On May 26, 2026, Micron Technology’s market capitalization surpassed $1 trillion for the first time, with its share price jumping more than 19% in a single day. The direct catalyst behind this historic milestone was UBS raising Micron’s target price dramatically from $535 to $1,625.
However, the way this target price was raised reveals a deeper shift beyond just price movements—investors are fundamentally rethinking how to value a memory chip company. The logic behind shifting from a "cyclical stock" to an "AI growth stock" hinges on whether three structural changes can reinforce each other.
Why the Valuation Framework Is Moving from Sum-of-the-Parts to Discounted Earnings
According to X user XinGPT’s analysis, UBS previously used a Sum-of-the-Parts (SoTP) valuation for Micron, splitting the company into two segments: the HBM business, valued at 6x P/S based on projected 2027 revenue of about $27.89 billion (roughly $132 per share); and the core DRAM+NAND business, valued at 3x P/S on projected 2027 revenue of about $187.7 billion (about $405 per share). This approach assumes Micron is fundamentally still a highly cyclical company, with HBM simply growing faster and deserving a higher multiple, so the two segments are valued separately.
The new framework, however, adopts roughly a 15x next-twelve-months (NTM) P/E, anchored on an estimated 2029 EPS of about $117, discounted back to 2028 at a roughly 12% cost of equity. The choice of 2029 is significant: UBS’s model already assumes the DRAM market will experience a moderate downturn by then. If Micron can still maintain EPS above $100 at that point, it demonstrates true "through-cycle" earnings power, not just a temporary peak.
This shift in valuation approach essentially moves the benchmark from "revenue structure" to "earnings stability." When a company’s profit volatility is structurally reduced, it becomes more logical to value it using a unified earnings multiple rather than segment-based revenue multiples.
How Long-Term Agreements Are Changing Earnings Volatility in the Memory Industry
For this valuation shift to be justified, there must be a structural mechanism supporting the core assumption of "reduced earnings volatility." That mechanism is the emergence of enhanced long-term agreements (LTAs).
Traditional long-term agreements in the memory industry typically only lock in shipment volumes, with prices set by the spot market. These contracts usually last less than a year and lack strong enforcement. As a result, memory chip prices are highly sensitive to spot market swings, with peak-to-trough price differences reaching 50% to 60%. Profits soar in upcycles and deteriorate rapidly in downturns, making earnings unpredictable—a key reason why the market has long applied a "cyclical discount" to Micron and its peers.
The new generation of enhanced LTAs differs in three ways. First, contract terms now span 3 to 5 years. Second, pricing uses a dual-track model with both fixed and floating components. Third, buyers commit to prepayments and capex support, making it much harder to cancel contracts.
UBS’s supply chain research estimates that by 2027, about 20% to 30% of industry-wide DDR shipments will be covered by these enhanced LTAs—about 20% for Micron, 30% for Samsung, and 18% for SK Hynix. More importantly, hyperscale cloud providers have already secured roughly 60% to 70% of all industry server DDR5 shipments under contract. This means major buyers are locking in future memory supply years in advance, shifting their focus from price volatility to supply security, even at the cost of fixed pricing.
LTAs impact Micron’s earnings in two key ways. First, DDR price swings from peak to trough are expected to be cut roughly in half. Second, UBS estimates that even if DRAM spot prices fall about 50% in 2029, Micron’s EPS can still remain above $100. The earnings floor created by LTAs fundamentally improves Micron’s profit predictability.
How HBM Supply and Demand Is Transforming Revenue Structure
Another pillar supporting the valuation shift is the rapid expansion of Micron’s HBM business. In its fiscal Q1 2026 earnings report, Micron disclosed that HBM generated $2.4 billion in quarterly revenue, accounting for 18% of total sales, with gross margins well above the company average.
In terms of capacity, Micron’s HBM output for 2025 and 2026 is already fully sold out, with "unprecedented" long-term supply contracts signed with customers. Management has said they can currently meet only about 50% to 67% of key customer demand, and this supply-demand imbalance is expected to persist beyond 2026.
According to Counterpoint Research, SK Hynix will hold a dominant 57% HBM market share in Q4 2025, followed by Samsung at 22% and Micron at 21%. While Micron is still catching up in market share, overall HBM demand is growing far faster than the combined capacity expansion of all three players. In this environment, short-term market share isn’t the main constraint on valuation re-rating. The key is that HBM has shifted from being an "optional high-end product" to an irreplaceable core component for AI servers. For example, a single NVIDIA DGX GB300 server cabinet uses over 20TB of HBM. This scale of per-unit consumption makes HBM demand highly inelastic.
Is There a Sustainable Foundation for AI-Driven Structural Demand?
AI-driven demand for memory isn’t limited to HBM. As Micron’s CEO noted on the earnings call, server shipments are expected to grow 18% to 19% year-over-year in 2025, with rapid AI data center expansion systematically driving up demand for high-performance, high-capacity memory and storage products.
Looking at global supply, Micron expects the memory chip supply shortage to persist beyond 2026. This is fundamentally different from past cycles, where supply-demand swings were driven mainly by end-consumer electronics. The current driver is sustained capital expenditure on AI infrastructure, which is more rigidly planned and executed.
However, the term "structural" still requires validation on two fronts. First, the lag between capital investment and capacity expansion—Micron has announced plans to invest about $200 billion in expanding US-based capacity, but new fabs typically take over three years from construction to stable production, meaning large-scale capacity won’t come online until after 2028. Second, the sustainability of AI capex—while tech giants’ AI infrastructure spending is at historic highs, any future capacity overshoot or slowdown in capex growth could put pressure on memory prices and profits. Even so, the fixed-price component of LTAs would still provide an earnings buffer.
Does Differentiation in the Competitive Landscape Offer a Lasting Advantage?
Among the three memory industry giants, Micron has a unique position as the only major US-based memory manufacturer—a geographic advantage that stands out as global tech supply chains are being restructured.
On the policy front, Micron has received about $6.1 billion in US government funding under the CHIPS and Science Act and has broken ground on the largest wafer fab project in US history in Clay, New York. On the demand side, hyperscale cloud providers are locking in years of future memory supply through enhanced LTAs, and Micron’s role as the core North American supplier gives it some geographic leverage in customer relationships.
However, competition in the HBM market remains intense. TrendForce data shows SK Hynix leads with 57% HBM market share, with Micron at 21%, trailing Samsung. The timing of HBM4 mass production will be a key variable in the next phase of competition—Micron’s HBM4 is expected to enter mass production in the second half of 2026. If the technology ramp goes smoothly, Micron’s market share could rise further.
Key Validation Points and Potential Risks in the Valuation Re-Rating
UBS’s $1,625 target price essentially reflects both higher earnings forecasts and a higher valuation multiple: projected EPS of $155, $167, and $117 for fiscal years 2027, 2028, and 2029, respectively, discounted at about a 15x forward P/E, with cumulative free cash flow exceeding $400 billion.
This forecast depends on three core variables: whether the actual LTA execution rate and price lock-in meet expectations; whether HBM shipment growth can keep raising its share of revenue; and whether AI-related capex can maintain its current pace over the next two to three years.
It’s important to note that Micron’s current forward P/E is about 8.4x, while UBS’s 15x forward P/E assumption implies a significant multiple expansion. Achieving this expansion depends on whether the market accepts that "the memory cycle has been structurally tamed." If LTAs fail to deliver the expected price stability in practice, or if buyers default during a downturn, the logic behind the re-rating will be challenged.
Conclusion
Micron’s shift in valuation framework represents a systemic re-pricing of the memory chip industry’s business model. The widespread adoption of long-term agreements now provides a level of earnings stability that was previously unattainable. HBM’s central role in AI infrastructure offers structural growth for revenue, and multi-year order visibility underpins the case for a higher valuation multiple. Whether this re-rating logic holds ultimately depends on whether LTAs can actually deliver on the promise of dampening earnings volatility, and whether AI-driven demand proves resilient enough to carry the industry through a full memory cycle.
FAQ
Q: What does Micron’s shift from SoTP to overall P/E valuation mean?
A: The SoTP (sum-of-the-parts) approach assumes Micron is still a highly cyclical company, with faster growth in HBM that warrants separate treatment. Switching to an overall P/E valuation means the market now believes Micron’s earnings volatility has been structurally reduced, and the company’s underlying business model is moving from highly cyclical to more stable.
Q: What are the key differences between long-term agreements (LTAs) and traditional supply contracts?
A: Traditional contracts typically only lock in shipment volumes, with prices set by the market and relatively short terms. The new enhanced LTAs last 3 to 5 years, use a dual-track model with both fixed and floating pricing, and include buyer prepayments and capex support commitments, making breaches much more costly.
Q: What could undermine the stabilizing effect of LTAs on earnings?
A: If the actual fixed pricing achieved is lower than expected, or if there are large-scale buyer defaults during a downturn, or if the proportion of shipments covered by LTAs falls short of market expectations, the smoothing effect on earnings will be limited.
Q: How does the expansion of the HBM business impact Micron’s earnings structure?
A: HBM products carry much higher gross margins than traditional DRAM and NAND. As of fiscal Q1 2026, HBM accounted for 18% of total revenue and is still rising. The growing share of HBM directly lifts overall profitability.
Q: What are the main points of contention around Micron reaching the $1,625 target price?
A: The main debate centers on whether LTAs can truly smooth earnings through a real downcycle. Other concerns include the yield ramp challenges in HBM capacity expansion, the sustainability of AI capex, and whether the valuation multiple can really expand from about 8x to 15x.
