OpenAI Enterprise & Partnerships: Codex Expansion, Superapp Pivot, ChatGPT Ads, Tokenmaxing Backlash, Brockman's Candid Retrospective, Cognizant Cyber Deal, Codex Reliability Bug
Key Questions
How many weekly users does Codex have and what is the growth trend for non-developers?
Codex has 5M weekly users, with non-developer growth occurring 3x faster than developer adoption. This positions it as a core OpenAI product expanding beyond traditional coding audiences.
What is OpenAI's superapp pivot and how does it relate to ChatGPT redesign?
OpenAI is pivoting toward a superapp model confirmed by FT, with a redesigned ChatGPT integrating coding, agents, and partner apps into a unified experience. This aims to create a 'super assistant' for agentic task execution reaching 1B monthly users.
What new advertising plans does OpenAI have for ChatGPT?
OpenAI is introducing ads in paid ChatGPT subscriptions and has opened a self-serve ad platform, while hiring ex-AWS VP Matt Wood to support this. It also seeks engineers for image, video, and conversational ad formats.
What partnerships has OpenAI announced recently with enterprise and payment companies?
OpenAI invested $150M in its Partner Network, partnered with Visa for agentic payments, and works with HP to deploy OpenAI Frontier with measurable enterprise metrics. Getty Images is another key collaboration.
What is the tokenmaxing backlash and how are companies responding?
Palantir's Karp criticized OpenAI's token model publicly, fueling backlash as Accenture works to limit small-task AI usage. This reflects growing enterprise concerns over inefficient token consumption.
What is the Cognizant partnership with OpenAI for cybersecurity?
Cognizant partnered with OpenAI on the Daybreak program using GPT-5.5 Trusted Access for Cyber to validate and deploy vulnerability fixes. It focuses on operationalizing frontier AI for cyber defense.
What reliability issues have been reported with GPT-5.5 Codex?
Multiple users report GPT-5.5 Codex short-circuiting to exactly 516 thinking tokens about 40% of the time, often returning incorrect results. This has raised new concerns about quality degradation and enterprise trust.
How is Codex evolving competitively against Microsoft products?
Codex is becoming a desktop Work OS and agent chassis for OpenAI's superapp pivot, directly competing with Microsoft Copilot. Brockman noted plugins failed due to immature models and envisions a future with almost no interface.
Codex core product with 5M weekly users, non-developer growth 3x faster. FT confirms superapp pivot. ChatGPT reaches 1B monthly users; OpenAI building 'super assistant' for agentic task execution. Redesigned ChatGPT integrates coding, agents, and partner apps. OpenAI drops $150M on Partner Network. ChatGPT market share slips below 50%. Pentagon testing OpenAI models. Ads in paid ChatGPT subscriptions; self-serve ad platform open. Hired ex-AWS VP Matt Wood. Amazon dropped Altman biopic 'Artificial' amid $50B deal. Getty Images partnership. Palantir's Karp publicly bashes OpenAI's token model on CNBC, reinforcing tokenmaxing backlash. HP deploying OpenAI Frontier with concrete metrics. Visa partners for agentic payments. Tokenmaxing backlash: Accenture scrambling to curb small-task AI usage. Cognizant partners with OpenAI for frontier AI cyber defense via Daybreak program, using GPT-5.5 Trusted Access for Cyber to operationalize vulnerability remediation. Brockman candidly admits Plugins failed because models weren't ready; envisions 'almost no interface' future, aligning with superapp pivot but contradicting current product reality. New reliability concern: GPT-5.5 Codex reported short-circuiting to exactly 516 thinking tokens ~40% of the time, returning wrong results. Multiple users corroborate quality degradation. Could erode enterprise trust and fuel tokenmaxxing critique. Codex is evolving into a desktop Work OS, becoming the agent chassis for the superapp pivot, directly competing with Microsoft Copilot.