Benedict Evans, former a16z partner and independent analyst, argues AI is roughly equivalent in scale to the internet or mobile: transformative, but not magic. His core framing: we are in 1997 for AI. The infrastructure is real, the hype is real, and almost nobody knows which specific applications will win or which jobs will actually disappear.
The conversation cuts through two failure modes. First, panic: Evans rejects the 'what percent of my job can AI do' framing entirely, insisting the real question is whether your role is a task or a job, a distinction with concrete implications for who gets replaced versus who gets augmented. Second, dismissal: the anti-AI backlash is documented here, including a former Google CEO getting booed at a commencement speech, and Evans explains why rational argument rarely moves people already emotionally committed to a position. There is also a counterintuitive data point worth reading for in full: professional services and consulting at AI companies are booming, not collapsing, a direct parallel to how VisiCalc and spreadsheets expanded accounting work rather than eliminating it.
Where value accrues in the stack, how Google, Meta, Apple, and OpenAI are fighting the distribution war, and why distribution is becoming the primary moat as model capabilities commoditize are all covered in specific terms. Evans does not predict winners. He maps the terrain. That is the value here.
[WATCH ON YOUTUBE →]