AI Rally Driven by Model Revenue, Not GPUs [Click e-Industry]

Expectations Rise for AI Model Revenue with GPT-6 Astra
"Software Stocks Also Need to Respond to the Cycle"

Artificial intelligence (AI)-related stocks are rebounding once again. While the most visible drivers are semiconductors, power, and data centers, some analysts argue that it is important to look further upstream. The key, they explain, is not simply the number of graphics processing units (GPUs) sold, but rather how quickly AI model developers such as OpenAI and Anthropic can grow their revenues.

On September 10, KB Securities pointed out that the recent volatility in AI-related stocks over the past three months stemmed from weakening expectations for revenue growth at model developers, which in turn raised questions about the profitability of data center investments. For example, AI model developers generate revenue by selling tokens. Hyperscalers and "neo-cloud" companies operate the data centers required to produce these tokens, while semiconductor and power companies supply the necessary equipment and infrastructure for those data centers. Ultimately, if expectations for model developer revenue growth weaken, doubts arise about the profitability of data center investments, and this pressure spreads to semiconductor and power stocks as well.

However, the landscape has recently shifted. KB Securities explained that revenue growth expectations for AI model developers have been revived, as OpenAI's release of GPT-6 Astra demonstrated a significant performance gap with competitors.

Accordingly, KB Securities emphasized that calculations regarding software stocks also need to be adjusted. At the beginning of the year, when Anthropic released Claude Co-Work, there were growing concerns that work software could be replaced by AI models. Subsequently, when China's OpenWeight model spread rapidly, there was a view that AI models could become widespread and universal, and some confidence returned to software companies enhancing their offerings using AI.

KB Securities researcher Ilhyuk Kim explained, "As the performance of AI models improves, workflows previously handled outside of software are increasingly being performed directly by AI models or their execution environments, which could undermine the value of the systems built up by software companies. However, as competition between AI models intensifies, companies are feeling greater challenges in managing token costs, and are now placing more importance on budget control than before." As companies strive for stricter cost control and authorization management, it is expected that the value of software companies will be less undermined.

Kim further noted, "When expectations for AI models rise, both the stocks that benefit from this as well as software companies can be invested in from a long-short strategy perspective. Rather than choosing one or the other, it is appropriate to manage long-term volatility by adjusting the weightings of each according to market trends."

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