Twenty-five years ago, $Intel (INTC.US)$ made history by becoming one of the first two technology companies to be included in the Dow Jones Industrial Average. Today, chip company $NVIDIA (NVDA.US)$ has seen its stock price rise sevenfold in the past two years, replacing Intel as a new member of this blue-chip index. This is just the beginning of the chip industry revolution.

Due to strong demand for chips driving generative ai technology, high-end ai chips are still in short supply. The involvement of only a few chip manufacturers in this lucrative market seems to ensure profit growth and stock price increases in the industry.

However, investing in chip stocks has been an extremely risky bet this year. Intel's stocks have fallen nearly 50%, while nvidia has tripled, becoming the best-performing company in the dow jones index.

What are the reasons behind this extreme stock price disparity? This is related to the three major challenges facing the industry.

Firstly, most chip manufacturers are currently lagging in designing and manufacturing high-performance ai chips. Nvidia's advanced design means that chip manufacturers face higher industry standards, and the era dominated by smart phone and auto manufacturers' demands is no longer the case. Quality standards are becoming increasingly stringent, and even for a few chip manufacturers with the technology and funds, long delivery times are still required.

Secondly, the two largest companies in the world capable of manufacturing high-end ai chips are currently struggling to recover from previous missteps. For intel, just one delay in factory renovations in 2015 hindered its chip acceleration process, beginning to undermine its leading position in global chip manufacturing.

This delay, combined with inadequate investment, has led to$Taiwan Semiconductor (TSM.US)$When production of 10-nanometer chips began in 2017, the technology gap with Intel narrowed in just two years. Since then, this gap has further widened, and Intel has handed over its manufacturing advantage to Taiwan Semiconductor.

Nvidia outsources the manufacturing of its physical chips, which is especially lucrative. However, intel's market share is negligible, whereas taiwan semiconductor's market share reaches 62%. The same applies to Samsung. High bandwidth memory chips are a crucial component of all nvidia ai chips, while Samsung lags behind sk hynix in this area. According to TrendForce, sk hynix is expected to account for over 52% of the market share this year, surpassing Samsung.

Thirdly, similar to the real estate and auto markets, the high-end chip market significantly outperforms other lower-end markets. In recent years, the profit gap between high-end chips and general memory chips or outdated chips used for traditional electronic products has been continuously widening.

This is partly due to the slowdown in demand for devices like smart phones and personal computers. However, the biggest factor is that the average selling price of high-performance ai chips is about five times that of traditional memory chips.

This exacerbates the trend in the chip industry where the winners take all. The two best-performing chip manufacturers globally this year happen to be nvidia's two major suppliers: taiwan semiconductor set a net income record in its latest financial report and raised its annual revenue forecast, while sk hynix reported record operating profit for the latest quarter. Both companies achieved operating margins exceeding 40% in the third quarter, more than double that of their peers. taiwan semiconductor's stock price has increased by 80% this year, and sk hynix has also risen by one-third.

This means that the risk of falling behind is higher than ever. Samsung's chip business profits in the latest quarter fell short of expectations, while intel launched its own ai chips to compete with nvidia but still recorded a historic quarterly loss of 16.6 billion dollars. intel also canceled its 2024 sales expectation of 0.5 billion dollars for the Gaudi ai chips. In contrast, nvidia's datacenter business, which is the segment that manufactures ai chips, achieved sales of 26.3 billion dollars in the quarter ending in July.

Traditional memory chips are increasingly oversupplied, and this is expected to further squeeze their prices and profit margins in the coming quarters. High-profit ai chips are increasingly becoming the only way to offset this weakness.

In a world dominated by nvidia's ai, the signal to global chip manufacturers is quite clear: if you can't beat them, join their supply chain.

Editor/Rocky