AI is the most important technology of the last 50 years, and the industry is still in the very early stages of adoption, with current technology being 'primitive' compared to future possibilities.
A successful turnaround in the semiconductor industry requires long-term, focused bets on core strengths, such as high-performance computing, and the willingness to endure short-term losses for future market leadership.
An open ecosystem, built on open-source software and modular chiplet architectures, will ultimately attract a larger developer community and prove to be a superior long-term strategy.
The primary bottleneck to AI progress in the near term is not chip design but the supply chain for data centers and power, indicating a massive, long-term need for more infrastructure.
It is a strategic necessity to increase semiconductor manufacturing in the United States for supply chain resiliency, but this is a complex, long-term process that cannot be achieved overnight.
2012
Joined AMD when the company was over 90% reliant on the PC market. Her leadership team determined the server roadmap was uncompetitive and made the decision to scrap it and start over.
October 2014
Appointed CEO of AMD when the company was near bankruptcy, with a share price around $2. Communicated to the board that the server turnaround would be a five-year project.
2014-2017
During the first three years of the server strategy reset, AMD's market share in the sector declined from 10% to a low of 1%.
~5 years ago
Engagement with policymakers in Washington D.C. was infrequent, occurring only once or twice a year.
Last Year
Significantly increased engagement with policymakers, meeting with them 12-13 times over the year, reflecting the growing strategic importance of the semiconductor industry.
Last Fiscal Year
AMD's server market share finished the year at slightly over 40%, marking the success of the long-term strategy initiated years earlier.
▶Long-Term, High-Risk Strategic BetsApr 2026
A core theme is Su's willingness to make difficult, long-term strategic decisions that may cause short-term pain. This is exemplified by scrapping AMD's uncompetitive server roadmap upon joining, telling the board it would take five years and three product generations to fix, and enduring a market share drop from 10% to 1% before seeing success.
This demonstrates a leadership style that prioritizes long-term technological superiority and market positioning over short-term financial results, a crucial trait in the capital-intensive, long-cycle semiconductor industry.
▶The Inevitable AI RevolutionApr 2026
Su consistently frames AI as the most significant technological shift in over 50 years, believing the industry is only in the 'very, very early beginnings' of its adoption. She sees AI becoming pervasive, from massive data centers to personal devices, and believes scaling laws will continue to drive its advancement.
Her conviction in AI's foundational importance underpins AMD's entire corporate strategy, justifying massive investments in specialized compute hardware and positioning the company to compete directly with established leaders like NVIDIA.
▶Navigating Semiconductor GeopoliticsApr 2026
Su's commentary reflects the complex geopolitical landscape of the semiconductor industry. She acknowledges Taiwan's current dominance in advanced manufacturing while being a 'big proponent' of bringing more production to the U.S. through initiatives like the CHIPS Act, all while needing to maintain access to the 'very important' Chinese market.
This highlights the delicate balancing act CEOs in critical tech sectors must perform, aligning with national security interests while serving a global customer base and managing a geographically concentrated supply chain.
▶Competitive Strategy via Open Ecosystems
AMD's strategy under Su involves deliberately choosing open platforms to foster growth and competition. This is evident in the shift to a chiplet-based architecture, which allows for more modularity, and the commitment to an open-source software strategy, which aims to attract a larger developer ecosystem than a closed, proprietary approach.
This positions AMD as a collaborative alternative in the market, potentially lowering the barrier to entry for developers and partners, which could be a key differentiator in the long-term battle for AI market share.