AI Newsletter

Mistral CEO Arthur Mensch accuses OpenAI and Anthropic of instrumentalizing the AI safety debate to obscure their own technical weaknesses. Simultaneously, he announces a new Mistral model that outperforms Chinese competitors in some areas. The attack escalates the culture war between open and closed AI approaches to a new, politically sensitive level.

With 1.4M USD trading volume, Polymarket prices in an Anthropic IPO by end of October 2026 at 80% probability – despite a sharp 61% decline over the month. Anthropic reportedly warned IPO investors explicitly about existential risks from AI, which is fueling valuation debates. A successful IPO would recalibrate the entire AI sector and set benchmarks for all private labs.

At 1.5M USD volume, OpenAI's market share in the bet for best AI model by end of October collapses to just 1% – a plunge of 56% within a week. Google leads with 66%, Anthropic holds 32%. The drastic shift signals the market views GPT-6.1 Sol as significantly weaker than Gemini 4 and Claude Opus 5.5 – a strategic setback for OpenAI after DevDay.

Meta's new personal agent app Muse has displaced ChatGPT from the top of the Apple App Store, while OpenAI counters with 'Dots' and Google with Gemini 4. Wall Street is shifting valuation focus away from base models toward concrete agent products for consumers and enterprises. This race for the interface layer could become more decisive for market share than pure model benchmark competition.

Wall Street expects S&P 500 earnings growth of 32% for 2026, directly fueled by AI investments and efficiency gains. Nearly half of all companies report measurable ROI from AI (BCG), while AI spending has risen to 3.3% of total revenue. The macro narrative is shifting: AI is no longer a cost center but the primary driver of results – with direct implications for valuation models.