Samsung made $38 bn in a single quarter: The next 5 years could be even bigger

In 2024, the company’s leadership issued a public apology. They admitted to “falling short of the market’s expectations” and failing to maintain “technological competitiveness.” For a company that once defined what it meant to be a global tech giant, that was a remarkable thing to say out loud.

The stock reflected every bit of that humiliation. For much of early 2025, Samsung was trading at tangible book value. Which, in plain English, means the market was valuing Samsung at exactly what its assets were worth on paper. No premium. No excitement. No future priced in. Just: here are the buildings, the machines, and the inventory. Take it or leave it.

Some investors left it. Others bought it. The ones who bought it made a lot of money.

So, what actually changed?

Two things happened in quick succession that rewrote the Samsung story entirely.

First, Nvidia approved Samsung’s high-bandwidth memory chips for use in its GPU clusters. This was the breakthrough Samsung had been chasing for months. High-bandwidth memory, or HBM, is the specialised chip that sits alongside Nvidia’s processors inside the AI data centres being built by Microsoft, Google, Amazon, and every hyperscaler in between. Without HBM approval, Samsung was essentially locked out of the most important supply chain in technology. With it, the doors opened.

Second, Tesla signed a chip manufacturing deal with Samsung’s foundry division. Samsung’s foundry business, which makes chips designed by other companies, had been struggling. It consistently trailed TSMC, the Taiwanese industry leader, and had developed a reputation for inconsistent execution. A Tesla deal was the kind of credibility it badly needed. The stock did what stocks do when a narrative flips. It shot up roughly 60% in a matter of months.

But here’s the bigger story: Samsung is not just recovering. It is sitting inside a structural boom.

In the first quarter of 2026, Samsung’s operating profit came in at 57 trillion won, or about $38 billion. That is more than eight times what it earned in the same quarter a year before!

The semiconductor division alone accounted for 61% of Samsung’s sales and 94% of its operating profits

The number of memory chips it sold rose about 20% compared to the previous quarter. But the average selling price of those chips rose 90%.

Read that again: Volume up 20%, prices up 90%.

That is not a recovery. That is a supercycle.

Why is the memory market so tight?

The memory chip market is essentially controlled by three companies: Samsung, SK Hynix, and Micron. Together, they account for close to 95% of the global market. When demand surges, there is simply nowhere else to go.

And demand is surging in a way that is genuinely different from previous cycles.

AI models need memory at two distinct stages. During training, large GPU clusters pull data through high-bandwidth memory at enormous speed. During inference, when the model is actually being used, memory is needed to maintain context, store user histories, and personalise responses. Every time you chat with an AI assistant and it remembers something you told it last week, that memory is stored somewhere. Multiply that by 800 million ChatGPT users and 650 million Gemini users, and the numbers quickly add up.

It is not just HBM anymore. Traditional DRAM and NAND memory are both growing rapidly too, precisely because the use of AI has broadened well beyond the GPU cluster.

Micron reported results this week that confirmed exactly this picture. Its DRAM business grew 69% year on year. NAND grew 22%. Analysts at KeyBanc noted that HBM capacity is already sold out for all of 2026, and that DRAM and NAND demand is expected to exceed supply throughout the year. Morgan Stanley’s semiconductor analyst described the situation on the ground in vivid terms: hyperscalers are locking in pricing for multiple years, while PC and server companies are only receiving about 70% of their memory requirements. Allocation and supply-line situations, he noted, exist “basically everywhere.”

When customers are scrambling to lock in multiyear contracts, they are telling you they expect prices to keep rising. They are paying a premium for certainty. And the memory producers are more than happy to collect it.

What about the risks?

The honest answer is that the memory business is, and always has been, cyclical. Samsung knows this better than anyone. After the last flash memory boom, it overbuilt capacity just as demand softened, and its operating profit fell by half in 2019. Management is clearly trying to avoid a repeat of that mistake. A new fabrication facility will start production later this year, and another, costing roughly $55 billion, will begin construction in July but will not be ready until 2030. Capital expenditure is rising in absolute terms, but is actually falling as a share of revenue. That is a deliberate choice.

The other risk worth taking seriously is efficiency. As AI models improve, they may require less memory per query. This argument is similar to the one made about DeepSeek at the start of 2025, when many analysts worried that more efficient AI would reduce the need for computing. In practice, every efficiency gain was swallowed by higher usage. The same logic likely applies to memory. More efficient models tend to attract more users, which generates more data, which needs more storage.

There is also the political dimension. South Korea’s government has floated proposals around redistributing chipmaking profits. A coalition of Samsung unions is demanding 15% of memory division profits be shared with workers and has threatened a multi-week strike. These are real complications, even if they are complications born of success.

The 30-year pattern that smart investors have been using

Here is something that does not get talked about enough. Over the past three decades, Samsung has touched tangible book value per share eight separate times. Each of those moments, in hindsight, turned out to be an extraordinary entry point.

From the bottom of the 1998 Asian currency crisis through to today, Samsung’s stock has compounded at roughly 24.6% annually. The stock is up about 470 times over that period. Investors who waited for tangible book value and bought each time it was offered have done extraordinarily well. Those who held through multiple cycles did even better.

Early 2025 was one of those moments. The stock hit tangible book value, and some fund managers made it a top three position precisely because of this pattern. Not because they had unique insight into Samsung’s short-term earnings, but because they trusted a 30-year rhythm.

Where does that leave us now?

Samsung is no longer the laggard it was in 2024. It has Nvidia’s blessing. It has a Tesla foundry deal. It has HBM capacity sold out through 2026. It has hyperscalers signing multiyear contracts because they expect prices to rise further. It has a memory market where three companies control 95% of the world’s supply, and demand is running well ahead of what any of them can produce in the near term.

The supercycle is real. The structural driver, which is the exponential growth in AI model usage and the memory that usage generates, is not going away. And Samsung, for all its historical stumbles, is one of only three companies on the planet that can address it at scale.

The question is not really whether Samsung will do well. The question is how long the cycle lasts, and whether the stock eventually trades back down to a point where the 30-year pattern offers another entry.

Until then, the business is running very, very hot.

Sonia Boolchandani is a seasoned financial writer She has written for prominent firms like Vested Finance, and Finology, where she has crafted content that simplifies complex financial concepts for diverse audiences

Disclosure: The writer and her his dependents do not hold the stocks discussed in this article. 

The website managers, its employee(s), and contributors/writers/authors of articles have or may have an outstanding buy or sell position or holding in the securities, options on securities or other related investments of issuers and/or companies discussed therein. The content of the articles and the interpretation of data are solely the personal views of the contributors/ writers/authors. Investors must make their own investment decisions based on their specific objectives, resources and only after consulting such independent advisors as may be necessary