From Gavin Baker on Koyfin's "Investing Wizards" series. · · Gavin Baker
“Digital processor markets, the barriers to entry are really underestimated. Code gets optimized for a specific flavor of CPU, GPU, or baseband, and that gives somebody said that it was an exorbitant privilege that the United States could borrow in its own currency. You know, the number one digital processor player enjoys the exorbitant privilege of having code optimized for their own specific architectures.”
On , Gavin Baker, Managing Partner and Chief Investment Officer at Atreides Management, LP, spoke about semiconductors during Gavin Baker on Koyfin's "Investing Wizards" series. on Gavin Baker.
Gavin Baker, Managing Partner and CIO of Atreides Management, has been active in public discussions on artificial intelligence, space infrastructure, and semiconductor supply chains. In June 2026, he interviewed SpaceX CFO Bret Johnsen at Mission Control, where they discussed the importance of launch capabilities and Starship's potential for rapid reusability. Baker noted that SpaceX is "the lowest cost per kilogram to space ever in the industry" and that Starship aims for "another 10x improvement." He also highlighted AI compute as a market that "really needs Starship to really happen" due to large payload requirements and cost focus. In May 2026, at the Sohn Investment Conference, Baker described the current AI environment as "the most extraordinary moment in the history of capitalism," citing Anthropic's rapid revenue growth. He discussed the role of TSMC's conservative capacity expansion in preventing a bubble, stating that "flinty old men and women who are safeguarding Morris Ching's legacy are helping us all avoid a bubble by enforcing a real-world physical constraint." Baker also emphasized the importance of Model Flop Utilization (MFU) as a key metric, arguing that "if you take your utilization from 40 to 80, you've dropped your cost per token in half." He joined the board of Aria Networks in April 2026, where he discussed the critical role of networking in AI factory efficiency.