AI Newsletter

Prediction markets with $2 million volume see Google with 74% as clear favorite for best AI model by end of October 2026 – an increase from 66% this month. Anthropic follows with 24%, OpenAI is practically irrelevant at 1%. This dramatic shift signals that Google's Gemini family has technically surpassed the competition and fundamentally shaken the previous power balance in the frontier model segment.

VC analyst Tomasz Tunguz predicts AI inference will grow from $160 billion to $350 billion, displacing databases as the largest software market. The catch: software margins could fall from typical 72% to 30–50% as compute costs burden the business model. This development forces SaaS companies to fundamentally reassess their pricing structures and cost models.

Mistral launched its most powerful to date, fully sovereign and green-energy-powered AI model – while simultaneously going literally into space: satellite Altair runs Mistral models as part of a planned 50-satellite AI constellation. In parallel, Mistral demonstrates it can deliver open-source models at OpenAI's level from nine months ago with significantly fewer resources, redefining cost efficiency of European AI.

McKinsey estimates the global agentic commerce market at $3–5 trillion by 2030, yet RevenueCat's 2026 report shows the brutal flipside: only 17.3% of all newly launched apps reach even $1,000 in monthly revenue within two years. Vertical AI agents with industry-specific solutions attract the largest funding rounds, while generic chatbot approaches are shunned by capital.

BCA Research warns of a massive $10 trillion AI revenue monetization gap – investments in AI infrastructure dramatically exceed measurable returns. Simultaneously, Nvidia's $500 billion financing apparatus ignites debate about systemic risks in the AI hardware ecosystem. This confirms McKinsey's finding that while 80% of companies report individual productivity gains, only 37% achieve real business results – a new dimension of this known problem.