Bernstein analyst predicts semiconductor supercycle driven by AI demand.

Huoxing Finance reports: On June 21, Bernstein’s renowned chip analyst Stacy Rasgon openly stated that this is the first time in his 18-year career that he has genuinely witnessed a semiconductor supercycle. With a Ph.D. from MIT and a background as an engineer, Rasgon’s data is staggering: the global semiconductor industry generated over $800 billion in revenue last year and is now racing toward $1.3 trillion this year. Every segment—from accelerators and memory to semiconductor equipment, network optical communications, power chips, and CPUs—is experiencing severe supply shortages. “The only consensus we hear is that no one has enough compute power. Take memory as an example: in AI chips, HBM may occupy over 85% of the silicon area, and producing 1GB of HBM requires about four times the silicon area of standard DRAM—meaning that even if fabs expand production at full speed, the actual increase in storage capacity remains extremely limited.” This supply-demand imbalance has even benefited Intel—its previously written-off inventory has been completely sold out, with customers saying, “We don’t care; just sell it to us.” Rasgon points out that the industry’s core focus is shifting from model training to AI inference—the key to commercial monetization—since training models themselves generates no revenue; only model usage creates value. According to Anthropic’s data, annualized revenue surged from approximately $9 billion in December last year to $30 billion by April this year—an almost vertical climb. In the chip competitive landscape, custom ASICs represented by Broadcom and NVIDIA GPUs are not a zero-sum game. “The real question is whether the opportunity is still growing—if it’s large enough, both will thrive.” Broadcom currently forecasts its AI revenue to reach $100 billion next year; ASICs currently account for around a few percentage points of AI chip market revenue and are expected to rise to 25–30% in the future—but will not fully replace GPUs. Regarding recently acquired AI inference startups like Groq by NVIDIA, Rasgon cites Jensen Huang’s view: “Not all tokens are equal; low-latency tokens hold higher value, and GPUs are not always the optimal choice for every task.” When asked about the most overlooked risk in the industry, Rasgon shifts focus from silicon back to the physical world—electricity. Estimates suggest that if NVIDIA’s projected annual infrastructure investment of $3–4 trillion materializes, the U.S. power grid would need to expand capacity by roughly 5% annually—a growth rate considered nearly impossible by power industry analysts. This means the next bottleneck will emerge in energy generation, cooling, and nuclear power. “But never underestimate human ingenuity—engineers always find a way when there’s money to be made.” Regarding Intel, new CEO Victor Peng’s pragmatic low-expectation strategy, coupled with better-than-expected yield rates for its new 18A process node and substantial investments from both government and NVIDIA, have significantly alleviated prior market concerns over its balance sheet. Rasgon concludes: As long as AI demand does not collapse, the full-supply-chain supercycle will continue. Capital markets must closely track capacity bottlenecks across every stage of the ecosystem.