The demand for GPU packaging boards is surging, and Nvidia Blackwell is stepping towards a path of aggressive expansion.

Wall Street investment giant Bank of America recently released a research report showing that the operating profit of the electronic component business under the leadership of Ibiden, a semiconductor packaging equipment leader based in Japan, is expected to achieve a compound annual growth rate (CAGR) of 35% in the next five years. It also significantly raised its EPS forecast for the fiscal year ending 2028. The main reason is that the demand for GPU packaging substrates (i.e. ABF advanced packaging substrates) provided by Ibiden for the large-scale expansion of AI chip production capacity is showing explosive growth, driving a significant increase in both volume and price of ABF. According to research data from Bank of America, Ibiden, a leader in packaging equipment, is significantly expanding its production capacity (expected to add an unprecedented 60-70% capacity from 2024 to 2026). It is also expected to seize the historical opportunity of accelerating expansion of ASIC. Starting from 2026, it is expected to supply ABF substrates to AI ASIC super customers, such as Google, Microsoft, and Meta, for their self-developed AI ASIC chip packaging. The Bank of America analyst team has extremely optimistic expectations for Ibiden's valuation and demand fundamentals, indicating that Nvidia's Blackwell series AI chip/AI server rack shipments will kick off a "super growth cycle" in the second half of the year. The Blackwell series AI chip includes the Blackwell architecture AI GPU and the AI GPU based on Nvidia's most advanced architecture, Blackwell Ultra. For example, the Blackwell architecture with double die+8-HBM3E or the next generation HBM system-12-HBM4 CoWoS advanced packaging requires at least 14 layers of ABF with an ultra-large effective area. It is understood that research data from Wall Street institutions such as Bank of America and Morgan Stanley show that since the second quarter of this year, the Blackwell architecture AI GPU products and GB200 NVL72 AI server racks have entered an accelerated growth trajectory beyond expectations. With the realization of mass production of Blackwell Ultra architecture products, the Blackwell series AI GPU/AI server rack products are expected to embark on a more aggressive path of "global shipment volume and AI computing power demand surge" in the second half of the year. Why are ABF packaging substrates closely tied to AI chip packaging? Nvidia's Blackwell series double-die AI GPUs and Google's TPUs and other AI ASICs are striving to securely connect high-power chips, higher HBM stack numbers, and thousands of high-speed I/Os through TSMC's CoWoS advanced packaging. This cannot be achieved without high-layer ABF packaging substrates - these advanced packaging substrates must simultaneously achieve ultra-fine circuitry, low dielectric, high heat resistance, and low warpage. Currently, the material system that meets these strict criteria and has been mass-produced and verified is the ABF advanced packaging substrate. For example, Blackwell-B200 needs to connect two large GPU dies with up to 12-high HBM4 stacks to the silicon intermediary layer. The number of I/Os compared to Hopper has increased by more than 145%, requiring 10-14 layers of any-layer interconnect (mSAP process) ABF advanced packaging substrate to meet the impedance and crosstalk requirements. The high-speed SerDes spacing of Blackwell-B200 is much less...