The chipmaker is making much more than CPUs thanks to new AI/ML applications, plans for new fabs, a revised GPU strategy, and plans to become its own wafer foundry.
Pat Gelsinger summarized the world economy of the future in a few words to reporters and analysts following his keynote address at the Intel Innovation 22 conference here earlier today.
“The most important ingredient in humanity’s future is silicon,” the Intel CEO said. “Where the oil reserves have defined geopolitics for the last five decades, fabs and technology supply chains are more important for the next five decades. Let’s build them where we want them; let’s think about these things.”
While that may sound a little self-serving, there’s more than a modicum of truth in the statement. With electric power clearly the future of consumer and industrial transportation and fossil fuels rapidly becoming a dirty term in many industries, technology powered by silicon-based data processors certainly will provide in increasing scale the muscle required to get things done for humans, as it has been doing for decades.
The event, which attracted about 1,000 attendees in person and several thousand online, concludes Sept. 28 at the San Jose McEnery Convention Center. It was loaded with news about the venerable company’s IT hardware, software, partnerships, use cases, future plans, and product roadmaps.
Key takeaways included the following:
- Intel introduced its latest high-end CPUs, 13th-gen Core i9 processors, aimed at the gaming and content creator markets. These are available with up to 24 cores (8 P-cores, 16 E-cores) and 32 threads and up to 5.8 GHz in single-thread performance. In other words, as in all tech conferences: faster, stronger, better – according to the manufacturer.
- The company is going all out to attract developers of all kinds who previously might not have worked in hardware. It will use its revamped Developer Cloud to enable new and future hardware platforms to be immediately available for pre-launch development and testing, such as the fourth-gen Xeon Scalable processors (Sapphire Rapids) and Intel Data Center GPUs.
- The new Intel Geti platform, introduced today, enables enterprises to more easily develop and deploy computer vision AI. Gelsinger said he expects that over time Intel will roll out AI capabilities throughout the company’s massive product line.
- Intel now will serve as a systems foundry, combining wafer manufacturing, packaging, software, and the chiplet ecosystem – a hardware workflow it has not attempted previously. The company has 15 fabs in operation and is building new facilities in places that include Ohio, Arizona, and Germany.
- The company previewed future high-volume system-in-package capabilities that will enable pluggable co-package photonics for a variety of applications. Research and development involving transforming code into light beams is being conducted in Scotland; the company showed a live demo of the process at the conference.
Intel’s take on GPUs
As Gelsinger promised investors and customers when he took over the CEO job from Bob Swan in February 2021, Intel is ramping up to challenge global market leader Nvidia in the burgeoning graphics processing unit (GPU) space. The company revealed several milestones in its lineup of data center GPUs in addition to pricing and availability for the first Arc GPUs for gaming.
Gelsinger described the near future of GPUs at Intel this way: “When I left Intel 12 years ago, I had a list of 10 things I wanted to get done in the data center business I was running at the time,” he told reporters following the keynote. “I paused upon leaving Intel because No. 10 wasn’t done – and that was discrete graphics and throughput architecture. Coming back, we’re gonna get it done.
“We now have all the product categories: integrated graphics, discrete graphics, GPU, and HPC. All of those products are shipping by the end of the year; the portfolio is now fully available and we’re fully committed to executing upon it. Discrete graphics for that community is a critical one for ISV engagement.
“There are only three kinds of semiconductor companies: You’re either big, you’re niched, or you’re dead. We’ve got to be big.”
Intel has some work to do in this sector, however. Nvidia currently owns 80% of the overall GPU market, and that includes 82% of the discrete graphics card market. On the other hand, in Q1 2022 Intel was the biggest vendor in the personal computer GPU market worldwide, owning a 60 percent share. Nvidia took a market share of 21 percent while AMD had 19 percent.
Geti computer vision brings new uses for AI
Intel’s newly launched Geti computer vision AI platform enables line-of-business enterprise team members the ability to develop their own AI models. Instead of having to spend dozens or hundreds of hours training an app via video to perform specific functions, it can do this by ingesting only about 10 to 12 images for the software to remember – and subsequently learn the use case with up to about 90 percent accuracy. Add a few other images and the accuracy percentage rises even higher, Intel SVP Nick McKeown told ZDNET.
The platform is currently being tested by select customers and partners; it will become commercially available in Q4 2022, the company said.
Geti and another new product introduced at the conference, OpenVINO, are complementary in that they serve different AI modeling needs. Enterprise users can upload data and build computer vision models with Geti, then deploy those models using OpenVino at scale running on Intel hardware. Geti can output an optimized OpenVINO model, ready to deploy, saving additional optimization steps, McKeown said.
At the end of his keynote address, Gelsinger introduced onstage a surprise guest — Linux creator Linus Torvalds (at right in photo), to whom he presented the first Intel Innovation award. The Intel CEO described the Finnish technologist, who got his start developing software in the 1980s after reading Gelsinger’s book about programming the Intel 386 processor, as “one of the true thought leaders in all of the history of IT.”
“What I found fascinating about this event were Intel’s focus on software and the developer cloud,” Rick Villars, a longtime industry analyst with IDC, told ZDNET. “Intel has to get its innovation into the hands of developers. Some of that is about being much more clear and visible about expanding their processor portfolio with GPUs, DPUs, and specialized systems. What’s really changed is how software gets to market, the whole as-a-service model, and the whole control plane-based idea of ‘I have a way of getting that software to thousands of places or individuals.’ That’s what Intel is now doing.”