저자: 미상 · 외국인
요약 보기
AI 붐에서 '삽 장사' 전략으로 메모리 반도체 기업에 투자하는 DRAM ETF를 소개. 2026.4.2 상장 이후 누적수익률 약 111%로 XLK·ARK·QQQ·VOO 대비 우위를 주장하나, 상장 기간이 짧고 액티브 운용 리스크, 메모리 경기 사이클 민감성, 최대낙폭 약 46%를 함께 지적. 결론적으로 장기 보유보다 가격 흐름 기반 단기 매매 관점을 제시하며 유료 강의 홍보.
다듬은 글
During the California Gold Rush , which began in 1848, hundreds of thousands of people rushed to the American West with one dream: to find gold and get rich. But here lay an interesting paradox. The people who made the most money weren't necessarily the ones who mined gold. These were the ones who sold shovels.
Because when hundreds of thousands of people rush after the same opportunity, the demand for the tools needed to do so almost inevitably increases. Almost 50 years later, a similar story unfolded during the second industrial revolution. Electricity, steel, and oil became critical resources and infrastructure for the new economy.
Companies like General Electric , US Steel, and Standard Oil did not necessarily produce final products for consumers, but they provided what the entire industrial economy needed. Electricity to run machines, steel to build factories, railways and bridges, and oil to fuel an economy that was becoming increasingly mechanized.
And now we are entering a new industrial revolution, the fourth industrial revolution . This time, it's not electricity or the internal combustion engine that's changing the world. This is artificial intelligence, AI. There are several major AI models today, including Gemini, ChatGPT, Claude, Grok, and others.
Hundreds of billions of dollars are being invested in the race to create increasingly powerful AI models. But the question arises: who will actually win? OpenAI, Google, Anthropic, or a completely new company that will emerge within the next few years? No one knows for sure, because the history of technology shows us time and again that today's leader can easily be replaced tomorrow.
So , instead of trying to guess which AI company will ultimately win, we can take a different approach, using a “ shovel-selling” strategy. The idea is simple. Look for companies that provide the infrastructure every AI company needs. Because Gemini needs infrastructure, ChatGPT needs infrastructure, Claude needs infrastructure, and Grok also needs infrastructure.
Regardless of which AI model ultimately wins, they will all need chips, data centers, networks, storage, and memory. In other words, AI companies are like people rushing to mine gold, and the companies that provide the infrastructure for AI are the ones selling shovels. And today I want to introduce you to an ETF that focuses on investing in these “shovel sellers” of the AI era.
This is the Roundhill Memory ETF with the ticker DRAM. So, what is a DRAM ETF? DRAM began trading on the stock market on April 2, 2026. This is the first ETF specifically targeting memory companies. The fund was created to invest in companies that manufacture and supply products such as DRAM, HBM, and NAND— critical components of AI systems and data infrastructure.
The thesis on which DRAM is based is very simple. It doesn't matter which AI company ultimately wins. DRAM is focused on one thing. As AI continues to rapidly develop and integrate into more and more areas of our lives, memory companies stand to benefit from the growing demand for AI infrastructure.
Before we look at this in more detail, let's take a look at the performance graph on the screen. DRAM trading began on April 2, 2026, so I use that date as a reference point for comparison. Since April 2026, DRAM has delivered a total return of about 111%, more than double that of the well-known technology ETF—XLK.
Its returns are also nearly three times higher than those of Katie Wood's actively managed ARK fund . Additionally, over the same period, DRAM significantly outperformed two other popular ETFs: QQQ and VOO. So why was DRAM able to grow so much more significantly than these other funds? To understand the reason, let's pay attention to the fund's assets .
The current DRAM portfolio is focused on companies such as Samsung Electronics (19.44% of the fund) and SK Hynix (16.75%). He also owns shares in Seagate Technology, SanDisk Corporation, and Western Digital. These are companies operating in the AI memory industry, and they have become some of the most profitable stocks in recent months.
But there is something very unusual about DRAM. Unlike a typical ETF, about 37 % of the fund's portfolio is invested in U.S. Treasury securities. So why does a growth-focused ETF hold such a large portion of U.S. government bonds? Because DRAM doesn't just buy stocks and hold them. The Fund uses equity swap agreements where U.S.
Treasury securities serve as collateral for these transactions. At the same time, these assets provide the fund with liquidity to rebalance the portfolio when necessary. This also helps reduce the impact of a sharp decline when stocks in the portfolio become significantly cheaper. However, despite its high performance, DRAM is not an ideal ETF.
There is one point that investors should pay special attention to. DRAM is an actively managed ETF. This means that DRAM's results depend largely on the team behind the fund. In particular, their ability to choose the right stocks and allocate capital effectively. And this is one of the biggest risks for DRAM.
Because in investing there is always one important question. Can an actively managed fund actually maintain its advantage over the long term? According to a study of the performance of actively and passively managed exchange-traded funds in the field of artificial intelligence, passive AI -ETFs that track an index outperformed actively managed AI- ETFs in the study sample.
This suggests that having skilled portfolio managers or people with deep AI knowledge does not necessarily mean that a fund will be a winner in the long run. And when we look at an example on the screen, it becomes even clearer. ARK is a very famous ETF from Katie Wood. This is an actively managed fund focused on innovative companies and technologies that have the potential to fundamentally change traditional industries.
However, over the past 10+ years, ARK has only provided a return of around 398%. Meanwhile, QQQ, a passive ETF that tracks the NASDAQ 100 index, delivered a significantly higher return—704%. This is an example of how even an actively managed fund with a very clear investment strategy and a team focused on high-performance technologies can struggle to maintain an edge over the long term.
And that's exactly the question we need to ask about D RAM. D RAM only started trading in April 2026. This means that the fund does not yet have enough history to prove that its current successes can be sustained for many years. The results so far are impressive, but we don't yet know whether the management team will be able to continue to identify promising companies in the next market cycles.
So just because D RAM is growing rapidly now doesn't mean it will necessarily continue to deliver high returns in the long term. There is another factor we must consider regarding D RAM—maximum drawdown. While D RAM has outperformed many other ETFs, the fund has also experienced very significant declines.
Over the same period from April 2026 to today, the maximum drawdown of D RAM reached about 46% . This is significantly higher than other growth-focused ETFs such as XLK , ARK , and QQQ . Just imagine that you invest $100,000. After a sharp drop, your account could be down to around $54,000. This is not a level of volatility that every investor can handle .
This means that while D RAM has the potential to generate very high returns, investors should also be prepared for periods when their accounts could fall very sharply. And this is especially important because D RAM is concentrated in a very narrow industry—the memory sector. And memory is a cyclical industry.
As investment in AI accelerates, demand for memory infrastructure also increases. This could allow DRAM to generate very strong growth during a favorable industry cycle. But when the cycle reverses, memory prices and profits for companies in the industry could also plummet. And as a result, DRAM sinkage can become very significant.
For those of you interested in the DRAM ETF, I think the most important question is will DRAM continue to rise? The real question is, how do you invest in DRAM? DRAM has a very clear investment thesis. AI is evolving. Data centers are expanding. The demand for computing power is growing. The demand for memory is growing.
And HBM is becoming increasingly important. And companies at the center of the memory supply chain can benefit from this trend. But there is a downside. DRAM has a very short history and is actively managed, so its performance depends heavily on the portfolio manager's ability to pick the right stocks.
The portfolio is also heavily concentrated in the cyclical memory sector. This means that the downturns could become very significant when the industry enters a difficult cycle in the future. For this reason, I don't think DRAM is an ETF that is well suited to a traditional buy-and- hold strategy. Instead, DRAM now makes more sense as part of an active investment strategy.
This means determining entry points based on price movement and technical analysis while adhering to very strict risk management. If you are interested in learning a trading approach to identifying quality entry points, you can join my course through a channel membership. It costs about the same as a few cups of coffee at Starbucks, so nothing too substantial.
Conclusion. Let 's go back to the California Gold Rush. Hundreds of thousands of people went in search of gold, but those who sold shovels made the most money. Then came the second industrial revolution . The economy needed more than just companies producing final products. She needed electricity. She needed steel.
She needed oil. And the companies that supplied these necessary resources became some of the biggest beneficiaries of this economic transformation. Today we are seeing a new version of the same story—the AI gold rush. Maybe Chat GPT will win. Maybe Gemini. Maybe Claude. Maybe Grok. Or maybe it will be a completely new name that hasn't even been released yet.
But no matter who wins, they will still need infrastructure. And within this infrastructure, memory is an integral piece of the puzzle. This is exactly what makes DRAM interesting. He doesn't try to predict which AI company will win. DRAM is just trying to do one very simple thing. Invest in companies that sell shovels to those looking for gold.
Comments 0