How can investors map the AI empire being built through chips, power grids, and policy?

Inside Taiwan tracks the physical foundations of AI: chip export rules, China’s self-sufficiency push, Intel’s 18A manufacturing test, OpenAI’s government data center strategy, and the looming power crunch, globally. We also examine a new lawsuit over AI hiring scores and what it signals about trust, transparency, and control right now.

Q1. What is the AI Overwatch Act, and what would it change if enacted?
It would give Congress a 30-day window to review and potentially block licenses for exporting advanced AI chips to China and other adversaries. The committee advanced it by 42-2, and the latest version also bans Nvidia’s top-end Blackwell chips.

Q2. What should investors watch in the Nvidia H200 export debate?
Watch policy volatility and enforcement friction. The U.S. approval framework is now contested politically, and reported Chinese customs uncertainty shows how “allowed” can still mean delayed, constrained, or repriced in practice.

Q3. What does Alibaba’s reported T-Head IPO preparation signal for China’s chip ecosystem?
Alibaba is reportedly preparing to restructure T-Head, including partial employee ownership, before exploring an IPO. For investors, it is a signal that China is mobilizing capital markets to accelerate domestic chip design across data center, AI, and IoT processors.

Q4. What is the investor-grade read on Intel’s Panther Lake and “18A”?
This is a manufacturing execution story. Intel has acknowledged yield challenges and says yields are improving monthly. The key is whether improving yields translate into competitive cost, reliable volume, and credible foundry traction versus leading incumbents.

Q5. What are the two constraints investors should treat as non-negotiable: energy and trust?
Energy is the hard ceiling: one industry estimate projects data center electricity use could more than double from about 460 TWh (2022) to over 1,000 TWh by 2026. Trust is the hard floor: a lawsuit alleges Eightfold AI created secret applicant “scores” without proper disclosures, highlighting rising legal and compliance costs for AI adoption.