AMD: Instinct AI Chip Sales Exceed Expectations As Microsoft Expands Consumption

BY DYLAN MARTIN

JULY 30, 2024, 7:19 PM EST

AMD CEO Lisa Su says a ‘steep ramp’ of Instinct MI300 GPUs allowed the product family to exceed $1 billion in quarterly revenue for the first time. This meant the MI300 chips made up more than a third of AMD’s reported $2.8 billion in second-quarter data center revenue.

AMD said sales exceeded expectations in the second quarter for its Instinct data center GPUs, EPYC data center CPUs and Ryzen client CPUs, which helped it boost gross profit by 17 percent and double data center revenue from a year ago.

On its earnings call Tuesday, AMD CEO Lisa Su said a “steep ramp” of the chip designer’s Instinct MI300 GPUs allowed the product family to exceed $1 billion in quarterly revenue for the first time. This meant the MI300 chips made up more than a third of AMD’s reported $2.8 billion in data center revenue for the second quarter.

Data center revenue was up more than 115 percent year-over-year, primarily due to the fast ramp of Instinct chips along with strong sales growth of its fourth-generation EPYC CPUs.

Meanwhile, client computing segment was up 49 percent-year-over-year to $1.5 billion while the gaming segment was down 59 percent to $648 million from a year ago and the embedded segment was also down 31 percent across the same period.

This added up to a total revenue of $5.8 billion for the second quarter, which was 9 percent higher than it was a year ago and up 7 percent from the previous three-month period.

The stronger-than-expected demand for Instinct chips prompted the Santa Clara, Calif.-based company to upgrade its forecast for full-year data center GPU revenue to $4.5 billion, which is $500 million higher than the guidance it issued in April.

For the third quarter, AMD expects total revenue to reach roughly $6.7 billion, plus or minus $300 million, which would represent a roughly 16 percent increase year-over-year and a 5 percent sequential increase, according to the company.

AMD’s stock price was up more than 7.5 percent in after-hours trading.

The chip designer reported its momentum as it seeks to whittle away at Nvidia’s dominance of the AI computing space, where its rival’s data center GPUs like the H100 have been in high demand due to a flurry of generative AI development.

“This has been an incredible ramp, and I’m actually really proud of what the team has done in terms of definitely the fastest ramp that we’ve ever done,” Su said on call.

One key customer that has contributed to AMD’s fast ramp of Instinct GPUs is Microsoft, which is using the GPU-only MI300X to power OpenAI’s GPT-4 Turbo model and multiple Copilot services, including Microsoft 365 chat, Word and Teams, according to Su.

The Windows giant also became the first large hyperscaler to launch public MI300X-based instances through Azure in the second quarter.

Among the first customers to use the new MI300X-based Azure instances was model repository Hugging Face, which is enabling one-click deployment for AI models.

Su said Dell Technologies, Hewlett Packard Enterprise, Lenovo and Supermicro all have Instinct server platforms in production. In addition, multiple hyperscalers and tier-two cloud providers “are on track to launch MI300 instances” in the third quarter, she added.

The company has also made progress in enabling customers to implement the latest AI models on its Instinct platforms, according to Su. This activity includes the MI300’s industry-first support of Stability AI’s new Stable Diffusion 3.0 image-generation model and its support for Meta’s recently launched Llama 3.1 models.

Su said AMD remains on track to launch its next-generation Instinct GPU, the MI325X, later this year, which will be followed by the MI350 series in 2025 and the MI400 series in 2026.

“Looking across our AI business, customer response to our multi-year Instinct and ROCm roadmaps is overwhelmingly positive, and we are very pleased with the momentum we are building,” she said.