On Tuesday, Advanced Micro Devices Inc. (NASDAQ:AMD) CEO Lisa Su said that the company's latest AI chip, the MI355, is ramping up production and gaining traction among customers seeking powerful and scalable alternatives to Nvidia Corporation's (NASDAQ:NVDA) high-end GPUs.
During AMD's second-quarter earnings call, Su said that the company began producing its MI355 chips in June and has seen "strong ramp" heading into the third quarter. She said customers are showing growing interest in AMD's data center GPUs, positioning the MI355 as a direct competitor to Nvidia's B200 and GB200 chips.
"What we're seeing with MI355 is very competitive versus the B200, GB200 family of products," Su said. "There's a strong desire to really use us at scale."
Su noted that AMD's previous AI chip, the MI300, had smaller deployments initially. However, with the MI355, she said, demand is broader — driven by the chip's performance in inferencing and training workloads, two key pillars of AI development.
See Also: July's 20 Most-Searched Tickers On Benzinga Pro — Where Do Opendoor, Nvidia, Apple Stock Rank?
Su added that the company expects continued revenue growth from its data center GPU business throughout the second half of the year.
"We're bullish on MI355 and where the AI opportunity is for us," she said. "And I think we're right on track to what we expected."
AMD sees the MI355 as a stepping stone toward its next-generation MI400 series, set to launch in 2026.
As Nvidia struggles in the AI chip market due to U.S. export bans and supply shortages, AMD has emerged as a viable challenger — especially in India and the UAE.
Nvidia's H100 chips, priced up to $40,000, remain the go-to for tech giants like Microsoft Corporation (NASDAQ:MSFT) and Meta Platforms, Inc. (NASDAQ:META). However, AMD's MI300 series offers a more affordable and power-efficient alternative.
The MI300X features 192GB of memory — double that of Nvidia's H100 — and the MI350X is expected to cost around $25,000.
In June, AMD unveiled its Instinct MI355X and MI350X GPUs at ISC 2025, showcasing significant performance gains for AI inference, particularly with support for low-precision formats such as FP4 and FP6.
Based on the CDNA 4 architecture, the chips feature 288GB of HBM3E memory with up to 8 TB/s bandwidth, reported Tom’s Hardware.
The MI355X, designed for liquid cooling, consumes up to 1400W and delivers 20.1 PFLOPS in FP4/FP6 tasks, while the air-cooled MI350X consumes 1000W and hits 18.45 PFLOPS.
Both surpass Nvidia's Blackwell B300 on paper but are limited to eight-GPU scaling, which may affect competitiveness. Real-world performance remains to be seen, the report added.
For the second quarter, AMD reported revenue of $7.69 billion, beating analyst estimates of $7.41 billion. However, adjusted earnings came in at 48 cents per share, narrowly missing expectations of 49 cents.
Segment-wise, data center revenue rose 14% year over year to $3.2 billion. Client and gaming revenue surged 69% to $3.6 billion, while embedded revenue declined 4% to $824 million.
AMD expects third-quarter revenue of approximately $8.7 billion, plus or minus $300 million, ahead of analyst forecasts of $8.15 billion.
Price Action: AMD shares have climbed about 44.50% year-to-date. However, the stock declined 6.34% in after-hours trading, falling to $163.25 at last check, per Benzinga Pro.
Benzinga’s Edge Stock Rankings indicate that AMD maintains strong upward momentum across short, medium and long-term periods. Additional performance details are available here.
Read Next:
Photo Courtesy: jamesonwu1972 On Shutterstock.com
Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.
Contact Us
Contact Number : +852 3852 8500Service Email : service@webull.hkBusiness Cooperation : marketinghk@webull.hkEnglish