OpenAI expands its AWS partnership: managed agents, ChatGPT in Slack and Teams
OpenAI will run managed agents on AWS and bring ChatGPT to Slack and Teams without individual licenses, with a private-inference preview coming this fall.
Per 财联社, OpenAI announced an expanded AWS partnership at DevDay: managed AI agents will run on AWS infrastructure, and ChatGPT arrives in Slack and Microsoft Teams with no individual license required. Also announced the same day: a $500 Pro tier with Ultrafast processing in ChatGPT and Codex, and a private-inference preview this fall combining confidential computing with “strict, verifiable controls.” OpenAI says 35 million people per week now build through ChatGPT Work and Codex.
The facts
- AWS: managed AI agents run on AWS infrastructure; OpenAI has a dedicated partnership page live.
- Distribution: ChatGPT in Slack and Microsoft Teams, no per-seat personal license needed.
- Tiers: the $500 Pro tier includes Ultrafast processing in ChatGPT and Codex.
- Private inference: preview this fall, pairing confidential computing with verifiable controls.
- Sourcing note: the newsflash is attributed to 财联社; the “35 million weekly” figure is OpenAI’s own and unverified. Don’t conflate this with the November 2025 $38B multi-year AWS agreement — today’s news is a capability expansion.
Our take
Read this with Dots and GPT-6.1 Sol: model commoditization (Sol), agent productization (Dots) and channel de-dependency (Slack/Teams/AWS) all happened in one DevDay — OpenAI is rewriting “sell subscriptions” into “sell an enterprise platform.” For procurement, ChatGPT inside Slack and Teams without per-seat licenses changes the seat-cost math directly; for the cloud race, a second major cloud is now formally in after Microsoft’s exclusivity ended.