At AMD’s recent “Advancing AI” event in San Jose, we had the opportunity to speak with Mahesh Balasubramanian, AMD’s Director of Data Center GPU Marketing. We previously interviewed him in 2023 when AMD was rolling out the AMD Instinct MI300X AI Accelerator, so it was good to reconnect and learn more about AMD’s advancements upcoming from the AMD Instinct MI350 to MI500 chips due out in 2026.
AMD is engaged in a long-term and high-stakes struggle to gain a bigger foothold in the AI compute market dominated by NVIDIA, which has a 90+ percent share. AMD has adopted an aggressive message to the market, saying “Only AMD powers the full spectrum of AI, bringing together leadership CPUs, GPUs and other accelerators, networking, and open software to deliver unmatched flexibility and performance.”
Speaking with Balasubramanian, he updated us about the AMD MI350 series, highlighting high-precision FP64 performance, critical for HPC, AI, engineering simulations and financial modeling. He said by early summer, the company will issue a whitepaper providing “a comparative analysis of native vs. emulated PF64 performance, emphasizing AMD’s technical leadership and commitment to providing robust HPC solutions.”
Balasubramanian emphasized three key components of AMD’s AI go-to-market strategy:
- **Performance:** Offer the most advanced hardware solutions in the market, ensuring customers receive top-tier performance.
- **Ease of Deployment:** Enable “effortless” adoption through open ecosystem solutions, simplifying implementation of AMD hardware and software.
- **Economics:** Build AI chips that allow customers to generate higher margins, with the MI355X delivering up to 40 percent more tokens per dollar compared to Nvidia’s Blackwell B200 GPU, according to AMD.
Balasubramanian also emphasized AMD’s success will depend heavily on strong execution and customer collaborations with partnerships such as OpenAI and Meta, positioning AMD at the forefront of AI, he said.
In her conference keynote address at Advancing AI, AMD CEO Dr. Lisa Su (who was joined on stage by OpenAI’s Sam Altman and Meta’s Yijin YJ) said the company’s guiding principle is: “Be flexible so that you are ready to pivot as quickly as this marketplace.” She also highlighted the company’s commitment to an annual AI product launch cadence, a strategy NVIDIA also has pursued over the past several years.
