Station F Expands AI Accelerator to Strengthen European Startup Pipeline

Nations are building domestic AI infrastructure and capabilities as a strategic priority, developing locally trained foundation models, AI factories (next-generation data centers), and workforce programs to ensure AI solutions reflect regional needs and comply with local regulations. The article outlines five key ingredients for national AI strategies: AI imperative, workforce development, locally trained models and data, ecosystem building, and AI factories. This shift reflects growing recognition that AI is reshaping economies and requires countries to develop sovereign computing capacity and expertise.

SoftBank announced a U.S. neocloud venture on Thursday, adding to hundreds of firms now competing in the AI server rental market. Together AI raised $800 million at an $8.3 billion valuation, while Nvidia said it will provide financial backing to younger cloud firms in exchange for a revenue share. The moves highlight intense competition in the sector, though Nvidia's backstop offer raises questions about the actual strength of demand for computing capacity.

Anthropic is in early-stage talks with Samsung Electronics to manufacture a custom AI chip, according to sources with direct knowledge of the project. The move mirrors OpenAI's strategy of developing proprietary chips to reduce dependence on external computing infrastructure and control costs. Google, Amazon Web Services, Meta, and Microsoft have all developed their own chips, while OpenAI unveiled Jalapeno, an inference chip designed for large-language models, last month.

Model routers, which automatically select the most cost-effective AI model for a given task rather than defaulting to expensive cutting-edge options, are gaining adoption among enterprises seeking to reduce AI spending. Companies like Snowflake and Palo Alto Networks have reported cost savings by routing basic tasks such as email summarization and document search to cheaper open source or older proprietary models. The routers take multiple forms, from standalone products to cloud provider features to internal IT-built applications, all aimed at maintaining quality while lowering costs as organizations grapple with rising AI model prices and employee overuse of premium models.