[DigitalToday reporter Jinju Hong] Now, 60 years after the symbolic concept of the semiconductor industry known as Moore's Law was introduced, the industry appears to acknowledge its slowdown while still using it as a baseline for technological progress. Recently, the concept has been expanding beyond semiconductors into a framework for explaining the growth pace of the generative artificial intelligence (AI) industry.

TechRadar, an IT media outlet, reported on April 19 that Moore's Law has explained both rising semiconductor integration and falling costs, and has become a key indicator for gauging the direction of progress across the broader technology industry.

Moore's Law is a concept presented in April 1965 by Intel co-founder Gordon Moore (고든 무어) as he forecast the next 10 years of the semiconductor industry. He observed a trend in which the number of circuit components such as transistors doubled every year, and projected that the trend would continue for at least 10 years. The law was later accepted as a basic industry principle that computing performance improves exponentially while costs fall.

In practice, the semiconductor industry maintained this trend for far longer than Moore initially expected. In the early 1970s, a single chip contained thousands of transistors. By the 2000s, that number had risen to tens of millions, expanding by tens of thousands of times in about 30 years. That amounts to an increase of more than 18,000-fold in less than 30 years.

But since the 2000s, questions have grown over whether the pace of these increases is gradually slowing. Pat Gelsinger (팻 겔싱어), Intel's chief executive officer, said in 2023 that the transistor growth cycle had slowed compared with the past, suggesting that exponential growth at the same pace as before has become difficult.

These changes show that the semiconductor industry is no longer in the same phase of exponential growth as in the past. Even so, Moore's Law is still acting as a valid benchmark across the industry.

In particular, similar patterns have recently been observed in generative AI, expanding the concept's scope of application. Sam Altman (샘 알트먼), OpenAI's CEO, mentioned Moore's Law in 2025 as he explained the pace of AI development. The analysis is that since the spread of generative AI, model performance has improved rapidly while usage costs have fallen sharply, resembling the semiconductor industry's growth path.

AI computing costs are in fact falling quickly. Altman also said the cost based on context tokens dropped to about one-tenth in a year, and that between 2023 and mid-2024 ChatGPT's per-token price fell to about one-150th. This shows that the structure of simultaneous performance improvement and cost reduction is being repeated in the AI industry as well.

In this trend, Moore's Law no longer refers only to the speed of semiconductor miniaturisation processes. The era of explosive growth in transistor counts has passed, but the industry's competitive principle of delivering higher performance at lower cost remains valid.

With the balance of performance and cost emerging as a key competitive factor in both the semiconductor and AI industries, an assessment is emerging that Moore's Law remains an ongoing benchmark as it changes form.