AMD stock slips after Q2 profit miss, but AI-driven $8.7 billion Q3 forecast sparks investor hope
Nillohit Bagchi | Aug 06, 2025, 03:32 IST
Highlight of the story: AMD posted a mixed set of numbers for Q2 2025, missing on earnings but beating revenue expectations. While export issues to China linger, the company’s strong AI forecast for Q3 suggests its long-term plans are on track. Investors now wait to see if AMD can finally close the gap with Nvidia.
AMD dropped its second-quarter report, and while the numbers didn’t scream ‘perfect,’ there was still enough to keep investors interested. The chipmaker, known for taking the fight to rivals like Intel and Nvidia, missed Wall Street’s earnings estimate by a whisker. But revenue came in strong, showing that demand across key segments hasn’t cooled off.
For Q2 2025, AMD posted adjusted earnings of 48 cents per share, just under the expected 49 cents. Revenue, though, stood tall at $7.7 billion, beating the $7.4 billion forecast. Year-over-year, that’s a 32% rise, driven by solid growth in personal computing and gaming.
AMD expects to bring in around $8.7 billion in revenue for Q3, give or take $300 million. That’s more than what analysts predicted, and it suggests confidence in its next-gen AI chips. CEO Lisa Su pointed to the rollout of the Instinct MI350 series, which AMD claims can rival Nvidia’s Blackwell chips in both training and inference.
One of the biggest challenges AMD faced this year was the temporary U.S. ban on exporting its MI308 AI chips to China. That move cost the company roughly $800 million in lost revenue and led to an operating loss of $155 million last quarter. Although the Trump administration has now relaxed those restrictions, AMD is still waiting for final clearance to resume shipments.
While AI grabs headlines, AMD’s traditional businesses are also pulling their weight. The gaming segment brought in $1.1 billion in revenue, a 73% increase year-over-year, driven by demand for custom console chips and GPUs. The client's business, which includes desktop and laptop CPUs, pulled in $2.5 billion, up 57% from last year.
Despite the upbeat forecast, AMD still trails Nvidia by a large margin in the AI chip race. Nvidia remains the go-to for training large models, thanks to its dominant GPUs and widespread developer support. But AMD is slowly building its case, especially in inference, where price and efficiency matter more.
AMD revenue beats estimates despite earnings slip
For Q2 2025, AMD posted adjusted earnings of 48 cents per share, just under the expected 49 cents. Revenue, though, stood tall at $7.7 billion, beating the $7.4 billion forecast. Year-over-year, that’s a 32% rise, driven by solid growth in personal computing and gaming.
AMD AI forecast shines a light on Q3
AMD expects to bring in around $8.7 billion in revenue for Q3, give or take $300 million. That’s more than what analysts predicted, and it suggests confidence in its next-gen AI chips. CEO Lisa Su pointed to the rollout of the Instinct MI350 series, which AMD claims can rival Nvidia’s Blackwell chips in both training and inference.
AMD expects to bring in a
China export ban: A costly delay for AMD
One of the biggest challenges AMD faced this year was the temporary U.S. ban on exporting its MI308 AI chips to China. That move cost the company roughly $800 million in lost revenue and led to an operating loss of $155 million last quarter. Although the Trump administration has now relaxed those restrictions, AMD is still waiting for final clearance to resume shipments.
Gaming and client segments push ahead
While AI grabs headlines, AMD’s traditional businesses are also pulling their weight. The gaming segment brought in $1.1 billion in revenue, a 73% increase year-over-year, driven by demand for custom console chips and GPUs. The client's business, which includes desktop and laptop CPUs, pulled in $2.5 billion, up 57% from last year.
While AI grabs headlines,
AMD is rivalling Nvidia in the AI segment
Despite the upbeat forecast, AMD still trails Nvidia by a large margin in the AI chip race. Nvidia remains the go-to for training large models, thanks to its dominant GPUs and widespread developer support. But AMD is slowly building its case, especially in inference, where price and efficiency matter more.