[Major U.S. Companies Begin Using Affordable AI Collectively: Mentions of Open-Source Models Surge 6x Year-on-Year] Beating AI News Flash: A Financial Times investigation has revealed that open-weight models are spreading from tech companies to traditional U.S. enterprises. According to AlphaSense data, in August and September of this year, mentions of open-weight or open-source models during earnings calls and investor meetings by U.S. companies increased approximately sixfold year-on-year. PNC Financial, logistics company CH Robinson, and Siemens have recently publicly discussed related applications. The most direct reason for companies switching models is the cost. Tinder's annualized AI expenditure rose from about $1 million in January this year to approximately $10 million in July, and it has now started delegating some routine requests to open-weight models. AT&T's scale is even larger. Currently, about 40% of its AI tasks have been handed over to open models, and the company plans to further increase this proportion to around 70%. Its internal AI system processes approximately 45 billion tokens daily, with simple tasks assigned to affordable models and complex tasks continuing to use cutting-edge models like OpenAI and Anthropic. [Original Link]