Updated: 4-AMD Expects Quarterly Revenue to Exceed Expectations, Data Center Revenue Hits Record High

Title and paragraphs 1-4 have been rewritten and analyst comments added.

Anhata Rooprai/Max A. Cherney

Reuters, November 3 - AMD AMD.O on Tuesday forecasted fourth quarter revenue to exceed Wall Street expectations, and driven by strong demand, its data center revenue reached an all-time high. Meanwhile, AMD’s multi-billion-dollar plan to challenge Nvidia's dominance in the AI chip market is accelerating.

This optimistic outlook eased investors’ concerns, which had been heightened due to the rising costs associated with large-scale AI infrastructure and Intel’s bleak prospects.

AMD shares surged over 8% in after-hours trading. The stock has more than doubled so far this year. Intel shares rose 2%, Nvidia gained over 1%, and Broadcom was up 1.2% as well.

Data center revenue soared 97% to a record $8.55 billion, surpassing expectations of $8.2 billion.

According to data compiled by London Stock Exchange Group (LSEG), the company expects fourth quarter revenue of about $17 billion, with a fluctuation range of ±3%, higher than the analysts' average expectation of $16.23 billion.

Hendi Susanto, portfolio manager at Gabelli Funds, stated that AMD’s chip lineup is poised to benefit from large-scale AI data center construction.

"Although AI-related stocks are generally under pressure, this stock remains resilient, indicating that investors believe AMD’s performance reinforces its technological leadership and strong momentum in continuing to grow market share," Susanto said.

As AI workloads shift to model operations in practical applications, demand for AMD’s GPUs and CPUs has increased, benefiting the company.

“AMD’s financial results not only provide the much-needed boost for AI infrastructure but also, to a lesser extent, for the PC market,” said Bob O’Donnell, chief analyst at TECHnalysis Research.

The company’s adjusted gross margin is expected to be approximately 55%, in line with the expected 55.27%, but slightly down from 56% in the third quarter.

https://www.reuters.com/graphics/AMD-RESULTS/STOCK/jnpwodqezpw/chart.png

AMD has accelerated the launch pace of its AI products, expanding from standalone chips to full rack-level systems to compete with Nvidia.

The company’s flagship Helios rack-mounted system—which combines the MI455X accelerator with the new “Venice” server chip—has now entered production, with plans to significantly ramp up capacity in the fourth quarter. The first systems are being tested at Meta META.O.

In February this year, the company agreed to sell up to $60 billion worth of AI chips to Meta (link) over five years. According to the agreement, the Facebook parent company could acquire up to a 10% stake in the chipmaker.

AMD’s third quarter revenue rose 45% to $13.38 billion, surpassing the expected $13.09 billion.

Adjusted earnings per share were $2.09, compared to the market expectation of $1.95.

The semiconductor industry is also facing a global memory shortage. Although manufacturers are racing to expand capacity, high-bandwidth memory for AI processors remains expensive and in short supply.

AMD’s client and gaming business segment (covering its PC chips and game console processors) saw an 8% drop in third quarter sales to $3.72 billion, below the expected $3.79 billion.

In addition to its chip business, in September AMD agreed to acquire World Labs—a startup founded by Fei-Fei Li focused on building 3D environment modeling AI systems— for $8.2 billion in stock (link), thus establishing a foothold in the field of physical AI system research.

https://www.reuters.com/graphics/AMD-RESULTS/REVENUE/zdpxeblwnvx/chart.png

(For the convenience of non-English native speakers, Reuters has automated its reports into several other languages. Due to potential errors in machine translation or lack of necessary context, Reuters does not guarantee the accuracy of the translated text, which is provided only for readers’ convenience. Reuters assumes no responsibility for any harm or loss arising from the use of the automated translation function.)