Good morning. It's Saturday, July 11th, 2026. Let's get into it.

Story one: AI has discovered two new superconducting materials. Researchers at Rice University, led by Professor Emilia Morosan, combined machine learning with quantum physics to identify candidate materials, then synthesized and confirmed both in the lab. The results are published in Physical Review Research. Superconductors — materials that conduct electricity with zero resistance — are essential for quantum computing fusion energy MRI machines and maglev trains. The problem has always been that finding new ones is slow, expensive, and largely guesswork. This AI-accelerated search cuts through that, creating a systematic framework that can screen far more candidates than trial and error ever could. We are not at room temperature yet, but the method matters as much as the discovery. When you combine physics-first models with machine learning pattern matching, you get something neither approach could do alone. Expect this pipeline to become the standard for materials discovery across energy, aerospace, and computing.

Story two: Apple is in discussions with PrismML about running exceptionally large AI models directly on iPhone. The Neural Engine in Apple's A-series and M-series chips is getting serious upgrades — the M5 generation's Neural Accelerator delivers roughly four times the on-device LLM throughput of M4. Right now Apple Intelligence still depends on cloud inference for heavy lifting, but the gap is closing. If Apple can run frontier-class models locally, it changes the privacy equation entirely. No data leaves the device, no API calls, no cloud latency. It also means Apple could stop being the one company in AI that depends entirely on partnerships — Google, OpenAI, Anthropic — and bring serious capability in-house. The engineering challenge is real: memory compression, thermal management, Neural Engine scheduling, Core ML integration. But Apple has been investing heavily in on-device compute for years. If this works, the iPhone becomes the most private AI device on the planet.

Story three: The United Nations Independent International Scientific Panel on AI has published its preliminary report, released alongside the first UN Global Dialogue on AI Governance in Geneva. The panel, forty scientists appointed by the General Assembly, warns that AI is outpacing both scientific understanding and governance frameworks. The core message: as AI capability concentrates in a handful of nations and corporations, shared international standards for safety, transparency, and accountability are critical — or governments and citizens lose their say in the outcome entirely. The Geneva dialogue, which ran July 6th and 7th, brought together governments, researchers, and civil society, but the structural tension is obvious. Washington, Brussels, and Beijing are pursuing openly competing visions of AI governance, and the UN's ability to find consensus among 193 member states is, by its own admission, nearly impossible. Still, having an independent scientific body inside the UN system is new. Whether it can actually bridge three competing superpower frameworks remains to be seen.

That's all for today. Saturday morning — get outside, touch some grass. I'll see you tomorrow.