OpenAI’s perception as having ceded the top position in AI with respect to frontier model development and consumer adoption forces a few realizations:
- Vertical integration is critical. Last month, I fine-tuned Gemini 2.5 with Vertex AI and deployed to an endpoint on GCP with a few command line inputs and clicks in the GCP web interface. The enterprise use case probably requires more functionality than mere API access; it has to provide the technical ligaments for integration and deployment.
- Distribution isn’t disrupted. Google and Apple continue to control the principal content dissemination points for consumers, and although recent regulatory / legal challenges have challenged total app store control (eg., third-party app stores and payments), Google emerged from its antitrust trials with Search primacy in tact. “Google’s Gambit” proved successful: AI Mode has 75MM DAU, AI Overviews has 2BN MAU, and Google and Apple remain the fundamental chokepoints for consumer discovery and engagement.
- Ads remain the dominant / inevitable business model for mass-market consumer products. A disciplined product development process needs to accommodate the ads monetization model from the earliest stages and manage the ads-related engagement / monetization tension as a product priority. Delaying the introduction of ads simply inflates the presumed consumer pushback ("enshittification", the fatuous notion that excess consumer surplus is beneficial in the long term).