The world's three major memory chip manufacturers—Samsung Electronics, SK Hynix, and Micron Technology—have all surpassed the $1 trillion market value mark. Even including Saudi Aramco, with a market value close to $1.8 trillion, the combined market value of these three storage giants is about 22% higher than the combined total of the world's top three oil companies.

This wave of enthusiasm even swept across the downstream of the industry chain.

Since March this year, flash memory manufacturer SanDisk's market value has nearly tripled, and its current scale can rival that of China, Asia's largest oil producer.

Such a strong rally at first glance seems to signal that memory stocks are about to face a correction risk, given the industry's historically highly cyclical nature. However, recent fundamental changes in business models have made future expected returns much more certain. Compared to these promising future earnings, even memory companies with market values exceeding one trillion currently seem quite "cheap."

| From "Commodities" to "Long-term Contracts"

Like oil, memory chips have long been widely regarded as highly volatile commodities. But AI is driving demand for memory chips far beyond the capacity of existing suppliers, pushing prices to unprecedented heights. Memory manufacturers, in turn, use their newly gained bargaining power to encourage customers to sign long-term agreements.

Enough such protocols will reshape the industry's business model, effectively curbing price volatility. At present, these efforts have begun to show results. In its latest March financial report, Micron Technology announced the signing of its first five-year supply agreement and stated at last week's investment meeting that it had made "substantial progress" with other customers on similar agreements.

A report released last month showed that five customers have signed long-term agreements, enough to cover more than one-third of the company's capacity in the next fiscal year.

SK hynix did not specify how many customers have signed long-term agreements. However, given that the company disclosed in its most recent earnings call that market demand over the next three years will "far exceed" its capacity, it is clear that the company has engaged in in-depth discussions with these customers about future needs. SK Hynix CFO Kim Woo-hyun stated bluntly during a conference call: "Today, memory has become critical, and customers view uncertainty in memory prices and supply as critical business risks." ”

The purchasing power of the largest investors in the AI sector means that a significant portion of memory production will be covered by more stable multi-year contracts in the future.

瑞银分析师蒂姆·阿库里估计，此类合同明年将覆盖高达30%的动态随机存取存储器（DRAM）总出货量。他认为，像微软、Alphabet旗下的谷歌和亚马逊这样的大型科技公司——超大规模数据中心运营商——已经占据了服务器所用DRAM芯片全球约三分之二的产能。

“换句话说，超大规模数据中心运营商似乎越来越愿意以价格取胜，换取多年供应的可见性和未来部署经济效益的更高可预测性，”阿库里在本周早些时候的一份报告中写道。

| "Cabbage Price" Under a Trillion-Yuan Market Value?

Long-term contracts make future earnings for memory companies more predictable. And these gains are quite substantial. Micron Semiconductor's adjusted earnings per share for the quarter ending February surged from $1.56 in the same period last year to $12.20. Wall Street expects Micron's earnings per share to exceed $60 for the fiscal year ending in August, while Visible Alpha forecasts that next fiscal year's earnings per share will be around $106.

Therefore, based on Micron Semiconductor's current P/E ratio slightly above $1 trillion, its expected earnings multiples for the next four quarters are less than 10 times. This puts Micron Semiconductor's price-to-earnings ratio in the bottom 10% among S&P 500 constituents, while SanDisk's P/E ratio is also close at about 10.5 times.

Korean manufacturers are valued even lower; according to FactSet data, Samsung and SK Hynix have P/E ratios between 6 and 7 times. In contrast, the average price-to-earnings ratio of the 30 constituents of the Philadelphia Semiconductor Index is about 26 times.

Memory may become the new oil, but Memory investors have not yet experienced a high price shock.

|Behind the breakthrough to a trillion-yuan market value

Just a few years ago, memory chips were still a relatively old and little-known field in the semiconductor industry. Many tech leaders and investors are focusing on the development of advanced processing chips, such as those designed by NVIDIA and chips produced by TSMC—both companies with market capitalizations exceeding $1 trillion.

The rapid rise in tech stocks has also raised concerns about an AI bubble and a possible adjustment in the global economy if AI fails to deliver on expected profits or delivers on its promise to revolutionize the workplace.

RBC analysts pointed out in a research report last week that South Korea's benchmark stock index has continued to climb in recent weeks, becoming the seventh largest index globally and being regarded as a "model of Asian AI tech stock rally."

However, Samsung and SK Hynix hold nearly half of the weight in this index, so if the tech sector experiences a downturn, the index's upward momentum will be severely affected.

Kingtech Perspective | Embracing the dual safety cushion of "low valuation + long-term contracts."

Currently, the three storage giants (Micron, Samsung, SK Hynix) have price-to-earnings ratios far below the semiconductor industry average, and long-term supply agreements have locked in profit floors for the coming years. For conservative investors, focus on these leading stocks with "cabbage price" characteristics, as they offer a very high margin of safety.

Be wary of the risks of a single market and focus on diversified allocation

Although the Korean stock market performed well due to the AI rally, Samsung and SK Hynix occupy nearly half of the index. If the technology sector pulls back, systemic risks are extremely high. It is recommended not to over-concentrate index funds in a single country or region when allocating, but to spread holdings across US, Korean, and A-share related industry chain targets to smooth out volatility.

Closely monitor the sustainability of AI computing power demand

The boom in memory chips is built on the foundation of the AI arms race. Investors need to closely monitor the capital expenditure plans of hyperscale cloud providers such as Microsoft and Google. As long as major companies continue to frantically hoard computing power, the "seller's market" logic of storage chips will not easily collapse.