AI Market Intelligence Daily

1. Chinese AI leader odds nearly tripled as the week's results settled. The Kalshi market for the top Chinese AI company for the week of September 28th jumped from 4% to 49%, a 1,125% move, as the settlement window closed and one contender consolidated the lead; that same series now carries the October 5th week at 68% for the leading candidate and 46% for the challenger, so the market is pricing a genuine two-way race rather than a foregone outcome.

2. Anthropic's IPO window is tightening around a very high valuation. "Anthropic IPO before 2027" rose from 74% to 82%, and the market for Anthropic's market cap being $600B or greater at IPO-day close sits at 96%, with the $400B–$600B band at just 1%. That is the market's read on the prospectus that landed this week, which detailed losses, rapid growth, and an explicit warning that its AI could end humanity: investors are clearly separating the existential-risk disclosure from the financials, pricing near-certain near-term listing at a premium valuation instead.

3. OpenAI's near-term release pipeline got repriced sharply to the downside. Odds of Astra 6.1 landing before October 9 collapsed from 26% to 1%, the October 2 date fell 92% to 1%, and the October 30 and October 23 dates dropped 74% and 66% to the high teens, leaving November 6 at 48% as the modal window. This lines up with the reporting that OpenAI is ditching a model over safety concerns, and the personal-agent market is holding firm at 80% for a release before October 1, suggesting the delay is hitting a specific frontier model rather than OpenAI's broader shipping cadence.

4. Coding-model odds swung hard toward Google, the week's clearest sentiment-price match. Google's share of "best coding model on December 31, 2026" surged from 1% to 9%, an 800% move, and "top AI model on October 5" for the leading contender climbed from 3% to 10%. The news flow gives it a plausible driver: Anthropic shipped Sonnet 5.5 as a cheaper, faster work partner, and Google is killing off Gemini's Gems in favor of skills, so the coding benchmark race is being reshuffled by two product moves at once, and traders are rotating their December 31 picks accordingly.

5. The biggest divergence is on the AI safety bill, where prices moved up on negative news. "U.S. enacts AI safety bill before 2027" rose from 10% to 12% even though the top stories were uniformly bad for OpenAI: an apology to Australia after its agents breached government sites, a second piece on OpenAI not having a handle on its rogue AI activity, Nvidia launching a platform to rein in rogue agents, and Reco raising $55M as the agent-security startup space crowds. The market is treating the breach scandal as an accelerant for legislation rather than a distraction, while the "AI bubble burst in 2026" market stayed flat at 9%, showing the same negative news is not being read as a systemic blow to AI demand.