AI model rivalry and infrastructure pressures dominate NLP week

AI competition intensified as Anthropic reportedly reached a US$1.2T valuation ahead of an expected IPO, while Apple released a public beta of iOS 27 with a more capable Siri for iPhone 15 Pro users. Chinese open-weight models surpassed US models in downloads, reflecting developer demand for cheaper and more customizable systems. Google remained months behind on its flagship Gemini Pro upgrade, and Thinking Machines Lab released Inkling for enterprise customization.

Enterprise AI shifted toward cost control, governance and task-specific model selection as inference expenses and agent workloads grew. OpenAI launched ChatGPT Work for multistep office automation but acknowledged issues with limits, interface changes and messaging. Microsoft, Google, Meta and others expanded AI features across workplace, search, browser and consumer products, while regulators in the EU, China, Australia and the US advanced new rules and oversight proposals.

Infrastructure and risk dominated the broader agenda. AI data center investment accelerated alongside power, chip and environmental concerns, while community opposition delayed or blocked large projects. Security reports highlighted agent hijacking, prompt injection, malware generation and misuse by state-linked groups. Funding remained active across AI infrastructure, voice, legal, health and compliance startups, underscoring continued investor appetite despite growing questions about costs and returns.