Photo Credit: anirudhlv from Pixabay
 

AMD’s (NYSE: AMD) recent second quarter earnings failed to impress the market. But AMD is working to improve its offerings that go head-to-head with Nvidia.

 
AMD’s Financials

AMD’s second revenues grew 32% to 7.69 billion, ahead of $7.42 billion expected by the market. Adjusted EPS came in at $0.48, missing the $0.49 market estimate.

By segment, Data Center segment revenues grew 14% to $3.2 billion, driven by strong demand for AMD EPYC processors that helped offset headwinds impacting AMD Instinct MI308 shipments to China.

The chip export ban by the US for export of chips to China has hurt the semiconductor industry. AMD is no different. It claims that its AI business revenue declined over the year due to the restrictions. AMD was not allowed to ship the MI308 chip to China, which cost it $800 million in the previous quarter.

However, as the US government has eased some of the restrictions, AMD expects to gain approval to resume shipping. Like Nvidia, AMD has also agreed to pay 15% of its revenues from sales to China to the U.S. Government.

Client and Gaming segment revenues grew 69% to $3.6 billion. Embedded segment revenues fell 4% to $824 million.

For the current quarter, AMD expects revenue to be approximately $8.4-$9.0 billion, compared with the analyst expectations of $8.3 billion.
 

AMD’s Growing Prowess

AMD is expected to be a big winner, benefiting from companies looking to diversify their reliance on Nvidia. Meta and OpenAI are looking up to AMD for its GPUs to reduce reliance on Nvidia’s offerings. Its latest chip, the Instinct MI350 range, already competes with Nvidia’s GB200 chips for training and inference. AMD is also offering the chip for rent from cloud providers starting this third quarter.

AMD believes that companies are looking for chips that can meet the growing need for inference to deploy chatbots or gen-AI applications. Inference uses more processing power than traditional server applications, and its MI350 series chips are equipped with more high-speed memory, allowing for bigger AI models to run on a single GPU. The MI355X has seven times the amount of computing power as its predecessor, and it competes with Nvidia’s B100 and B200 chips. AMD claims that its Instinct chips have already been adopted by seven of the 10 largest AI customers, including OpenAI, Tesla, and Cohere. Oracle is also planning to offer clusters with over 131,000 MI355X chips to its customers.

Additionally, AMD announced details about its next-generation AI chips, the Instinct MI400 series, expected to be released next year. Pioneering the rack-scale system, AMD will allow for the assembly of MI400 chips into a full server rack. The technology will allow the integration of chips in a manner that it could still be used as a single system. Essentially, while the end user will see the rack-scale system of chips as a single system, the chip set will allow for advanced workloads needed for AI customers for their LLM models.

AMD’s rack-scale technology will help it compete with Nvidia’s Blackwell chips, which come in configurations with 72 GPUs stitched together. With the release of some of its newer chips, including the MI400 series, AMD is looking to compete more aggressively with Nvidia both on capability and price.

Gaining market share from Nvidia is not going to be easy. According to a recent research, Nvidia accounted for 92% of the add-in board GPU market in the first quarter of 2025, an 8.5% increase compared to its previous position. By contrast, AMD’s share fell 7.3 points to just 8% during the same time.

The recent Nvidia and Intel deal is also not going to fare too well for AMD. Prior to the deal, AMD had the advantage of offering a CPU and GPU combination that Intel and Nvidia did not. The Nvidia deal will end that competitive advantage.

AMD’s stock is currently trading at $159.79 with a market capitalization of $259.3 billion. It was trading at a 52-week high of $186.65 earlier last month and has recovered from the 52-week low of $76.48 in April.

More By This Author: