Microsoft’s Homegrown AI Models, Trump’s AI Executive Order, OpenAI to Merge Codex & ChatGPT

Microsoft’s recent Build Conference signals a strategic shift toward cost-effective, in-house AI models and specialized hardware, including new coding and reasoning models alongside a "DevBox" desktop unit. While Marvell gains market attention following NVIDIA’s endorsement, the semiconductor landscape remains competitive, with Intel’s "Crescent Island" inference chip attempting to challenge incumbents through lower-cost memory architectures. Meanwhile, the White House’s latest AI executive order establishes a voluntary framework for model safety, reflecting a delicate balance between fostering innovation and addressing cybersecurity risks. OpenAI is consolidating its product suite by integrating Codex into ChatGPT to enhance agentic capabilities for knowledge workers. Simultaneously, Meta faces internal friction as employees push back against the "Model Capability Initiative," a tracking tool designed to train AI models by monitoring workplace computer activity, highlighting broader industry tensions regarding the use of employee data for AI development.