Huawei's Tau Scaling Law Targets 1.4nm Chips; India Emerges as AI Data Center Hub

China's Huawei charts a path around U.S. chip controls with a new scaling principle as India's AI data center market accelerates past core tech growth.

🗄️ Huawei Unveils 'Tau Scaling Law' — A Post-Moore's Law Path to 1.4nm-Equivalent Chips

Decoded: Huawei Technologies revealed at a semiconductor symposium in Shanghai on May 25 that its high-end chips will achieve transistor density equivalent to 1.4-nanometer processes by 2031 — without relying on cutting-edge lithography equipment that U.S. sanctions have blocked China from accessing. The strategy is built around a new principle Huawei calls the Tau Scaling Law: instead of shrinking transistors further — the basis of Moore's Law, which is nearing physical limits — Huawei's approach focuses on reducing the time it takes electrical signals and data to travel through chips and computing systems. China's most advanced proven chipmaking capability is widely assessed at around 7nm, while TSMC currently manufactures at 2nm and plans to introduce 1.4nm for mass production in 2028. Reaching near-frontier density through alternative means — advanced packaging, chiplets, and interconnect optimization — would allow Chinese AI chip designers to close the performance gap without the extreme ultraviolet lithography machines Washington has blocked. Industry analyst He Hui of Omdia described the approach as a credible way to extract more performance when leading-edge lithography is constrained. (Reuters, Bloomberg, May 25, 2026)

Why it matters: Huawei's Tau Scaling Law is the most specific public articulation of China's post-EUV chip development strategy. If the 2031 target has technical credibility — and independent analysts say the approach is architecturally sound — it means U.S. export controls on lithography equipment may not be able to permanently cap Chinese AI chip performance. For investors, TSMC (TSM) remains the dominant advanced-process producer through the end of the decade, but the path China is building is not through TSMC's equipment — it is around it. A 1.4nm-equivalent chip built through interconnect and packaging innovation could be sufficient for inference-scale AI workloads even if it does not match TSMC's raw transistor density. The Huawei announcement also reframes the export control debate: the question is no longer whether China can catch up, but how long it will take.

📊 Schneider Electric: India AI Data Center Revenue to Outpace Core Business Over Next 5 Years

Decoded: Schneider Electric expects its India data center infrastructure business to grow faster than its broader India operations over the next four to five years, driven by surging demand for AI-ready infrastructure, Reuters reported on May 25. Data centers currently account for 15 to 20 percent of Schneider Electric's India revenue and are expanding at a double-digit rate, according to Sumati Sahgal, the company's vice president for Secure Power in India. Schneider provides power management and cooling infrastructure — the physical layer that determines whether AI workloads can run continuously without throttling or failure. India's AI data center build-out is accelerating as global hyperscalers including Google (GOOGL), Microsoft (MSFT), and Amazon (AMZN) have committed multi-billion-dollar India cloud infrastructure investments in 2025 and 2026, and domestic enterprises are deploying AI inference capacity at a pace that has consistently outrun available power infrastructure. (Reuters, May 25, 2026)

Why it matters: Schneider Electric's India data center forecast is a read-through on the global AI infrastructure buildout extending to emerging markets at scale. When a major power and cooling infrastructure supplier projects its fastest-growing segment will outpace its core business in the world's most populous market, it signals that AI infrastructure capex is not concentrated in the U.S. and Asia-Pacific hyperscaler corridor — it is becoming a global construction cycle. For investors, the power delivery and cooling layer of AI data centers — less visible than GPU procurement — is increasingly the binding constraint on deployment velocity. Schneider Electric's position in that layer, alongside Eaton and Vertiv, means infrastructure suppliers are likely to see sustained demand regardless of which AI model wins the software race.

Stay decoded. See you tomorrow.

— The Get AI Decoded Team