ChatGPT reached 100 million users in 2 months. No consumer technology has moved that fast. Benedict Evans uses that number as his anchor point, then pulls back to ask the harder question: speed of adoption is not speed of transformation. The internet took 20 years to move 20% of US retail online. The iPhone sold only 5.4 million units in its first 12 months. Cloud computing is still only roughly one-third of enterprise workflows, 25 years after Marc Benioff first pitched software in a browser.
Evans argues that generative AI has cleared the awareness hurdle faster than anything before it, but that the infrastructure, behavioral, and organizational conditions for real transformation are still being built. The interesting tension in this piece is not whether AI matters. It is whether the unprecedented speed of initial adoption tells us anything useful about the speed of downstream economic change, or whether we are pattern-matching on the wrong metrics.
Read the full piece for the data Evans uses to map adoption curves across past platform shifts, and for his specific framing of where enterprise AI deployment actually stands today versus the hype cycle. The argument is more calibrated than the headline suggests, and the charts are doing real analytical work.
[READ ORIGINAL →]