The article explores why OpenAI and Anthropic focus on selling tokens rather than launching their own application-based companies. Commenters argue that while frontier models are powerful, building and managing products requires significant human involvement and capital. One key point is that these companies need revenue quickly due to tighter access to capital from rising borrowing rates. Additionally, some believe AI will exhibit Jevons Paradox — lower prices driving higher demand. The models still have limitations, making it safer for others to discover these on their own dime. OpenAI has a $4B revolving credit line but has barely used it, and they haven’t issued bonds. Despite rumors, there’s no clear evidence they’re committed to $500B in Oracle compute. Meanwhile, their Forward Deployed Engineers work inside existing companies using subscriptions. Selling long-running agents remains the most reliable revenue strategy for now.
Another angle is that OpenAI is already experimenting with applications through its product portfolio, though at scale this would mean launching 100,000 SaaS products — a risky and costly endeavor. The rush toward AGI means these companies are burning cash, and any new ventures add debt and uncertainty. Until they can amortize past investments, focusing on token sales and enterprise integrations makes more financial sense.
The debate reflects broader questions about monetization strategies in the AI industry as commoditization pressures mount and the gap between model capability and practical usability widens.
Source: Hacker News · Summarized by HeadlinesBriefing