"Will the Semiconductor Industry Be in Trouble?" Stock Prices Have Fallen, but Investments May Increase [Weekend Money]

Amid growing concerns in the memory semiconductor market over the possibility that global artificial intelligence (AI) companies such as Anthropic may slow down the pace of AI development, a report has concluded that the actual impact on DRAM demand will be limited.

According to a report by Kim Unho, a researcher at IBK Investment & Securities, "The news of a slowdown in AI development has initially raised concerns about delays in investments in high-performance graphics processing units (GPUs) and high-bandwidth memory (HBM), leading to an overall decline in memory-related stock prices. However, since the grounds for memory industry expansion lie in agent AI, the impact on the memory sector will remain limited."

Earlier, Dario Amodei, CEO of Anthropic, stated, "The pace of improvement in AI capabilities is excessively fast, necessitating intentional speed control." He pointed out that as frontier AI models are increasingly participating directly in next-generation model development, risks involving security, safety, and controllability are escalating rapidly. In reality, there have been reported cases where OpenAI’s new models have attempted to hack external systems autonomously. This shared concern has resonated with other global big tech leaders, including Demis Hassabis, CEO of Google DeepMind; Sam Altman, CEO of OpenAI; and Elon Musk, CEO of SpaceX.

Nevertheless, IBK Investment & Securities forecasts that this moderation in development speed will have limited effects on the overall memory semiconductor industry. Kim noted, "Even if the speed of developing training models slows, agent AI support remains sufficiently feasible with the models already in place. Investment in computing power (CPU, DRAM, NAND) for the agent AI market, which is expected to grow further, will continue independently."

On the contrary, the emerging focus on security and controllability could serve as a catalyst for additional infrastructure investment. Kim explained, "Paradoxically, strengthening AI security will be a variable that expands further infrastructure investment," adding, "It will drive even stronger demand for edge AI."

Ultimately, as the scope of agent AI expands, it is expected to underpin the growth of the memory market. While agent-function-enhanced models such as OpenAI’s Astra and Meta’s Muse are gaining momentum—currently with functionality concentrated on cloud servers—the market is likely to further expand into edge and on-device agent AI in the future.

Kim predicted, "As models with enhanced agent functions are rolled out in earnest, the growth of the memory market will continue steadily."

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