AI's Evolving Economics: Harry Stebbings Highlights Shift from Traditional Margins to Future Value

Venture capitalist and 20VC podcast host Harry Stebbings recently sparked discussion on social media regarding the evolving relevance of traditional business margins in the age of artificial intelligence. Stebbings asserted that while margins remain "absolutely important," a clear vision for technological advancement allows companies to build for "the margins of the future," a crucial distinction in today's rapidly changing market. This perspective suggests a re-evaluation of immediate profitability metrics in favor of long-term strategic positioning within the AI landscape.

The sentiment echoes discussions from his podcast, where industry leaders have highlighted the intense competition and cost pressures within the foundational AI model space. Aidan Gomez, CEO of Cohere, previously described selling access to AI models as quickly becoming a "zero margin business" in an interview with Stebbings on 20VC. Gomez explained that "there’s so much price dumping. People are giving away the model for free," leading to very tight margins for companies focused solely on model sales.

This competitive environment forces a strategic shift, with many looking towards the application layer for sustainable revenue. The high compute costs associated with training and running advanced AI models, often referred to as a "GPU tax," significantly impact profitability. A recent SaaStr discussion involving Stebbings, Rory O'Driscoll, and Jason Lemkin noted that some SaaS companies are already seeing a 10% "GPU tax" on their revenue due to AI inference expenses.

Stebbings' emphasis on "the margins of the future" suggests that current low or negative margins in core AI infrastructure might be a necessary investment for future market dominance and value creation. Companies are betting on optimization and the eventual commoditization of foundational models, allowing for higher profitability in specialized applications and services built atop this infrastructure. This long-term view prioritizes strategic positioning and innovation over immediate, conventional margin expectations, reflecting the dynamic and capital-intensive nature of the AI industry.