🤖AI Newsletter
September 6, 2026 · 10:32 Uhr
1Polymarket: Anthropic leads model race with 84% probability
Polymarket With $1.9 million trading volume and 84% probability, Polymarket sees Anthropic clearly ahead of OpenAI (14%) and Google (3%) as the best AI model by end of September 2026 – momentum continuing to rise in the past week. This market assessment is remarkable as it remains unchanged despite GPT-6 Astra launch, and Anthropic's Fable model offers users concrete benefits with 45% cost savings. For enterprise customers, this signals a possible paradigm shift away from OpenAI as the default provider.
2Gemini 37 Flash: Google DeepMind attacks inference market
@UndercodeUpdate / aireleasetracker.com Google DeepMind has released Gemini 37 Flash as a cost-optimized, agentic coding model, directly targeting the booming AI inference market. The model is explicitly designed for low costs with high agent performance, competing directly with both Anthropic's Fable and GPT-6 Astra. Given Google's infrastructure advantages, this could significantly increase price pressure in the enterprise segment.
3AI infrastructure boom: Gartner sees 96% IaaS growth to $42 billion
Gartner / @IOPn_io Gartner forecasts 96% growth in AI-optimized cloud infrastructure for 2026, reaching $42 billion – infrastructure is now the limiting factor for AI adoption in the enterprise sector. Jensen Huang and investors like JasonL Capital see AI-infrastructure plays ($IREN, $MRVL, $MU) as the true winners of the AI cycle, while model providers face margin pressure. This infrastructure bottleneck is both a strategic opportunity and geopolitical risk, as computing capacity becomes a critical resource.
4Anthropic IPO: Market capitalization of $1.8 trillion priced in at 80%
Polymarket New and significant versus previous reports: Polymarket now prices in with 80% probability that Anthropic will reach a market capitalization of at least $1.8 trillion on IPO day – an increase of 40% within one month. This would instantly make Anthropic one of the world's most valuable technology companies and would have massive implications for valuations across the entire AI sector. Combined with the October IPO window (63% probability), the picture of an impending IPO of historic dimensions is becoming clearer.
5Productivity paradox: 84% of developers use AI, gains only 10-30%
secondtalent.com / @AllenTanCheeHoe Despite near-universal AI tool adoption among developers (84%), measurable productivity gains remain limited to 10-30% according to new data – McKinsey confirms: 80% of employees see individual gains, but only 37% of companies measure significant EBIT impact. The paradox points to a fundamental implementation and organizational problem that tool adoption alone cannot solve. For startups like Tenor (YC S26) and platforms like Cursor, this creates a multi-billion dollar market for the next layer: AI orchestration at the enterprise level rather than individual tools.
Situation Report
The AI industry is at a strategic turning point in September 2026: Anthropic solidifies its leadership position in model rankings despite the GPT-6 Astra launch and is on the verge of an IPO with a potentially $1.8 trillion valuation, which would recalibrate the entire sector. At the same time, the real competition is shifting from model quality to infrastructure capacity – Gartner data shows 96% IaaS growth, and computing capacity is becoming a scarce, geopolitically relevant resource. The productivity paradox – high tool adoption with low enterprise gains – signals that the next value creation layer lies in AI orchestration and integration, not in further base models. Escalation risks exist primarily in the infrastructure bottleneck and in the increasing concentration of AI power among few, well-capitalized players in the US and China.
Tokens: 2,453(1,498 in · 955 out)
