Last year was a tough one forAdvanced Micro Devices.
The microprocessor company was hurt in particular by a sharp falloff in personal computer demand, as a pandemic-era surge in PC purchases faded away. The stock lost about half of its value—a decline of more than $100 billion—and fell a little behind the valuation of its rival, Intel.
That was the background as AMD CEO Lisa Su stepped to the CES stage Wednesday evening at the Venetian Resort in Las Vegas to deliver the opening keynote for the giant tech trade show’s 2023 edition. Su announced updates to AMD’s Ryzen processor lineup that include integrated artificial intelligence capabilities. That echoes a theme that will resonate all week at CES: AI has arrived, and it is showing in every conceivable kind of device.
The addition of AI features into AMD’s processors reflects the first instance of the company integrating technology from Xilinx, which AMD acquired in February 2022 for about $35 billion in stock. AMD contends that the Ryzen AI engine in its new processors is 20% faster than the same capability in the Apple(AAPL) M2 processor. AMD noted that Intel (INTC) doesn’t currently offer AI capabilities in its x86 based processors.
AMD also said there will be more than 250 new AMD-based notebook designs coming to market from HP Inc. (HPQ), Lenovo, and other major PC manufacturers this year, up more than 25% from last year.
Newsletter Sign-up
The company unveiled at the show were the Ryzen 7040 Series processor; the Ryzen 7045HX processor for high-end applications; a new series of gaming processors called the Ryzen 7000X3D; and the Radeon RX 7000 Series Graphics chips.
AMD also unveiled the Instinct MI300 accelerator chip for supercomputers and AI applications. The chip will be used in El Capitan, a supercomputer under development at the Lawrence Livermore National Laboratory. AMD said the new chips offer considerable performance and efficiency improvement from the current generation AMD data center accelerator, which is used in the Hewlett Packard Enterprise (HPE) Frontier supercomputer at the Oak Ridge Leadership Computing Facility in Tennessee. AMD said that AI training models that take months to run now could be done in weeks with the MI300.
“This is an incredible time to be in the semiconductor industry as the last few years have highlighted how virtually every product, service, and experience in our lives are powered by chips,” Su said.
Comments