Stanford MS&E435 Economics of the AI Supercycle | Spring 2026 | Infrasctructure, Enterprise AI, SaaS

Organizations are struggling to capture AI productivity gains not because of a lack of intelligence in current models, but because they have failed to integrate internal business context and re-engineer their operational processes. Successful AI investment requires focusing on long-term, secular applications rather than chasing transient tech hype.

• Context gap — AI models often fail in enterprises because they lack the tribal knowledge (the "John and Jane" insight) employees keep in their heads [05:40].

• Productivity lag — Historical shifts, like the electric engine replacing steam, took 40 years to show economic impact because businesses had to fundamentally redesign their factories to use the new technology [18:32].

• Software reality — The industry is seeing lower barriers to entry and lower switching costs, which means incumbents must innovate their business models or risk being replaced by faster, cheaper competitors [09:12].

• Operational refactoring — Real gains come from re-engineering workflows, not just faster models; one Databricks team reduced a nine-month task to three months by simply changing how people and processes were organized [24:11].

• Investment horizon — True value typically accrues to applications and long-term secular trends (like healthcare or education) rather than the most popular tech trends of the moment, which often turn out to be overhyped "multicast" problems [26:44].

• How does the speaker define artificial general intelligence?

• How does the history of the electric engine explain current productivity challenges?