人工智能：AI基础设施价值链

Our fact base and framework for comparing fundamentalupside and identifying winners across the AI value chain. The AI boom versus bubble debate has raged on since 2023. The range of outcomes couldscarcely be wider, from advanced superintelligence to a crash of epic scale. While it is clearlydifficult to pinpoint the exact trajectory, we see AI spend remaining healthy for the foreseeablefuture. However, we primarily focus on the more analyzable question: Where is AI infrastructurespend going, and who could benefit? We break down the GB200 NVL72 data center bill of materials and estimate an all-in AIdata center capex at $36Bn per GW. This spend is notably dominated by the GPU and byNvidia gross profit dollars. Networking is the other big-ticket item, and there is clear upsidefor foundry, HBM, WFE, and mechanical and electrical equipment. We further estimate incremental AI profit dollars as GWs * TAM/GW * Market Share *Incremental Margins. Based on this framework, we find that Ibiden, Unimicron, and other PCBand substrate names could have further upside, while Intel, Cisco, and server OEMs have alower upside potential relative to their prominence in the debate. PORTFOLIO MANAGER'S SUMMARY The range of outcomes in AI could scarcely be wider: at one extreme, we could see a crash ofepic proportions, wiping out potentially trillions of dollars of shareholder value. At the other, wecould achieve advanced superintelligence and the obsolescence of the human race. We hope for a golden mean in which AI boosts productivity, driving the next leg of economicgrowth and helping to develop technological solutions for many of the world’s problems.However, we lack the power to meaningfully influence the outcome, or even to advance a high-conviction prediction. Instead, we focus on what we can do: help investors make money. While we lack conviction in the long-term outcome, we see AI spend remaining healthyfor the foreseeable future.The medium- to long-term outcome depends heavily on the scalinglaws, which are not knowable, alongside building the next generation of frontier models andseeing how capable they are. However, we observe that AI believers dominate top-level decision-making at most key technology firms. Moreover, given the continued progress in model capability,we lack visibility on any downside catalyst that would cause decision-makers to change theirviews — suggesting AI spend will remain healthy. We model an estimated AI data center capex at $36Bn per GW ($6Mn per rack).Datacenter capex is dominated by the graphics processing unit (GPU), which we estimate at 38% oftotal costs. Networking is the other big-ticket item at ~12% of spend. Storage is relatively small,while spending on mechanical and electrical equipment is significant but less concentrated. Based on this analysis, we construct a framework for estimating AI upside acrosscompanies and sectors.At a high level, our framework is extremely simple: first, GWs ofcapacity coming online * TAM/GW * company market share = incremental revenue by company.Furthermore, incremental revenue * incremental margins = incremental profit dollars. While these estimates of AI upside are imprecise, we find that PCB, substrate, GPU, ASIC,and electricals stocks could still have further upside.Beyond industry favorites such asNVIDIA and Broadcom, we find that Ibiden, Unimicron, GPU and ASIC names such as AdvancedMicro Devices (AMD) and Mediatek (covered), as well as electrical names such as Eaton, could allsee very large upside opportunities relative to current profit footprints. While the US-China AI race remains a hot topic, the US is clearly adding more compute,and China is not close.However, this adds context to the US bans across AI and semicap. Stacy A. Rasgon, Ph.D.Daniel ZhuAlex WangChad DillardQingyuan Lin, Ph.D.David Dai, CFA March 27, 2026 TABLE OF CONTENTS SURVEYING REVISIONS AND STOCK MOVES17A quick scan of how the GenAI boom has translated into fundamentalrevisions and stock moves COMPARING FRONTIER MODELS35Who is leading — and, more importantly, how industry-wide modelprogress is trending THE GPU DEPRECIATION DEBATEAddressing one bear case: Why is it reasonable to depreciate GPUsover a six- to seven-year lifespan? THE DATA CENTER BILL OF MATERIALS55What actually goes into a GW of data center capacity? OUR AI UPSIDE FRAMEWORK69GPU, GPU components, and electrical names appear to have the mostupside leverage to the theme US VERSUS CHINA COMPUTE CAPACITY ADDITIONS83The US-China AI race adds a geopolitical dimension, but the US islikely to remain ahead in the near-to-medium term SIGNIFICANT RESEARCH CONCLUSIONS THE TRILLION-DOLLAR QUESTION: IS THE AI BOOM A BUBBLE? AI has been the biggest theme driving markets.It is hardly controversial to say that, since theNovember 2022 launch of ChatGPT and NVIDIA’s subsequent beat and guide-up in May 2023,AI has been the single biggest theme driving public markets. Hyperscale capex guidance pointedto $400Bn+ in