Huawei Technologies' leading semiconductor scientist has issued a stark warning that major Western chipmakers, including artificial intelligence powerhouse Nvidia, will inevitably encounter insurmountable physical constraints as they push to develop ever-more-powerful processors. The cautionary message from Liao Heng, delivered during a rare four-hour public discourse streamed across China in late July, underscores deepening technological divergence between US-aligned and Chinese semiconductor ecosystems amid escalating geopolitical tensions.
Liao's extended interview with a prominent Chinese media personality served as a platform for articulating Huawei's emerging vision for semiconductor advancement. Beyond his technical warnings about Western foundries, he discussed the severe operational disruptions the Shenzhen-based technology conglomerate endured when American sanctions severed its connections to global chipmaking suppliers approximately five years ago. Most significantly, Liao showcased his company's unconventional innovation framework, the Tau Scaling Law, positioning it as a viable alternative trajectory for chip development when traditional miniaturisation approaches become physically impossible.
Central to Liao's thesis is the notion that foundries like Taiwan Semiconductor Manufacturing Company and Intel have approached structural boundaries in their ability to continually compress transistor dimensions on silicon. "There has to be a limit in how they scale up with ever-increasing compute die and more HBM," Liao stated, referring to the high-bandwidth memory components critical for artificial intelligence systems. "The industry is still pushing, but once they cross that physical limit, there will be an avalanche." This assertion reflects a viewpoint gaining traction within semiconductor research circles: that the era of relentless transistor miniaturisation, the cornerstone of Moore's Law for decades, is approaching exhaustion.
While Taiwan Semiconductor Manufacturing Company, Intel, and Samsung Electronics have maintained ambitious technology roadmaps extending years forward, buoyed by irreplaceable advanced lithography systems manufactured by Dutch equipment maker ASML Holding, Huawei operates under fundamentally different constraints. Comprehensive Western sanctions prevent the Chinese technology company from acquiring the cutting-edge fabrication machinery necessary to pursue conventional miniaturisation strategies. This restriction has forced Huawei's engineering leadership to reimagine chip architecture from foundational principles, positioning constraints as opportunities for breakthrough innovation rather than mere obstacles.
Huawei's Tau Scaling Law represents this conceptual pivot. Rather than pursuing the traditional path of making individual transistors progressively smaller, the framework emphasises enhancing data transmission velocities between distinct components within computing systems. As the efficiency gains from further transistor size reduction diminish and encounter physical barriers, this alternative approach offers a complementary avenue for boosting computational performance. The company intends to showcase its first smartphone processor engineered according to this paradigm shortly, incorporating a technique designated LogicFolding that optimises how information flows throughout the chip architecture.
Liao projected that the technology sector will gain definitive insights into Huawei's alternative methodology's competitive viability when the new smartphone processor undergoes comprehensive analysis by influential semiconductor research organisations including SemiAnalysis and TechInsights later this year. These independent teardowns and technical evaluations will provide the industry with concrete performance data, enabling precise assessment of whether Huawei's unconventional approach genuinely narrows competitive distances with internationally dominant suppliers. The stakes extend beyond corporate competition; the outcome will influence whether China's semiconductor ambitions can materialise despite international isolation.
According to Liao, the technological bifurcation separating American and Chinese semiconductor strategies reflects irreversible geopolitical fragmentation reshaping global commerce. He articulated a strategic imperative for both regional blocs: "To survive, each side has to build its own complete manufacturing and supply capabilities, even without serious clashes between the two sides." This formulation suggests that complete decoupling of semiconductor supply chains represents not merely a contingency measure but an inevitable structural reorganisation of the global technology economy. The implication carries significant weight for Southeast Asian nations heavily dependent on both American and Chinese technology inputs, potentially forcing difficult alignment decisions.
Liao's exceptional public engagement signals intensifying Chinese resolve to achieve technological self-sufficiency during the artificial intelligence epoch, when semiconductor capabilities determine competitive positioning across virtually all technology sectors. His willingness to discuss strategy and philosophy publicly reflects institutional confidence that Huawei's semiconductor initiatives have progressed sufficiently to merit broader explanation and, implicitly, international scrutiny. This contrasts sharply with conventional corporate practice of guarding advanced technology details closely.
During his extended podcast appearance, Liao articulated an alternative conceptual framework for understanding the entire artificial intelligence value chain, characterising it as an eighteen-storey pagoda structure. This metaphor deliberately surpassed the layer-cake analytical model previously popularised by Nvidia chief executive Jensen Huang, implicitly positioning Chinese thinking as more sophisticated and comprehensive. Crucially, Liao emphasised that Chinese technology enterprises operating across all value-chain tiers must actively support and coordinate with each other, developing resilient domestic ecosystems capable of functioning independently from Western suppliers.
Liao specifically commended Liang Wenfeng, founder of Hangzhou-based artificial intelligence research laboratory DeepSeek, whose breakthrough model in early 2025 catalysed multi-billion dollar stock market adjustments by demonstrating that Chinese developers could train sophisticated artificial intelligence systems using substantially fewer computational resources than American counterparts required. Liao attributed DeepSeek's achievement to necessity-driven architectural innovation, where limited computing capacity forced the research team toward more efficient algorithmic designs. He characterised this dynamic as analogous to maximising spatial utility within a confined apartment while competitors inhabit spacious villas, suggesting that resource constraints can paradoxically stimulate creativity and efficiency.
Huawei itself is developing proprietary artificial intelligence processors specifically optimised for supporting computational-efficiency-focused model architectures, employing what Liao termed a co-design mechanism integrating hardware and software development. This integrated strategy contrasts with historically siloed hardware and software engineering, instead treating chip design and algorithmic requirements as mutually reinforcing disciplines. "We have to put in more effort around design to trade higher complexity for less compute resource consumption," Liao explained, articulating the fundamental philosophy underpinning Chinese semiconductor strategy during an era of international technological competition and constraint.
