OpenAI's confidential S-1, filed with the SEC in June 2026, prices ChatGPT's 500 million weekly active users as the primary deployable asset. Alibaba's Qwen API completed the same transition (consumer chat to enterprise per-token billing) through 2024, booking the resulting revenue against Alibaba Cloud's Singapore cluster expansion without a public disclosure event requiring it to value the retiring product and the replacement simultaneously. Ars Technica reported the same week that OpenAI's own product leadership has declared chat obsolete and is rebuilding ChatGPT around ambient agentic interaction, with no disclosed enterprise contract count and no published seat pricing in its new form. Alibaba faced that pricing problem in private.

The disclosure requirement is the structural constraint. OpenAI's S-1 must price the H100 and A100-trained ChatGPT base, built 2022-2024 and generating the consumer revenue the current valuation rests on, alongside the agentic surface replacing it, on the same document. Baidu's ERNIE Platform made the same consumer-to-enterprise migration in 2024 under internal budget allocation. OpenAI has B200 NVL72 hardware access that BIS October 2023 controls deny to PRC labs; the S-1 must price the agentic deployment built on that hardware, and that document is in the SEC pipeline as of June 2026. Baidu ran this transition without a public capital-markets disclosure requirement. OpenAI runs it with one.