The direction

Where the AI race is heading

AI competition is moving away from frontier models alone toward control of the stack that turns models into governed execution. That shift is being carried by agent environments tied to enterprise identity, data, workflow, security, and everyday work surfaces, and by model access that is increasingly routed through managed clouds, compatibility layers, owned product surfaces, and access controls. The infrastructure race underneath is widening into financing, datacenter capacity, packaging, custom silicon, and power. What is particular now is that the enterprise operating layer and the physical power layer are hardening at the same time, so the contest is increasingly about who can execute, where, and at what industrial scale.

The move

What actors actually did

OpenAI and Broadcom launch custom AI chip, Anthropic accuses Alibaba of model distillation

OpenAI and Broadcom introduced Jalapeño, a custom AI accelerator chip optimized for LLM inference and designed for gigawatt-scale deployment, with the announcement made on June 24. The chip promises significantly improved performance per watt and was developed within nine months, marking a move toward dedicated AI hardware for large-scale applications.

Anthropic publicly accused Alibaba on June 24 of performing the largest known distillation attack against its Claude AI model. Anthropic alleges that nearly 25,000 fraudulent accounts were used to extract model capabilities in over 28.8 million exchanges, raising concerns about AI model security and intellectual property.

OpenAI released an update to GPT-5.5 Instant in ChatGPT on June 24, aiming to improve conversational quality, context retention, instruction handling, and coherence across advice, planning, and shopping features. Meanwhile, Microsoft announced that its Digital Crimes Unit, working with Europol and industry partners, used AI-assisted analysis including Copilot to disrupt the Amadey and StealC cybercrime tools—disabling criminal control over more than 18,000 infected devices on June 24.

Sources