2025.41: It's OpenAI's World, We're Just Living in It (stratechery.com)

OpenAI is increasingly being framed as the "Windows of AI": not just a provider of models but a platform aiming to control both the software layer and the hardware/OEM ecosystem beneath it. Sam Altman’s recent interview and company signals make clear that OpenAI is placing “company-scale” infrastructure bets now to meet anticipated model demand, aiming to centralize developer and consumer activity on its stack. That positioning has major implications for the AI/ML community — it could reshape where investment flows (turning speculative infrastructure spending into what some characterize as near‑trillion-dollar opportunities), concentrate margin and influence up and down the supply chain (even over chip vendors), and standardize the model stack across consumer and enterprise uses rather than bifurcating them. At the same time, product moves like Sora underscore a behavioral shift from AI-enabled consumption to AI-enabled creation: early metrics (roughly 30% of active users becoming creators) suggest rapid adoption once tooling becomes easy and social. For researchers and practitioners this means rising demand for scalable, low-latency inference, robust content-generation safety and moderation, and APIs/SDKs that support creative workflows. If OpenAI succeeds as the dominant platform, developers, hardware partners, and downstream services will need to adapt to a more centralized model distribution and monetization ecosystem — changing where innovation, control, and economic value in AI accrue.