Morgan Stanley’s 89.3% Micron Margin Forecast Rests on Scarcity, Not HBM4 Alone

• Morgan Stanley raised its Micron fiscal-2027 gross-margin estimate to 89.3%, but that is an analyst forecast, not company guidance.
• Micron’s latest margin gain came mainly from price: DRAM prices rose in the low-60% range sequentially while bit shipments grew only in the low single digits.
• Long-term contracts can support margins, yet price ceilings, rival HBM4 suppliers, demand destruction and new capacity all limit the case for a permanent reset.

Morgan Stanley’s forecast puts Micron close to retaining 90 cents of gross profit from each dollar of fiscal-2027 revenue. The evidence supports a formidable scarcity trade. It does not yet show that HBM4 or customer contracts have abolished the memory cycle.

A near-90% forecast built mostly on price

Morgan Stanley analyst Joseph Moore raised his Micron price target to $1,200 from $1,050 and kept an Overweight rating. His revised model lifts fiscal-2027 revenue to $266.867 billion from $211.737 billion, gross margin to 89.3% from 86.3%, and earnings per share to $168.52 from $122.05, according to a report on the note. Another summary says Moore expects AI to keep DRAM demand well above supply beyond 2027.

Those are Morgan Stanley estimates, not Micron guidance. Micron reported an 84.9% non-GAAP gross margin for the 13 weeks ended May 28, 2026, and guided to approximately 86% on the same basis for its 14-week fiscal fourth quarter. The company’s prepared remarks attribute the latest quarter’s 10-percentage-point sequential improvement primarily to higher pricing, with execution and product mix also helping.

The price-volume split is the clearest test of what drove the result. Sequential DRAM revenue rose 67% as average selling prices increased in the low-60% range and bit shipments only in the low single digits. NAND revenue rose 99% as prices increased in the mid-80% range and bit shipments in the mid single digits. On a year-over-year basis, Micron said DRAM prices rose in the low-260% range and NAND prices in the mid-310% range. This was much more a repricing of existing supply than a surge in physical volume.

The historical margins below are deliberately not presented as a continuous series: they mix quarterly and annual periods, GAAP and non-GAAP results, company guidance and an analyst estimate. They establish the size of the swing, not a like-for-like trend.

Micron described fiscal 2018 as a record year and reported a 61% fourth-quarter GAAP margin in those results. By fiscal 2023, its annual filing showed a negative 9% gross margin, down from 45% a year earlier, after DRAM and NAND prices fell, inventory was written down and underused facilities added $382 million of cost. That filing also said customers were generally reluctant to sign long-term fixed-price purchase contracts. The new agreements are therefore a meaningful change in commercial structure—but they have not yet been tested through a downturn.

Contracts create a floor, a ceiling and new liabilities

Micron says it has signed 16 strategic customer agreements across data-center, consumer and automotive markets. They represent roughly 20% of DRAM volume and one-third of NAND volume over their terms. Most run from calendar 2026 through 2030; automotive contracts generally run for three years. When its planned agreements are complete, Micron expects about half or more of company revenue to be covered.

The take-or-pay contracts bind customers to specified volumes. Most use fixed prices or price bands, while a minority leave prices to market conditions. Micron says floors in the banded agreements would still produce gross margins well above the peak quarter of any past cycle. That remains a company projection, not an observed recession or oversupply result.

The accounting is narrower than a headline revenue commitment. Micron reported about $5 billion of remaining performance obligations at the May 28 quarter-end. Including agreements signed after that date, it put the figure at approximately $100 billion, based on minimum committed volumes and minimum prices. The company explicitly said this is not total expected revenue and excludes agreements without fixed prices or bands.

Micron also projects $22 billion of deposits and related financial commitments from signed agreements, including about $18 billion in cash deposits. Those cash flows are financing, do not increase free cash flow and are to be returned to customers later in the contract terms. By contrast, Morgan Stanley raised its fiscal-2027 free-cash-flow estimate to $140 billion from $104 billion, according to the analysis. The bank’s figure therefore depends on operating assumptions, not on treating deposits as earnings.

The largest contracts generally cap existing-product prices around second-calendar-quarter 2026 market levels. When all targeted agreements are executed, Micron expects fixed prices or ceilings at or near those levels to cover about 40% of revenue. Customers receive supply assurance and protection against further price spikes; Micron receives minimum volumes and a floor but surrenders some shortage upside.

Micron’s own quarterly filing adds the legal and operating counterweight. The contracts may constrain available supply, reduce the company’s flexibility as market conditions change and lead to disputes if customers do not honor their commitments. A contract can redistribute cyclical risk; it cannot eliminate counterparty, allocation or execution risk.

TrendForce’s February 2026 estimate of global HBM bit output shares for Micron, SK hynix and Samsung in 2024, 2025 and 2026F. Source: TrendForce HBM4 Analysis.

HBM4 is valuable, but not exclusive

Micron says its 12-high HBM4 volume ramp is running twice as fast as the comparable HBM3E ramp and that it has already shipped more than $1 billion of HBM4 revenue. Yet neither Micron nor Morgan Stanley disclosed an HBM4 product margin, and the company’s own price-volume data show broad DRAM and NAND pricing—not HBM4 alone—drove the record companywide margin.

The competitive evidence also narrows the differentiation claim. A February industry analysis expected Nvidia to use Samsung, SK hynix and Micron for its Rubin platform because no single supplier could meet its HBM4 requirements. At that point it described Micron’s validation progress as relatively slower than its two rivals and expected all three to complete validation by the second quarter of 2026.

Samsung separately said it had begun HBM4 mass production and shipped commercial products. It also projected that its HBM sales would more than triple in 2026 from 2025 and said it was expanding HBM4 capacity. Those are Samsung claims, but they establish that Micron does not control the supply response.

HBM’s economics cut both ways. Micron says HBM requires more wafers and cleanroom space than conventional DRAM to produce the same number of bits on the same node, helping constrain non-HBM supply while demand is strong. The same filing warns that if HBM demand weakens and suppliers redirect capacity to conventional DRAM, the resulting supply increase could pressure prices. Qualification, yield, packaging and allocation determine who captures HBM4 value; the product category itself does not guarantee durable margins.

New capacity supports the shortage—until it does not

Micron estimates fiscal-2026 capital spending of approximately $27 billion, net of government incentives. Its first Idaho fab is projected to produce initial DRAM wafers in mid-2027, a second Idaho fab in late 2028 and its first New York fab in 2030 and beyond. A Singapore HBM packaging facility and the acquired Tongluo fab in Taiwan are both expected to add meaningful capacity from 2027.

That schedule helps explain Morgan Stanley’s fiscal-2027 scarcity assumption: much of Micron’s greenfield wafer supply arrives at the end of that fiscal year or later. It also identifies the eventual reversal mechanism. Micron says its competitors are expanding too, and warns that state-backed investment—including by China and affiliated companies such as CXMT and YMTC—could create DRAM and NAND oversupply.

Public support changes the private economics of that buildout. Micron has agreements for up to $6.4 billion of U.S. CHIPS Act grants and says qualified U.S. semiconductor-manufacturing investment receives a 35% investment tax credit. The incentives depend on milestones and can be reduced, terminated or clawed back. The company also lists construction labor, specialized equipment, electrical power, water and permitting among the constraints that can delay fabs or raise their cost.

Scarcity can weaken demand before new fabs arrive. Micron warns that sustained high memory prices may cause customers to cut purchases, redesign products with less memory or seek alternatives. It also says data-center buildouts depend on energy, water, capital and local acceptance. Those constraints matter because roughly half of Micron’s fiscal-2025 revenue came from the data-center end market: the buyers paying the “memory tax” must still finance and power the systems around it.

What would resolve the durability question

The next results need to show whether contracts and product mix can replace price inflation as the source of margin strength.

• Micron must approach the 89.3% estimate over a full fiscal year on a consistent accounting basis; one quarter near 90% would not establish a new floor.
• Price, bits, cost and mix must separate. The latest quarter was dominated by price, while Micron expects the rate of price increases to moderate and blended DRAM cost per bit to rise as advanced products and new capacity ramp.
• Contract disclosures need to distinguish revenue protected by floors, revenue limited by ceilings, market-priced volume and actual customer performance during weaker demand.
• HBM4 qualifications, yields and allocations must hold as Samsung and SK hynix compete and expand supply.
• New fabs and packaging lines must start on time without raising costs faster than revenue—or producing the oversupply that ended earlier booms.

Morgan Stanley’s 89.3% forecast is credible as a scarcity case: Micron enters fiscal 2027 with record pricing, unusually firm customer commitments and capacity that takes years to build. A durable regime change requires different proof. Margins must remain exceptional after price increases cool, competitors qualify more HBM4 and the first wave of new supply reaches customers.