OCP Summit signals full-stack collaboration, GF Securities: Semiconductor pullback creates a buying opportunity, bullish on NVIDIA, Marvell, and the CPU supply chain

GF SEC analysts attended the OCP Global Summit in San Jose last week, where the conference reflected a shift by hyperscale cloud providers toward full-stack co-design. The analysts said the current pullback in semiconductor stocks creates buying opportunities in quality names such as NVIDIA Corporation (NVDA.US) and Marvell (MRVL.US). Analyst Jeff Pu said in an investor report: "NVIDIA Corporation's networking roadmap now covers both co-packaged optics (CPO) and the new USFP pluggable standard, the latter developed jointly with Cisco Systems, Inc. (CSCO.US) to improve density and serviceability inside AI factories. USFP modules reach up to 12.8 Tb/s per module and 400 Gb/s per lane, achieving efficiency close to CPO while retaining the field-maintenance flexibility of a pluggable solution." USFP stands for Ultra Small Form-factor Pluggable, a new optical transceiver standard used to connect large-scale AI data center clusters. Pu noted: "Our survey of US neoclouds shows that power availability and permitting remain the decisive factors for expansion. However, there is no power shortage now or in the foreseeable future, because most sites are far from city centers and many use stable gas-fired generation. On power costs, regional spreads remain wide, driving new projects toward the southern corridor and some overseas regions such as Italy. On shells, modular data center shells can be completed in as little as three months, but liquid cooling infrastructure remains the bottleneck." The survey also showed that CPUs are becoming more important in inference, especially agentic AI, which is expected to double the CPU total addressable market. Pu said: "GPU utilization in decode workloads averages only 20% to 30%, and SRAM-based accelerators perform poorly on long-context efficiency. Token costs are falling 5 to 10 times a year, but semiconductor and memory costs are rising. The bottleneck is NAND, DRAM and thermal management. In the agentic AI factory model, orchestration and retrieval run on CPUs, reshaping unit economics: the CPU-to-accelerator ratio shifts from 1:4 to 1:8 in the chat Siasun Robot&Automation era to 1:1 in agentic deployments." He added: "Despite overall market weakness, we advise investors to buy quality stocks on dips. We remain bullish on NVIDIA Corporation, the PTFE segment, Marvell (DSP, Alphabet Inc. Class C companion, Maia 300) and the CPU supply chain (AMD, Intel Corporation)." Risks include slower AI demand, infrastructure deployment, supply constraints and competition.