Keep pulling the thread on Benedict Evans.
Agentic coding has achieved absolute product-market fit, with customers actively pulling the technology from developers' hands.
Benedict Evans predicts that foundation models will become commodity infrastructure sold at marginal cost, similar to mobile networks or ISPs.
Approximately $1 trillion to $2 trillion of CapEx is expected to be invested in the AI industry over the next couple of years.
AI models are becoming 100 to 200 times more efficient each year.
AI models lack the network effects that enabled operating systems like Microsoft Windows and Apple iOS to capture significant value.
Benedict Evans believes that foundation models and chatbots are not the end-product, and that economic value will be captured further up the application stack.
There is no clear path for a large language model to achieve a sustainable, fundamental competitive differentiation similar to the network effects of Instagram, YouTube, or Google Search.
Generative AI is fundamentally different from previous platform shifts because its physical limits and rate of improvement are currently unknown and unpredictable.
Benedict Evans predicts that a significant percentage of existing SaaS companies will be wiped out by the shift to AI.
Microsoft, Meta, and Google are projected to spend 50% of their revenue on CapEx in the current year.
The big four technology companies have provided collective guidance for $700 billion in CapEx for the current year.
OpenAI's strategy has shifted from an 'everything all at once' approach to a more focused strategy on coding.