Gelsinger combined his vision for the company’s future with new announcements and the short and medium term road map to highlight the extensive changes coming, particularly from the rise of AI.
SAN JOSE – At the 2023 Intel Innovation Forum here last week, company CEO Pat Gelsinger outlined his vision of how Intel – and developers who work with them – are poised to rule what he calls the Siliconomy. The Siliconomy, a term Gelsinger coined, reflects the fact that the digital economy is based on silicon, and that the digital economy has been growing two and a half times faster than GDP over the last decade.
“AI represents a generational shift in computing that is giving rise to the Siliconomy,” Gelsinger told his developer-centric audience. “We have a $574 billion silicon industry, powering an $8 trillion tech industry. As new levels of silicon increase the ability of semiconductors, we have had a 4x increase in connected devices over the last five years.”
This is great news for developers, Gelsinger stressed.
“Developers rule,” he said. “You run the global economy. Intel Developer Cloud (cloud.intel.com) will give you access to the coolest hardware and software as early as possible.” Gelsinger than announced the general availability of Intel Developer Cloud, which had been in beta, as well as a free week of Intel Developer Cloud access for all the conference attendees. Gelsinger said that the purpose of the Developer Cloud is to give developers quick access to pre-production hardware, where they can build and test applications on things like the Intel Gaudi 2 accelerators, the 4th Gen Xeon processors, the CPU Max series and Intel Data Center GPUs. It also gives access to Intel software for functionality like AI, deep learning, high-performance computing, and rendering. Freemium, commercial and enterprise solutions will all be available on the Developer Cloud.
“We are looking at a broad developer demographic for Developer Cloud, from startups to large players like Dell, which will use them for onboarding its Gaudi use cases,” Gelsinger said.
“AI is now 50 years old, but for 40 years of that, nothing happened,” Gelsinger noted, making a rueful reference to his own work on the 80486 chip over 40 years ago, which they mistakenly thought would put AI on the map. “The last ten years have been incredible, with AI redefining itself every two to three years, and impacting every domain along the way. Intel Gaudi rivals the market leader [NVIDIA] and at a much better TCO.”
Intel now has multiple partnerships brewing with strategic partners around major AI initiatives.
“We have a new partnership with Stability.ai, where we are building a large AI supercomputer to be one of the top 15 in the world,” Gelsinger said. He also noted that Dell has a major partnership with them around Gaudi, Xeon and Dell PowerEdge servers, and that Alibaba and Acer were two other partners engaged with Intel in major joint AI products.
“We do not want to compete with our cloud partners,” Gelsinger stressed.
Gelsinger then highlighted elements of the road map, which he termed extremely robust, and emphasized that Intel remains committed to Moore’s law so developers can develop with confidence on their platforms.
Citing Moore, Gelsinger said that “nothing can go on forever, no physical quantity can continue to change exponentially forever, but it can be delayed. And he was often amazed by the creativity of how we continue to find workarounds to barriers. We at Intel see ourselves as the stewards of Moore’s Law, and this relentless pursuit for computing at efficiency and scale, and we will not rest, we are committed to continuing this pursuit.”
Gelsinger then announced that the new Intel Core Ultra processors, formerly code named Meteor Lake, will launch on December 14, This contains Intel’s first integrated neural processing unit [NPU] for AI.
“This is the first client made using our Intel 4 process mode,” Gelsinger said. “It is power efficient and at price points for high volume development.”
Launching concurrently with the Intel Core Ultra on December 14 will be the 5th Gen Intel Xeon processors. IT will deliver innovative E-core efficiency to compete in the cloud and strong P-core performance for critical workloads like AI, which Gelsinger said would enable them to tackle any workload.
“We have addressed every phase of the AI continuum,” he stated.
“It’s AI inferencing and development, have led to a 90% increase in downloads in the last year,” he said. “You write once and bring AI everywhere.”
Gelsinger emphasized as well that the channel would play a key role in this.
“The channel is a priority for us here,” he said. “Our AI-enhanced Xeon offerings have become more important for sales, so that’s why channel programs and channel field resources are of interest. There is industry interest in a good alternative to the market leader here.”
Gelsinger also announced that ARM is now supporting the OpenVINO platform, so that it is now on both ARM and x86.
“This brings us closer to the vision of any model, on any hardware, anywhere,” he said.
Since Intel believes that inferencing will be increasingly hybrid, Gelsinger said that Intel will release a Hybrid AI SDK, coming early next year, designed for model and application development built into a low-code environment.
Also coming in early 2024 is Project Strata, an edge-native software platform designed to manage the exploding array of new cloud-native applications outside the data centre, which will help developers build, deploy, run, manage, connect and secure distributed infrastructure, applications and AI models at the edge.
“We believe that the next decade won’t be cloud-native but edge-native, and will need app development to do the hard work,” Gelsinger said. “This is what Project Strata does. It’s a modular edge native software platform to accelerate Ai at the edge. It will bring back vision and passion and will launch in early 2024.”
Gelsinger also announced that on the day before the event began, Intel had announced that they were moving to the next generation of packaging, with glass substrates.
“We are excited to be leading the industry in bringing this to market,” he said.
Gelsinger said that all this change was necessary as he sought to improve Intel’s long-standing issues with execution after his return from VMware.
“There are three types of semi-conductor companies – the big, the niche, and the dead,” he commented. “We were too big to be niche. We do the best process technology on the plant, so we made the decision that we would be a foundry at scale. We made some hard organizational changes, including putting much more into capital and paying less to shareholders.”