AI Newsletter

OpenAI improves its most widely used model GPT-5.5 Instant through better context recognition and more precise interpretation of user intentions – an incremental optimization of the existing product without radical innovation. The update addresses user experience and competitiveness against competing LLM providers (Claude, Gemini), but could generate limited commercial leverage for revenue growth.

China's GLM-5.2 achieves near-equivalent performance to Western top-tier models (Claude Opus) at drastically lower costs (1/5 of the price), which significantly intensifies price pressure on Western AI providers like Anthropic. This could put Western AI business models under pressure and lead to market consolidation as cost-effective Chinese alternatives increasingly establish themselves as equivalent substitutes.

OpenAI reduces its dependence on NVIDIA and other chip manufacturers through in-house development of an inference chip with Broadcom, which reduces costs long-term and improves margin development. The "Jalapeño" chip addresses a core bottleneck in the AI industry and could significantly increase OpenAI's operational efficiency starting in 2026 as well as shake up the competitive landscape in the GPU/AI chip market.

OpenAI founds DeployCo, a specialized B2B subsidiary, to integrate AI models directly into enterprise structures – with its own engineering team and direct customer feedback loop for model improvement. This aims to dominate the high-margin enterprise market and reduce dependence on API-based access.

Anthropic integrates Claude directly into Slack workflows to provide enterprise customers with AI-powered productivity in their existing work environment. Internal usage (65% code generation) demonstrates product maturity and positions Anthropic as a competitor to OpenAI/Microsoft in the B2B AI market. This addresses the growing market for embedded AI solutions and could unlock significant revenue potential through Slack integration.