An interview with analyst Benedict Evans on the role of productization in AI adoption, AI bubble, OpenAI, comparing Nvidia to Sun Microsystems, and more
In this conversation, Ben Bajarin and Jay Goldberg engage with Benedict Evans to explore the current state of AI development, its historical context, and future predictions. X: @hobie_tweets , @howardlindzon , @buccocapital , and @benbajarin X: Hobie / @hobie_tweets : This is correct. Them not winning Apple, and they getting out-executed by Anthropic in enterprise might be their undoing Howard Lindzon / @howardlindzon : supply is coming ... @buccocapital : “What Sam Altman is desperately doing is swapping paper for assets and distribution and product and people before the music stops.” - @benedictevans [image] Ben Bajarin / @benbajarin : This was one of my favorite insights from @benedictevans on our latest episode of @T_h_e_Circuit Is AI a product or a prompt? Productizing AI, with use case-specific UIs, is the role of the entrepreneur. We agree that AI is essentially a feature of all software. [video]
Context & Ripple Effects
The discussion extends a recurring distinction in the coverage between AI as a destination product and AI embedded in existing software. Apple's earlier integration of generative AI across its products illustrated the latter approach, while broader platform strategies have differed over how tightly models, infrastructure and applications are combined.
It also shifts attention from model availability to commercialization: the value may accrue to entrepreneurs that turn general-purpose capability into specific interfaces and workflows. That complements the view that early AI deployments may be led by enterprise implementations rather than consumer novelty.
First-order effects
- AI startups are pressured to demonstrate use-case-specific products and interfaces, not just prompt access to broadly capable models.
- OpenAI and Nvidia are assessed less on technological prominence alone and more on whether their assets, distribution and ecosystem positions translate into durable product adoption; the Nvidia–Sun analogy explicitly raises that distinction.
Second-order effects
- Incumbent software vendors with established customer touchpoints gain an advantage if AI adoption is driven by workflow integration, reinforcing the distribution logic visible in competing AI integration and modularization strategies.
- Model providers and infrastructure suppliers face more pressure to support builders that package AI into repeatable products, since raw capability becomes less differentiated at the application layer.
Third-order effects
- If productization remains the adoption bottleneck, AI competition should increasingly sort around workflow ownership, distribution and implementation rather than around standalone model access.
- The Sun comparison signals a broader caution: leadership in a pivotal technology layer need not determine where long-term application and customer value concentrates.
The trend: AI is moving from a model-led race toward a commercialization phase in which workflow-native products and distribution determine adoption.