Latest / Stock Picking with Fexingo: Individual Equities, Research, and Building a Concentrated Portfolio / How Micron Is Benefiting from the AI Memory Boom
Transcript
- Lucas: Luna, I want to talk about a stock that doesn't get nearly the attention it deserves right now, even though it's sitting in the middle of one of the biggest technology waves we've seen in decades. Luna: I'm guessing you're not talking about NVIDIA again, because we've covered that one a few times. Lucas: No, not NVIDIA. I'm talking about Micron Technology, ticker M-U. The memory chip maker. And I think the AI narrative around it is actually more durable than most people assume. Luna: Micron — that's DRAM and NAND flash, right? The stuff that goes into every computer and phone. But what's the AI angle there? Memory isn't computing. Lucas: Right, but the thing is, AI models — especially the large language models that everyone's been deploying — they consume an enormous amount of memory. Each time you run an inference, the model parameters have to be loaded into memory. So the more models you deploy, the more DRAM you need. And Micron is the only us based company that manufactures DRAM at scale. Luna: Okay, so it's an enabler play. Sort of like how NVIDIA sells the shovels, Micron sells the buckets. Lucas: Exactly. And the numbers back it up. In their most recent quarter — which they reported in March — Micron posted revenue of about $8.6 billion, up 58 percent year-over-year. That's not a one-time spike. Data center revenue alone more than doubled, driven by high-bandwidth memory, or HBM, which is a specialized type of DRAM that's optimized for AI workloads. Luna: Fifty-eight percent growth — that's impressive. But memory has always been cyclical. How do we know this isn't just another upswing that's going to crash when supply catches up? Lucas: That's the million-dollar question. And historically, memory has been brutally cyclical. But I think the structure of this cycle is different. For one, the supply side is much more disciplined. The three major players — Samsung, SK Hynix, and Micron — have been very careful about adding capacity. They've learned from the overinvestment that crushed prices in 2018 and 2023. And secondly, the demand driver this time is structural, not just a consumer upgrade cycle. AI data center buildouts are multi-year projects. Luna: So you're saying this cycle could last longer than the typical 18 to 24 months. Lucas: I think so. And Micron's guidance for the current quarter — which we're in now, June 2026 — calls for revenue around $9 billion, plus or minus. That would be another sequential increase. The company's CEO, Sanjay Mehrotra, has described this as the 'most exciting period in the company's history.' And the market seems to agree — the stock is up about 40 percent year to date, even with the broader market up around 10 percent. Luna: Forty percent year to date? That's significantly outperforming the S&P 500, which is near 7,500. Lucas: It is. And yet the stock still trades at about 18 times forward earnings. That's not expensive for a company growing revenue at nearly 60 percent. Compare that to NVIDIA, which trades at over 30 times forward earnings even after its own run-up. So you could argue there's room for multiple expansion if Micron sustains this growth. Luna: But what about the risks? I mean, there's always a 'but' with memory stocks. Lucas: There are definitely risks. Number one: NAND flash pricing has been soft lately. That's the storage side of their business, and it's not growing as fast as DRAM. Number two: geopolitical exposure. Micron gets about 15 percent of its revenue from China, and trade tensions could flare up again. And number three: the cyclical risk is still real. If AI demand disappoints — or if hyperscalers like Amazon and Google decide to design their own memory solutions — the whole thesis breaks down. Luna: And what about competition? Samsung and SK Hynix are giants. Micron is the number three player. Can they really hold their own in HBM? Lucas: Actually, Micron has a technical edge in HBM right now. Their HBM3E — that's the latest generation — is already in production and has been qualified by NVIDIA for their H200 and B100 GPUs. They were the first to deliver a 24-gigabyte HBM module. So they're not just following; they're leading in a key segment. And because HBM is so complex to manufacture, it's a higher-margin product that provides a buffer against commodity pricing. Luna: That's a good point. The higher margins come from the packaging complexity. So it's not just a commodity play anymore. Lucas: Exactly. And that's why I think Micron deserves a fresh look. The market still thinks of it as a cyclical memory maker, but the reality is shifting toward a structural growth story. For stock pickers who are looking beyond the obvious AI names, this is a compelling candidate. Luna: Before we go deeper, I want to mention something. You know, conversations like this — where we dig into a company's fundamentals and challenge the conventional wisdom — that's exactly the kind of content that keeps this show ad-free. And if you find value in these episodes, you can support us by buying us a coffee. It's at buy me a coffee dot com slash fexingo. No pressure, but every bit helps us keep doing this. Lucas: Absolutely. And we're grateful to everyone who's already contributed. It really does make a difference. Okay, back to Micron. One thing I want to highlight is the company's balance sheet. They have about $9 billion in cash and short-term investments, against $13 billion in debt. Net debt is manageable, and free cash flow turned positive last quarter after a period of heavy capital spending. That's a sign that the investment cycle is starting to pay off. Luna: So they're not just growing revenue; they're generating cash. That's a good sign for the sustainability of the dividend, too. I think Micron pays a small dividend, right? Lucas: They do — it's about 25 cents per share per quarter, which yields roughly 0.5 percent. Not a big income play, but it shows management is confident in the cash flow trajectory. And they've been buying back shares as well. In the last quarter, they repurchased about $200 million worth of stock. Luna: Let's talk about valuation one more time. You said 18 times forward earnings. What's the growth rate implied by that multiple, and how does it compare to history? Lucas: Historically, Micron has traded at a discount to the broader market because of its cyclicality — sometimes as low as 6 or 7 times earnings during downturns. But during upcycles, it can trade at 15 to 20 times. So 18 times is toward the high end of the historical range. But the key question is whether this earnings growth is sustainable. If earnings per share can grow from the current $8.50 trailing to $11 or $12 over the next year, then the forward multiple drops to around 15. That's not expensive. Luna: So the bull case relies on earnings continuing to grow at a rapid pace. What could derail that? Lucas: The biggest risk is a slowdown in AI data center spending. If the hyperscalers pause or reduce their capex, DRAM demand could soften quickly. Also, there's the possibility of a trade war escalation that cuts off Micron's access to China or other markets. And then there's the commodity risk — if Samsung and SK Hynix decide to ramp up production aggressively, price declines could squeeze margins. Luna: But isn't that less likely now because of the capital intensity? I mean, building a new fab costs tens of billions of dollars. Lucas: Exactly. Entry barriers are higher than ever. And the three major players have signaled discipline. So while the risk is still there, it's lower than in previous cycles. That's why I think Micron is worth considering for a concentrated portfolio. It's not a safe stock, but it's a high-conviction one. Luna: Alright, Lucas. So if a listener wanted to buy Micron today, what's the right price? Or are you not giving price targets? Lucas: I don't give price targets, but I can tell you what I look at. I'd want to see the stock pull back to around $125 or $130 before adding a position. At today's price near $140, it's not a screaming bargain, but it's not overvalued either. For a long-term investor, dollar-cost averaging in over time makes sense. The key is to have conviction in the structural demand story and be willing to hold through volatility. Luna: That's a fair approach. And one last thing — how does Micron compare to Intel? Intel has been struggling, but they also have memory exposure through their Optane business, which they shut down. So they're not really a competitor. Lucas: No, Intel is a completely different story. They have a foundry business, a CPU business, and they exited the memory space. Micron is pure-play memory. So if you want exposure to the memory cycle, Micron is the way to go. Intel is more of a turnaround story with execution risk. Luna: So Micron is the better play for AI memory demand, while Intel is for a bet on manufacturing recovery. Got it. Lucas: Exactly. And that's the kind of distinction that active stock picking can uncover. In a world where index funds are crowding into the same mega-cap names, finding differentiated plays like Micron is where we can add value. Luna: Love it. Thanks, Lucas. That's a great deep dive on a stock that doesn't get enough love. For our listeners, if you want to hear more episodes like this, remember you can support us at buy me a coffee dot com slash fexingo. And as always, do your own research before making any investment decisions. Lucas: Thanks, Luna. And thanks to all of you for listening. See you next time.