AMD is firmly committed to data center and AI dominance. To achieve this, it will have to contend with both old (Intel) and newer competitors (Nvidia). In San Francisco it brings its strengths to the table.

AMD chooses a slogan for its annual conference in San Francisco that leaves little to the imagination Advances in AI. Not very original, because these days every tech conference is about AI. But this slogan is very relevant to AMD’s current course: the data center is the goal, AI shows the way. PCs are barely mentioned during the event, but even then a reference to AI isn’t long in coming.

“We believe that High performance computing is the basic building block of innovation. AI is the most exciting application of HPC. “AMD wants to be the end-to-end market leader in AI,” explains CEO Lisa Su in her opening speech, explaining the company’s focus. The confidence that Su exudes shows AMD’s strong belief in achieving this goal.

Rapid progress in the server market

When I heard the name Advanced micro devicesAs AMD is fully aware, your first reflex may be to see if there is a sticker from the company on your PC. For many, AMD is synonymous with laptop and desktop processors, and for almost fifty years AMD was also the company’s main source of revenue.

That changed when Su, who celebrated her tenth anniversary as CEO during the conference, took over. Su believed that the company should expand its horizons beyond the PC. This resulted in the first generation of Epyc processors in 2017, named after Naples, with which AMD delved into the deep waters of the server market.

Seeing Naples and then dying is out of the question with AMD. During her keynote, Su proudly holds what is now the fifth generation of Epyc in her hand. It’s hard to see the pitch from the press stand, but everyone tries to capture it as close as possible from all angles.

This year’s trip to Italy will end in Turin. With up to 192 cores and based on the latest Zen 5(c) architecture, Epyc Turin shouldn’t lack the power to handle heavy (AI) workloads in data centers.

“You should know that five years ago we represented maybe two percent of the server market. In the first half of 2024, our market share increased to 34 percent. Data centers now account for fifty percent of AMD’s business. Thanks to these new innovations, we believe we can continue our rapid rise. “Epyc Turin is a beast,” says Dan McNamara, head of AMD’s server business, sounding as confident as its CEO.

Epyc Turin is a beast.

Dan McNamara, VP Server Business Unit AMD

Old and new acquaintances

By unleashing the Epyc Turin beast in the AI ​​space, AMD aims to devour its competitors. A competitor knows it inside and out: the eternal feud with Intel is also taking place in the server industry. Now that Intel is struggling, AMD sees an opportunity to deliver a knockout blow. While we wait for the market numbers to finally turn around, a technological victory over Intel is already being announced.

As many superlatives as AMD attaches to Epyc Turin, the company has realized that good CPUs alone are not enough to rule in the data center. A certain company called Nvidia has won many souls with its Hopper GPUs. With the Instinct accelerators, AMD is now entering Nvidia’s territory and has no intention of backing down.

“Until now, AI applications have been largely supported by CPU capacity. New developments such as Agentic AI lead to a shift in market demand and increase the need for GPU capacity,” says Su. With the MI300X, AMD already had a direct competitor for the Nvidia Hopper H100, and now they have found a challenger for the H200 with the MI325X. The response to Blackwell is well underway and is expected in spring 2025.

It was written in the stars that AMD and Nvidia would meet in the data center. Both companies battled in secret for years over who could make the best graphics cards for PCs, when only gamers and hobbyists were awake at night. The AI ​​craze presented an opportunity to come to the fore and it can be said that Nvidia has succeeded best (for now), although AMD obviously sees it differently.

Now Nvidia and AMD are looking each other in the eye again at the highest level. At the same time, they also have to come to terms with each other, because AMD is working with Nvidia to link Epyc processors with Hopper GPUs. Keep your friends close and your enemies closer: Nowhere is this saying truer than in the tech industry.

Enthusiasts don’t have to wait long for the latest chips from AMD. In a flurry of announcements, Dell, HPE, SuperMicro and Lenovo take the stage to unveil their latest AMD Epyc Turin and MI325X servers. “Maybe next time we’ll have to start handing out order forms,” jokes Forrest Norrod, head of Datacenter Solutions Business.

Feed the beast

AMD’s arsenal is not yet complete with CPUs and GPUs. In data centers, the network plays a crucial role in connecting hardware. Since acquiring Pensando, AMD has also added DPUs (Data processing units) on offer. The chip manufacturer is happy to focus on this as part of the event.

“The combination of CPUs and GPUs makes the parts stronger. Networks used to be relatively simple, sorry to all the network specialists in the room. AI has made this much more complex,” says Norrod.

Soni Jiandani, who joined AMD’s networking division after acquiring Pensando, agrees with this statement. “AI encompasses an entire ecosystem. We must innovate on all fronts. AI systems connect to the front end, GPUs communicate with the back end of the network. The beast needs to be fed.”

AMD wants to distinguish itself as the party that is present everywhere in the data center. In the presence of representatives from Cisco, Microsoft Azure and Oracle, among others, Jiandani can show the press the latest Pensando Salina DPU. It operates on the front end of the network, but AMD is also thinking about the back end with Pollara, an Ethernet-compatible network card that brings additional performance, scalability and efficiency to a data center network.

From the PC to the data center and back again?

The renewed focus on data centers doesn’t mean AMD has completely forgotten its roots in the PC industry. But it takes until the last ten minutes of the keynote before we see the first PC. Lenovo guest speaker Vlad Rozanovic announces a ThinkPad PC with Ryzen AI Pro. The introduction of the Ryzen AI Pro label for Copilot+ PCs appears to serve primarily as an invitation for Microsoft to re-promote Copilot.

Vamsi Boppana, AI specialist at AMD, and McNamara admit that there is even a certain contradiction in Copilot+’s presence at the event. The explosion of AI means endless investment in data centers, which AMD would obviously like to see, while the idea behind Copilot+ is to take AI workloads out of data centers and run them locally on the device.

“Different workloads require different technologies and methods. You don’t always have time to go through the cloud,” says Boppana. The gentlemen are then asked whether AMD has one Bag to the cloudStrategy as Lenovo strives for. “If people ever walk around with Epyc processors in their pockets, we will be very happy,” replies McNamara with a joke.

AMD has grown up in PCs and now wants to take over the reigns in the data center. When the company felt it was reaching its limits in the PC market, it ventured into the server and data center industry and is reaping the benefits today. AMD is ready to fight Nvidia for the AI ​​crown.

Or will AI mean that AMD will soon have to take the opposite path towards PCs (and Intel) again? Whichever path AMD takes, it always seems to run into old friends.