Now Playing
OpenAI's $1.25T Round, Microsoft's Oracle Exit & AI Safety's Broken Layer
Audio is available on Spreaker — see link below.
OpenAI closes an $11B funding round at a $1.25 trillion valuation while Microsoft quietly exits its Oracle GPU deal — and regulators are exposing a structural flaw in how AI safety actually works. Today's briefing covers capital concentration, compute strategy, and the accelerating shift of AI accountability to the deployment layer.
Audio is available on Spreaker — see link below.
OpenAI just closed an eleven-billion-dollar funding round at a one-point-two-five trillion dollar valuation. That number deserves a moment.
While OpenAI was raising capital, Microsoft was quietly cutting one. The company is winding down a compute rental arrangement with Oracle, where it had been leasing GPU capacity for Bing AI workloads.
There's a structural problem with how AI safety has been built, and a recent investigation makes it hard to ignore. The core issue is that model-level refusals, the systems that cause AI tools to decline certain requests, don't reliably work in either direction.
Regulators in Singapore, the EU, and the UK are now converging on a different approach. Rather than certifying model behavior, their frameworks require deployers to show evidence of risk assessment, lifecycle controls, and named human accountability.
That regulatory logic is being codified in industrial settings too. ISO/IEC forty-two-thousand-and-one and the established functional safety standards, IEC sixty-one-five-oh-eight and sixty-one-five-one-one, now establish risk-based classification, human oversight, and lifecycle management as baseline expectations for industrial AI deployment.
Pulling these threads together, the pattern is consistent. Capital is concentrating around the largest players.
Chapter summary auto-generated from the verified script. Listen to the full episode for the complete content.