Micron Stock: Let’s Talk About Those Long-Term Contracts

Micron (MU) just reported blockbuster Q3 earnings. Revenue surged 4x year-over-year, earnings per share jumped 15x, and gross margins hit a massive 84.6%. The reasons are straightforward: an unprecedented AI buildout has driven explosive demand and pricing for high-bandwidth memory that sits alongside AI accelerators from the likes of Nvidia (NVDA) and AMD (AMD), with the supply crunch spilling into the broader memory market, too.

And now the three dominant suppliers are committing staggering capital to build new capacity, simultaneously, right now.

Sound familiar?

It should. This is exactly how every DRAM bust has started. There is always a narrative given for why this time is different. The outcome, so far, has always been the same. The latest narrative is that long-term customer contracts will tame the memory industry’s boom-and-bust cycle. However, there are important questions worth asking.

The Cycle, Explained

For decades, DRAM has followed a brutally predictable script. Demand surges. Suppliers race to build. New fabs come online simultaneously. Supply floods the market, prices collapse, margins go down deeply, and the industry spends years nursing losses before demand recovers.

Now look at the numbers. Samsung is committing roughly $73 billion to capex and R&D this year. SK Hynix is raising about $29 billion through a U.S. listing to fund new fabs and equipment. Micron has raised its fiscal 2026 capex forecast to $27 billion and expects even higher annual capital spending in fiscal 2027. Together, these three memory suppliers are deploying roughly $130 billion in a single year. Take a detailed look at how past memory cycles have played out for Micron

Prior cycle peaks saw the industry collectively spend $30 to $40 billion in a big year. This is three to four times that. Here is How low can Micron stock go if the cycle turns.

These are greenfield fab programmes – multi-year construction projects that take years to produce wafer output and cannot be reversed once underway. The capital is being locked in now, against demand projections that stretch years into the future.

The Contracts Are Real, But So Are the Questions

The bull case rests on something new: multi-year take-or-pay contracts that require customers to purchase committed volumes regardless of spot prices. Micron has signed 16 such agreements, 14 of which represent roughly $100 billion in minimum contracted revenue.

Once all planned agreements are completed, the company expects more than half of its revenue to fall under these contracts, with about 40% covered by fixed or ceiling pricing. For contracts with price bands, Micron says the floor prices support gross margins above any prior cycle.

But it does not eliminate risk; it redistributes it.

Roughly half of Micron’s revenue would still remain outside these agreements. More importantly, if AI infrastructure spending ultimately disappoints, or newer AI models require less memory than expected, customers could find themselves committed to purchasing capacity they no longer need. Enterprise AI deployment is proving more challenging than many expected, with workflow integration, governance, and organizational adoption remaining significant bottlenecks. If companies struggle to generate meaningful returns on AI investments, the pace of future AI infrastructure spending could slow, testing assumptions that today’s demand will persist for years.

When Certainty Becomes The Risk

When every supplier, every customer, and every analyst converges on the same narrative – tight supply beyond 2027 and AI demand that is structural and unbounded – consensus itself can become a warning sign. If reality comes in even modestly below the bull case, the adjustment can be sharp because nobody built in a cushion.

You might be thinking: Micron trades at just 8x estimated FY2027 earnings. See Micron valuation multiples vs. peers. That’s cheap. If the company earned that amount every year, investors would theoretically recoup their purchase price in about eight years. But there’s a catch. Those estimates assume a very lofty $148 per share in consensus earnings for FY’27, versus just $8 last year. The real question isn’t whether 8x is cheap. It’s whether earnings at that level are sustainable.

The cure for tight supply has always been the investment that tight supply incentivizes. No supplier can afford to sit out this expansion without risking market share. Ironically, that competitive dynamic has fueled previous memory cycles. The DRAM cycle has been declared dead before. The contracts are new. HBM is genuinely different. But neither changes the basic economics of supply eventually catching up with demand. The questions worth asking are the ones nobody is asking right now.

Opportunities like Micron highlight how individual semiconductor stocks can surge dramatically during technology transitions, but they also carry concentrated exposure to industry cycles, capacity expansion, and execution risk. A disciplined portfolio approach helps smooth these risks while still participating in long-term growth themes. The Trefis High Quality (HQ) Portfolio has consistently outperformed its market benchmark since inception, delivering cumulative returns of over 105%.