With AMD finishing its 2024 fiscal year with more than $5 billion in revenue from its Instinct data center GPUs, the company’s CEO, Lisa Su, says the company expects the AI chip segment to reach tens of billions of dollars in annual revenue in the ‘coming years.’
AMD CEO Lisa Su said the company expects its Instinct data center GPU revenue to reach tens of billions of dollars in annual revenue in the “coming years” after it generated more than $5 billion from the AI chip segment last year.
Su made the comments during AMD’s fourth-quarter earnings call on Thursday, where she also revealed that the company plans to start shipping its next-generation Instinct MI350 GPUs in the middle of the 2025, sooner than its original target window for the second half of the year. She added that AMD plans to start sampling the chips with customers by March.
[Related: Intel Cancels Falcon Shores AI Chip To Focus On ‘Rack-Scale Solution’]
“The customer feedback on MI350 series has been strong, driving deeper and broader customer engagements with both existing and net new hyperscale customers in preparation for at-scale MI350 deployments,” she said.
AMD beat Wall Street’s expectations for fourth-quarter revenue with $7.7 billion, which was 12 higher sequentially and a 24 percent increase from the same period. The two main drivers were its data center segment, which nearly doubled in sales again year over year to a record $3.9 billion thanks in part to a “strong ramp” of Instinct GPUs, and its client segment, which grew 52 percent year-over-year to a record $2.3 billion on the strength of its Ryzen chips.
Touting AMD as the “only provider with the breadth of products and software expertise needed to power AI across servers, edge devices and PCs, Su said the company made “outstanding progress building the foundational product, technology and customer relationships needed to capture a meaningful portion of this market.”
“We believe this places AMD on a steep long-term growth trajectory, led by the rapid scaling of our data center AI franchise from more than $5 billion of revenue in 2024 to tens of billions of dollars of annual revenue over the coming year,” she said.
While AMD’s $5 billion-plus of Instinct GPU revenue in 2024 is a fraction of the nearly $80 billion Nvidia made in data center revenue for the first three quarters of its recently completed fiscal year, the figure is nonetheless a significant increase from sales that were likely in the hundreds of millions in 2023 for the company’s product line.
AMD’s Expectations For Instinct MI325X, MI350, MI400 GPUs
Su said the company began volume production of its Instinct MI325X GPU in the fourth quarter of last year and touted “new customer wins” thanks to its “significant performance and [total cost of ownership] advantages compared to competitive offerings.”
The MI325X is the follow-up to AMD’s Instinct MI300X that launched in late 2023 as the company’s first significant competition to Nvidia’s data center GPUs.
“We have also made significant progress with the number of customers adopting AMD Instinct. For example, we recently closed several large wins with MI300X and MI325X at lighthouse AI customers that are deploying Instinct at scale across both their inferencing and training production environments for the first time,” Su said.
Among the customers adopting Instinct now are Meta, which uses the MI300X exclusively to power its 405-billion-parameter Llama model, and Microsoft, which uses the same GPU, to “power multiple GPT-4-based Copilot services” in addition to “flagship instances that scale up to thousands of GPUs for AI training and inference and HPC workloads.”
More than a dozen cloud service providers—including IBM, Digital Ocean and Vultr—use Instinct platforms for new instances, and Su said that number will grow this year. She added that more than 25 MI300 server designs are in production with the largest OEMs and ODMs.
As for the MI350 series, Su said there’s “very strong customer demand” for the GPU line, which is expected to boost AI performance by 35 times, the largest leap in the company’s history. And by moving up the MI350 series’ production ramp to the middle of this year, the move “improves our relative competitiveness,” she added.
“It’s a more powerful GPU. [Average selling prices] go up, and you would expect larger deployments that include training and inference in that timeframe,” Su said.
Then in 2026, AMD plans to launch its MI400 series GPUs, which is expected to enable “powerful rack-scale solutions that tightly integrate networking, CPU and GPU capabilities at the silicon level to support Instinct solutions at data center scale,” according to the CEO.
“We see significant traction and excitement around what we can do there with rack-scale designs and just the innovation that’s going on there,” she said.
How AMD Performed Elsewhere
While AMD’s Instinct GPUs played a major role in the company’s data center growth in the fourth quarter, it also cited an increase in sales of its EPYC server CPUs.
“2024 marked another major inflection point for our server business as share gains accelerated driven by the ramp of 5th-Gen EPYC ‘Turin’ and strong double digit percentage year-over-year growth in 4th-Gen EPYC sales,” Su said.
This was driven by a 27 percent increase last year in EPYC-based cloud instances, which now amount to more than 1,000, over 100 of which were stood up in the fourth quarter alone.
AMD also saw EPYC CPU sales grow in the “strong double-digit percentage” last year for on-premises enterprise deployments, with more than 450 EPYC-based server designs available from top OEMs and ODMs, more than 120 of which came online in the fourth quarter using the company’s latest Turin processors from the likes of Dell Technologies, Lenovo and others.
“We are seeing growing enterprise pull based on the expanding number of EPYC platforms available and our increased go-to-market investments,” Su said.
Growth in AMD’s data center and client segments offset a 59 percent year-over-year decline in AMD’s gaming segment, which made $563 million in the fourth quarter, and a 13 percent decrease in the company’s embedded segment, which generated $923 million in the period.
The company said it expects to make roughly $7.1 billion, plus or minus $300 million, in the first quarter of this year, which would represent a sequential decline of approximately 7 percent and a 30 percent increase from the same period last year. This is slightly above the average estimate of Wall Street analysts.
AMD’s stock price was down more than 8.5 percent in after-hours trading.
