Intel CEO on CHIPS Act: ‘Get that frickin’ thing done’
Intel CEO Pat Gelsinger has once again called for the US House and Senate to reach an agreement over the CHIPS Act as a matter of urgency.
The CHIPS Act is a piece of US legislation designed to facilitate the onshoring of a larger proportion of the semiconductor supply chain in the face of rising tensions with China.
Once passed, the act will unlock tens of billions of dollars in federal funding for semiconductor research and manufacturing, much of which will land in Intel’s pocket. The US House and Senate agree on the need for the legislation, but have been slow to iron out the specifics.
At Intel Vision 2022, Gelsinger explained the company’s current fab building and expansion projects are “either on track or ahead of schedule”. However, he also warned that the CHIPS Act is necessary to “accelerate the industry”.
Read our full report here.
Comments from the press Q&A
Pat Gelsinger, CEO, on ramping up manufacturing capacity:
“We’re on or ahead of schedule for all of our building an expansion projects, but it’s super important for the US House and Senate to finish the CHIPS act. Get this frickin’ thing done, because we need it to accelerate the industry. ”
Sandra Rivera, CVP and GM, Datacenter and AI, on hyperscalers building this own custom silicon:
“Hyperscalers are some of our big volume drivers, and they are counting on us to deliver sustainable, durable capability for their data centers. For their large-scale deployments, they still rely on Xeon.”
“But they also want durable innovation. We can deliver some of that with Xeon, and there is also the opportunity with our foundry business to enable hyperscalers to innovate on unique IP.”
Gelsinger, on the importance of software:
“I have more software engineers working for me now than I did at VMware. The role of software has become more important and we don’t believe that will change.”
“We’re going to do a lot more SaaS, and a lot more SaaS acquisitions, pulling silicon differentiation through SaaS services. Silicon plus SaaS equals solutions, and you’re going to see us doing a lot more of that solutioning.”
That’s a wrap
To close out the keynote session, Pat Gelsinger is back on stage.
“There’s so much that is simply not possible, yet. But we are setting a torrid pace, with our purpose at the center, to create world-changing technologies that improve the lives of everyone on the planet,” he says.
“Today’s rennaisance will set the stage for tomorrow.”
Gaudi 2 accelerator demo
The COO of Habana Labs is on stage to deliver a demonstration of the deep learning training performance advantages of the new Gaudi 2 accelerators, mentioned earlier in the blog.
In comparison to Nvidia’s A100 GPU, the Gaudi 2 delivers 2x the training throughput for both ResNet-50 and BERT, the most popular vision and NLP models, respectively.
“The higher the throughput, the faster the training, and the faster customers can achieve their goals,” says Eitan Medina, Habana COO.
Koduri invites to the stage Rick Stevens, Associate Director of Argonne National Labs, home to the Aurora supercomputer, which is currently under development.
Aurora is made of a large number of compute blades, each featuring 2x Sapphire Rapids CPUs and 6x Intel Ponte Vecchio GPUs, and each connected with thousands of other blades. When it goes live, the supercomputer will be among the most powerful in the world.
Announcement: Argonne will now academic invites to submit requests for time with Aurora while it’s still under development.
Intel Arctic Sound-M GPUs
Koduri holds aloft Intel’s new data center GPU, Arctic Sound-M, first announced earlier this year.
Artic Sound-M GPUs feature up to 4x Xe media engines and up to 32x Xe cores and ray tracing units, and can apparently support up 40+ HD gaming streams and 30+ HD video streams at once. “It’s a media supercomputer on a single chip,” says Koduri.
Arctic Sound-M will be available to customers next quarter.
Raja Koduri, head of Accelerated Computing and Graphics (AXG), hits the stage to talk about what’s next in the high-performance computing space for Intel. It’s a space Intel dominates, but AMD has been gaining ground in recent years.
ANNOUNCEMENT: New AI accerators from Intel’s Habana Labs
Intel has lifted the lid on a second generation of Gaudi accelerators that could reduce the time it takes to train large-scale AI models significantly.
The Gaudi 2 processors are built on a 7nm process, feature 24 integrated 100GbE RoCE ports and boast the largest quantity of memory of any accelerator on the market (96GB HBM2e).
The new processors are a product of Israel-based Habana Labs, which was absorbed by Intel back in 2019, and are designed for servers dedicated to deep learning workloads.
Read our full write-up here.
Rivera holds aloft a Xeon Sapphire Rapids wafer. Very shiny.
The company is already providing customers with 4th generation Xeon SKUs, but there’s still no news as to when we can expect wide availability.
ANNOUNCEMENT: IPU roadmap
Sandra Rivera, head of data center and AI, takes to the stage to deliver an update on Intel’s data center business line.
Rivera announces an extension of the company’s infrastructure processing unit (IPU) roadmap, all the way out to 2026.
The company plans to deliver three new generations of infrastructure processing units (IPUs) within the next four years, we’re told. The first new products – Oak Springs Canyon FPGAs and Mount Evans ASICs – will land by the end of the year.
We’ll have more details on this later.
ANNOUNCEMENT: 12th Gen Intel Core mobile workstation processor
Some news! Intel announces the launch of the 12th Gen Intel Core-HX mobile workstation processor, the final product in the family.
The new chip delivers up to 16 cores, up to 5Ghz clock speeds and “unrivalled performance”, we’re told.
It’s all about the PC
Pat Gelsinger exits stage right, replaced by Michelle Johnston-Holthouse (EVP and GM of Client Computing), who’s here to talk about the humble PC.
“The PC is the human touchpoint. We rely on to focus, create, connect and drive forward our business. I truly believe it’s here to say,” she says.
“I’d encourage you to imagine what the last two years would have been like if you’d had to work from your phone. The PC enables the workforce of the future.”
There’s ove 140 million commercial PCs that are four years or older, and this is holding business users back, adds Johnston-Holthaus.
IDM 2.0 and supply chain
Gelsinger takes a moment to nod towards the company’s IDM 2.0 strategy, the objective of which is to position Intel at the forefront of chip manufacturing during a period of unprecedented demand.
“We will build leading edge capacity in the US, to support the demand for these four superpowers,” he pledged.
“We’re just getting started delivering that geographically balanced supply chain capacity.”
The company has already sunk billions into new fabrication facilities across the US and Europe this year, but the implicit suggestion here is that there’s more to come.
An inflection point
“The world is at a strategic inflection point, a moment in time where things can go incredibly well, or incredibly poorly,” Gelsinger adds.
“Transformation is inevitable, it applies to all. Every business is becoming a technology business.”
There are four superpowers driving forward innovation, says Gelsinger: pervasive connectivity, ubiquitous compute, artificial intelligence and cloud-to-edge infrastructure.
“Each of these superpowers is impressive on its own, but when they come together, that’s magic.”
“If you’re not applying AI to every one of your business processes, you’re falling behind. We’re seeing this across every industry.”
Hi there, Pat
The Intel CEO takes to the stage to welcome the audience back to the first major in-person event the company has put on since the start of the pandemic (and since his return to Intel).
“Today, the pace of technology is the fastest of your life, but the slowest of the rest of your life,” he says. “We’re sitting at the precipice of a digital renaissance.”
We’re waiting now for Intel CEO Pat Gelsinger to take the stage for his keynote address. He should appear in roughly 20 minutes’ time.
We expect new product and service announcements, and maybe an update on the progress of the company’s IDM 2.0 manufacturing strategy, announced last year.
Good morning from Dallas
Intel Vision begins today and TechRadar Pro will be here with all the latest announcements.
First up is a keynote from CEO Pat Gelsinger, followed by more tightly focused sessions with the Client Computing and Data Center and AI business groups.
Rounding out the day is a roundtable session with Raja Koduri and Jeff McVeigh, during which we should hear more about the companies efforts in the graphics space, both from a device and data center perspective.
What to expect
Intel has kept its cards relatively close to its chest, but we do know we’ll be hearing from spokespeople across the company’s Client Computing, Accelerated Computing and Data Center business groups.
Essentially, there should be something here for everyone to get excited about, with product and service updates spanning the processors powering your devices, to the chips propping up the cloud.
In the meantime, here’s some reading on the latest news to come out of Intel: