AMD is intensifying its push into the AI chip market with the launch of its new Instinct MI350 series and a string of strategic acquisitions aimed at challenging Nvidia's dominance. CEO Lisa Su unveiled the new MI350X and MI355X AI accelerators during a company event in California, emphasizing their competitive performance at a lower cost compared to Nvidia's flagship chips.

Su claimed the MI355X delivers up to 40% more performance per dollar than Nvidia's H100 and B200, with the new chips set to become available through cloud providers starting in Q3 2025. AMD also revealed its next-generation MI400 series is scheduled for release in 2026.

To strengthen its position, AMD has acquired six AI-related companies in recent months. These include:

 ● Brium, an AI inference compiler startup that enhances AMD's software stack efficiency.

 ● Enosemi, a silicon photonics specialist, to support next-gen optical interconnect solutions.

 ● Untether AI, whose engineering team will boost AMD's AI compiler, SoC, and digital design capabilities.

 ● ZT Systems, offering rack-level deployment expertise for large-scale AI infrastructure.

Previously acquired firms Nod.ai, Pensando, Silo AI, and Mipsology, all of which enhance AMD's AI software and hardware integration across the full stack.

These moves are part of AMD's strategy to become a “full-stack AI challenger” to Nvidia's CUDA-dominated ecosystem. AMD also introduced ROCm 7, its latest open-source AI development platform, aiming to chip away at Nvidia's software lead by offering flexibility and broader compatibility.

While Nvidia still controls over 70% of the global AI chip market, according to Omdia, AMD is targeting the AI inference market during the technology's mass adoption phase. Its chips have already been adopted by major players such as Meta, Microsoft, and Oracle, and the company claims to supply 7 of the top 10 AI-focused firms.

Despite the aggressive push, AMD faces an uphill battle. Nvidia's head start and entrenched developer ecosystem pose significant challenges. Still, with a growing hardware portfolio, an open-source software stack, and a string of acquisitions to bolster its capabilities, AMD is making a serious play for relevance in the next phase of the AI compute race.