KB Securities Maintains Target Price at 160,000 Won
"Samsung to Benefit Most from AI Ecosystem Diversification"

On November 27, KB Securities maintained its "Buy" investment rating and target price of 160,000 won for Samsung Electronics, stating, "Samsung Electronics is expected to be the biggest beneficiary of the diversification of the artificial intelligence (AI) ecosystem, and we present it as our top semiconductor pick."


"AI Ecosystem Diversifying from Nvidia to Google and Others"

On this day, Kim Dongwon, Head of Research at KB Securities, said, "Google's expansion of its AI ecosystem through Tensor Processing Units (TPUs) is increasingly likely to lead to increased memory supply from Samsung Electronics, higher utilization rates for advanced process foundries, and increased Galaxy sales driven by Gemini AI."

[Click eStock] "Samsung Electronics to Benefit More from Google TPUs... Top Semiconductor Pick" View original image

Recently, Google launched Gemini 3.0, demonstrating the performance of its in-house AI chip, the TPU. Google has also concretized its AI revenue model by integrating TPUs with search advertising, YouTube, Gemini, and other services.


Kim stated, "This is the second case marking a structural shift in the AI market toward investment efficiency following the DeepSeek shock," and added, "From next year, the AI market is expected to rapidly shift from a focus on training to inference, and the AI ecosystem will diversify from being Nvidia-centric to including Google and others."


Accordingly, demand for not only high-bandwidth memory (HBM) but also standard DRAM is expected to surge simultaneously. Kim said, "Next year, Samsung Electronics' HBM supply volume will more than double compared to this year," and "TPUs optimized for inference AI will require much more storage computation than repetitive training, leading to a simultaneous surge in standard DRAM supply."


"TPUs to Significantly Lower Total AI Service Costs"

Kim particularly pointed out, "TPUs have many areas where matrix computation and inference efficiency are higher than Nvidia's graphics processing units (GPUs), and the cost per server is only about half that of GPUs, so TPUs will significantly lower the total cost of AI services."


Samsung's HBM4, which has a speed advantage, is expected not to require redesign in the future, making it highly likely to pass quality certification early. Kim said, "With the expansion of the inference AI market and the proliferation of AI application services, server data processing volume will surge, and next year's server DRAM demand will increase by 35% compared to this year," adding, "However, supply will remain below 20%, resulting in a severe supply shortage and continued sharp price increases."



He added, "Since 70% of Samsung Electronics' total DRAM production capacity is standard DRAM, the company's operating profit next year is expected to approach 100 trillion won, up 129% from this year."


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing