin, who became Intel’s global channel chief in December to lead the new Intel Partner Alliance program. “The work that he did at VMware, he has obviously gotten deep and rich in software, and he is now coming back with a lot of energy and passion around making sure that we invest at an appropriate level in this particular area.”
Getting Intel’s ‘Mojo Back’
Gelsinger knows Intel wasn’t at the top of its game when he returned in February. The chipmaker had been facing increasing pressure from AMD on the CPU front, Nvidia’s GPU-powered AI growth engine continued to roar, and Intel was working on fixing a manufacturing issue that resulted in a six-month delay for products using its 7-nanometer process.
“I was dismayed a bit that we simply had lost that lean-forward mentality and that we’re going to be delivering leadership products, we’re going to be defining categories, we’re going to be creating new standards and interfaces,” he said.
But that’s changing, Gelsinger promises, with new advancements coming over the next few years in manufacturing and architecture, plus increased outsourcing to external foundries, that he said will result in more competitive products as part of Intel’s new integrated device manufacturing strategy, IDM 2.0.
“We’re ready to get the mojo back in that respect,” he said.
His comeback plan is nothing, however, without strong software support and an exhaustive roster of partners who know how to take advantage of Intel’s processors. That’s why Gelsinger is pushing Intel to “dramatically” ramp up partnerships with software providers and expand its software offerings.
He also wants to see channel partners up their software expertise.
“We have to grow the partners. Some of those will be ISVs. Some of those will be SaaS providers as well,” he said. “Many of those skills need to become part of our channel partners’ repertoires as well, as they increase their cloud and SaaS capabilities and their software capabilities.”
The top executive responsible for this software charge is Gelsinger’s former right-hand man and CTO at VMware, Greg Lavender, whom he hired as Intel’s CTO in June to lead the company’s new Software and Advanced Technology Group. The group’s goal, according to Gelsinger, is “to drive a unified vision for software” while ensuring it remains “a powerful and competitive differentiator” for Intel.
Lavender has been a self-described “hardware-software person” most of his career, which has given him a diverse range of perspectives in the IT industry. His industry view was honed during time spent in research labs, academia and startups as well as at large infrastructure companies like Sun Microsystems and Cisco Systems. He also knows what it’s like to be an end customer, having built out Citibank’s global private cloud as the bank’s CTO.
“I wasn’t surprised when [Gelsinger] called me up and said, ‘I need you to come over here to help me with my software business because we’ve got a great hardware business, but to be competitive, obviously in the AI [and machine learning] space, in the cloud space, we need to bring a software-first focus,’ which he talks about a lot inside the company,” Lavender said.
Intel has long gotten the foundations right when it comes to software, according to Lavender, and this includes the compilers, the firmware and the BIOS software that are so critical to ensuring that applications run correctly and as fast as possible on Intel’s processors.
What Intel needs now, Lavender said, is to create “more sophisticated software” to let Intel’s vast ecosystem of partners realize more value from a “total system” perspective rather than just a single processor or a single server in a cluster. This includes working with the largest players in data center and cloud infrastructure, ensuring a wider impact for the channel.
“What I bring and where Pat is taking us is whoever that software Infrastructure-as-a-Service provider is, whether it’s VMware, whether it’s Google, whether it’s [Amazon Web Services], whether it’s [Microsoft] Azure, whether it’s private clouds, hybrid clouds, we’re going to be there,” he said.
Intel’s new software plans won’t just impact the data center market, though.
As an example, Lavender pointed to the new efficiency cores and performance cores that will be introduced later this year in Intel’s next generation of CPUs for client devices, code-named Alder Lake. To help users get the most out of systems with Alder Lake, Intel is incorporating a new silicon-based feature called Intel Thread Director that will tell Microsoft’s new Windows 11 operating system how to balance workloads on the two core types to optimize performance and efficiency.
But Lavender said Intel can do more to open the hybrid processor’s capabilities to a greater ecosystem of partners, and he thinks Intel could do it in other areas, like security and power management.
“We have these efficiency cores and performance cores. Don’t just leave it up to the operating system to decide what to do. Let the OEMs, ODMs and the channel partners have access to APIs that can do whatever they want to do, let’s say, in telco or at the edge,” he said.
Corey Kirkendoll, CEO of Plano, Texas-based 5K Technical Services, said Intel has already proven that a tighter integration of silicon and software can open up new business opportunities for channel partners with the Intel vPro platform, a set of silicon-based remote management and security capabilities that has allowed him to patch computers quicker and more efficiently and increase services revenue.
“Any time that the software is written or has the ability to take advantage of the core features of the processors is always a win,” he said. “The only challenge I see with that is they have to make sure that they have a strong ecosystem of partners that are early to the game, to have something that is truly ready for prime time when it’s released.”
Tipping The Scales With Software Optimization
Miles Ward, CTO at SADA Systems, knows firsthand how Intel’s investments in software can help open new opportunities for channel partners and tip the scales in the chipmaker’s favor in the face of increasingly competitive processors from AMD.
SADA, a Los Angeles-based cloud-native solution provider, recently began a cloud optimization pilot program with Intel that is helping customers optimize Google Cloud instances powered by the chipmaker’s Xeon Scalable CPUs, which can result in significant cost savings.
“I think they’re trying to make sure that, for the places where they do have the performance advantage, they’re able to crow as loudly as their competitors can,” Ward said.
The pilot is initially focused on SADA’s largest customers, and so far, the results have been promising: Participating customers have reduced compute costs on average by roughly 5 percent, which is a big deal considering the hundreds of thousands, if not millions of dollars they’re spending. That not only benefits customers but also helps SADA compete against other cloud resellers.
“The biggest motivation for most customers is going to be some material cost savings,” he said. “Google has some general price advantages, but if you’re then talking about millions to tens of millions of dollars of annual consumption, a couple percentage points [of savings] is bigger than a discount incentive I might provide or other kinds of incentives to drive their movement.”
In several cases, Ward said, Intel’s optimization work has made a crucial difference in shifting the advantage from AMD- to Intel-powered instances. Part of this is made possible by new features in Intel’s Xeon Scalable CPUs that can, for example, accelerate encryption and machine learning workloads.
But for large customers with specific application needs, such optimization work also requires time spent by SADA’s engineers to test various configurations, which are then reviewed and tweaked by Intel’s engineers as part of SADA’s pilot program with the chipmaker.
“It’s the Intel engineers that are reviewing those configurations and suggesting patches and alterations to the configuration process and other kinds of instructions we want to put in at the machine or network level to get to optimized results,” Ward said.
The pilot with SADA is part of a wave of new investments Intel is making to provide more technical resources and expertise to partners for optimizing software on the chipmaker’s products. In the first six months of this year, Intel has added more than 100 partner- and customer-focused technical sellers, most of them focused on cloud and IoT applications. That workforce has increased by as much as 20 percent since 2018.
“We see that deep integration with our customers as being more and more important,” said Michelle Johnston Holthaus, Intel’s executive vice president, CRO and general manager of the Sales, Marketing and Communications Group. “Not only does it help our customers and offer value that I would say a lot of our competitors can’t, it also helps drive affinity for Intel’s brand at the edge, which is super important for us as well.”
Intel is adding other kinds of cloud resources for partners too, including a new training program, a development sandbox and a cloud optimization tool. These efforts will help Intel reach its wider channel, said Kalvin, head of Intel’s global partners and support.
“We’re stepping back and saying, ‘OK, how do we take the learnings that we have from that deep, enriched work that we’re doing on cloud placement and then putting that into solutions, documentation, training that we can then bring to our partners at scale around the world,’” said Kalvin.
Playing Catch-Up To Nvidia In AI
Gelsinger sees a lot of challenges facing Intel, but there’s one that is coloring his software investment strategy more than others: artificial intelligence. Even Arm, the British chip designer that is enabling large customers like AWS to create competitive homegrown chips, doesn’t rank as high.
“The architectural disruption that I’m more concerned about in the data center is the AI one, not the Arm one,” he said.
To Gelsinger, this means fighting an “uncontested” Nvidia, Intel’s Silicon Valley neighbor that has dominated the accelerated computing space with its graphics processing units.
It’s why he has tasked Lavender with “rapidly pulling together a singular AI software stack” to support Intel’s range of business and hardware offerings, including the upcoming Ponte Vecchio data center GPU Intel promises will deliver breakthroughs in AI and high-performance computing.
Lavender knows Intel is now in catch-up mode, but he doesn’t believe Intel has to “go head-to-head” against Nvidia in every one of the verticals the GPU maker has built software for, like medical imaging and smart cities. Instead, he said, Intel will lean on the wider software ecosystem, which is part of the chipmaker’s broader goal in enabling new business models for partners.
“Nvidia’s done a good job of creating the market; therefore they have lock-in and therefore they have price leverage with their customers,” he said. “If we can come out with competitive products, and we enable the same software ecosystem that the rest of the world is using … and let other people make money, not just Nvidia, then I think that the market will accept that.”
What also differentiates Intel’s approach is how it enables software developers to accelerate those workloads not just on GPUs but also CPUs, FPGAs and other products in Intel’s portfolio. This is made possible by a new effort started a few years ago: Intel oneAPI, a set of toolkits that lets developers use a single programming model for different types of architectures. It also supports GPUs from Nvidia and AMD as well as CPUs based on Arm architectures.
Gelsinger said oneAPI’s open-standards approach serves as an important contrast to the one offered by CUDA, Nvidia’s parallel computing platform and programming model for GPU-accelerated applications.
“Nvidia has become too proprietary, and that’s widely seen in the industry, and so we’re going to fill out that stack with oneAPI but do it in a way that’s much more favorable and open to the industry and their innovations,” Gelsinger said. Nvidia did not comment as of press time.
But for all this to work, Intel needs to make an unprecedented push in getting software developers to use the company’s toolsets and oneAPI programming model, Lavender said.
“We’ve got all the great software and hardware, but how do I enable them to want to choose it when they have a choice, and that’s really where we want to reach out in a new way that historically we haven’t,” he said.
Leveraging Intel’s Software Assets
One channel partner who is eager to do more business with Intel is World Wide Technology, the St. Louis, Mo.-based solution provider that ranked No. 9 on CRN’s 2021 Solution Provider 500 list. Jim Kavanaugh, the company’s CEO, believes one way Gelsinger could make this happen is by taking a page out of Nvidia’s playbook and offering software and systems that partners can resell.
“I’m not saying that’s the way that Pat and Intel should be doing things, but it’s something that, if I was in his shoes, I’d be thinking about,” he said. “Does it make sense to do some of that?”
While WWT has long considered Intel a key strategic partner for many of the solutions it delivers, Kavanaugh has seen the art of the possible with the “rocket ship” he calls Nvidia. That view is reinforced by the fact that while WWT’s Nvidia GPU business is smaller, the five-year growth rate for that business is roughly 40 percent, double that of WWT’s growth rate with Intel.
“We’re showing a little over 20 percent growth year over year [with Intel], which is good growth, but we’d like to see even better growth,” Kavanaugh said.
One key area to WWT’s growth with Nvidia are the products channel partners can resell. Whereas Nvidia used to only sell GPUs that go inside systems, the chipmaker now sells completely integrated AI systems under the DGX brand that partners can sell like they do servers from OEMs.
The GPU maker also sells a portfolio of paid enterprise software that now consists of 10 offerings. One of the latest, a software suite for VMware’s vSphere platform called Nvidia AI Enterprise, makes it easier to manage AI workloads in traditional data center infrastructure, and Nvidia said it sees a multibillion-dollar opportunity over time with that product alone.
Intel, on the other hand, has become more abstract to channel partners who have left behind their system building practices to focus on reselling OEM systems, cloud instances and software.
That has been a challenge, Kavanaugh said, but the two companies have made substantial progress in working together. Much of this is focused around WWT’s Advanced Technology Center, where WWT works with Intel to test new and emerging workloads on a variety of Intel-based solutions from different OEMs and ISVs.
“Intel, to a certain degree, can be very abstract to the customer or to the solution, but they’re very instrumental to the technology advancements and how that solution works,” he said. “So I think it is important that you can really understand the benefits and leverage the benefits of the Intel technology and thought leadership.”
Gelsinger understands the benefits of a chipmaker offering paid software. After all, he helped lay the foundation for GPU virtualization solutions like Nvidia AI Enterprise as VMware’s CEO by fostering a strategic partnership with Nvidia.
He isn’t ready to give too many details, but Gelsinger said Intel is considering new paid software services that channel partners could resell. These would add to the paid software offerings Intel already has, like Intel Unite for collaboration and Intel Data Center Manager for real-time monitoring and management as well as support services for Intel oneAPI and OpenVINO. He calls this a “somewhat natural progression” that mirrors the software industry’s move to “SaaS-oriented” delivery.
“I do expect that you’ll see more in that area: How do we leverage our software assets? How do we have unique monetized software assets and services that we’ll be delivering to the industry, that can stand in and of their own right? And yeah, that’s a piece of the business model that I do expect to do more of in the future,” he said.
Gelsinger is careful to emphasize that Intel wants to work together with its large ecosystem of partners for any new paid software services it may introduce.
“Having done a lot of this for the last eight years at VMware, I’ve gotten a deep appreciation for it and the challenges,” he said. “I’d also say, though, what are the ecosystem-friendly views of those services? How do we build on what others do?”
But for channel partners to truly take advantage of where Intel is going next, Gelsinger said they also need to evolve beyond just selling systems and build practices focused on software and services.
“I think many channel partners got too comfortable with the sheet-metal-oriented business models, and the ones that I have the most respect for have, at a minimum, started to deliver more of the software, SaaS and additional services on top of it and, in many cases, built entirely new business practices,” he said.
Kirkendoll, the CEO of 5K Technical services, said Gelsinger couldn’t be more on point.
“The box is almost a commodity today. That’s the necessary vehicle that gets you there. It is always about the value-added [software,] services and support that you add on top,” he said. “And if Intel can help us be better at that, I agree, 100 percent.”