Why I Still Believe Samsung Is Worth Investing In

When gold miners crowd the hills, sell shovels instead.

If you cannot sell shovels, invest in the companies that do.

If shovels have become too expensive, move further upstream:

Identify the most profitable and irreplaceable players along the entire value chain, and invest there.

Wall Street can short, but the AI gold mountain will not be ignored.

These are my reflections during the “Great AI Era.”

Over the past year, Samsung Electronics has nearly doubled. I missed the initial run-up. Yet I increasingly believe Samsung occupies the core shovel-seller position in this wave of AI infrastructure.

The Great AI Era: What Is the Real “Shovel”?

In recent years, Nvidia has been the undisputed shovel seller, reporting soaring profits quarter after quarter.

Yet a Sword of Damocles hangs overhead:

Google’s TPU, Amazon’s Trainium, and proprietary silicon from Chinese tech giants like Alibaba and Huawei.

Capable hyperscalers refuse to surrender their margins to Nvidia indefinitely, turning in droves to in-house silicon.

This sword has hovered in the air since last year’s “DeepSeek moment.” It will not drop suddenly; rather, it will gradually slice into Nvidia’s margins—first replacing Nvidia silicon in inference workloads, then progressively encroaching on training clusters, eventually showing up in Nvidia’s forward order books.

Tracing the supply chain upstream: tech giants will develop in-house chips to defend margins, but custom ASICs cannot be conjured from sand alone. High Bandwidth Memory (HBM) is the indispensable bottleneck.

Whether running on Nvidia GPUs or in-house ASICs, AI models cannot train or infer without memory bandwidth.

At its core, AI model training is:

Compute × Bandwidth × Storage

Custom silicon can bypass Nvidia GPUs, but it cannot bypass HBM.

TPUs require HBM; Trainium may avoid HBM in certain architectures, but still relies on high-speed DDR from Samsung and SK Hynix.

Samsung’s Strategic Moat

• One of the world’s three leading DRAM manufacturers
• A core supplier of advanced HBM
• In-house leading-edge foundry fabrication capabilities

In the current supply chain hierarchy, Samsung and SK Hynix sit right at the front, commanding the high-margin HBM product category.

Against the backdrop of US-China geopolitical competition, both American and Chinese tech firms require advanced memory technology. Regardless of which camp expands faster, the colossal market size is sufficient to buoy earnings for Samsung and SK Hynix.

Amid cross-strait geopolitical risks surrounding Taiwan, South Korean semiconductor infrastructure carries markedly lower existential risk compared to TSMC, while offering the advanced foundry alternatives sought by US supply chains.

This provides Samsung with an attractive profile: high margins coupled with comparatively muted geopolitical exposure.

HBM Growth Potential: Are Latecomers Just Holding the Bag?

I prompted an AI model to simulate the relationship between earnings growth and share price performance.

Because buying individual South Korean shares directly can be cumbersome, I chose the EWY ETF—a basket of South Korean equities where Samsung and SK Hynix account for roughly ~45% of total holdings.

I ran the following projection:

If Samsung and SK Hynix see their HBM earnings grow 30% annually, while other divisions remain flat and EWY’s P/E multiple holds steady, how would share prices evolve?

Based on HBM’s current profit contributions, over five years this implies SK Hynix +108% and Samsung +54%, translating into EWY +34.6% (roughly a 6% annualized return).

Yet a 30% growth rate is exceptionally conservative. Nvidia’s average net income growth over the past three years hovered around ~88%. Modeling alternative scenarios:

How you structure this forecast depends on your outlook on AI infrastructure buildout. Consider Nvidia’s net income trajectory over the past five years:

• +125%
• -55%
• +581%
• +145%

Let the Music Play On

“The duck is the first to know when spring waters warm.” Programmers are the first ducks in the pond.

Many worry the AI cycle is nearing its tail end or teetering toward a dot-com style bust. But as a software engineer who uses AI tools every day—writing code with AI, generating this Hugo blog with AI, and cross-referencing market data with AI—this feels like the closest thing to an Industrial Revolution I have ever witnessed. If we are genuinely embarking on a new industrial era, what is there to fear?

Samsung and SK Hynix have already surged over the past year. I did not track or hold them previously, but I have resolved to practice what I preach: accumulating quality positions on dips.