
Falling inventory at memory companies is usually a positive signal, but it does not prove that the industry cycle has already reversed. You need to determine whether the inventory decline comes from real demand recovery, deliberate production cuts, discount-driven inventory clearing, or accounting changes after inventory write-downs. A more reliable method is to examine days of inventory, DRAM and NAND average selling prices, bit shipments, product mix, and gross margin together. For investors tracking Micron, Samsung Electronics, SK hynix, and other memory companies, inventory decline is only the first step; ASP increases and sustained gross margin improvement are the real confirmation signals.

Falling inventory usually means supply-demand pressure at memory companies is easing, but you should not judge the cycle by the phrase “inventory is down” alone. The more important question is whether inventory is falling because customers are truly buying again, or because suppliers are cutting prices, reducing output, or lowering book value through inventory write-downs. Only when lower inventory comes together with rising ASP, stable or growing bit shipments, and improving gross margin does it look more like a healthy recovery signal.
Memory inventory is highly cyclical. DRAM, NAND, HBM, and enterprise SSDs are all affected by end-market demand, cloud procurement cycles, smartphone and PC replacement cycles, AI server investment, and supplier capital expenditure. When demand slows abruptly, memory companies can face high inventory and falling prices. When suppliers cut production, customers finish destocking, and AI or server demand recovers, inventory pressure can ease quickly.
But falling inventory can mean very different things. You should distinguish at least four situations:
| Reason for Inventory Decline | ASP Trend | Bit Shipments | Gross Margin Impact | Cycle Implication |
|---|---|---|---|---|
| Real demand recovery | Rising or stabilizing | Stable growth | Improves | Healthy recovery |
| Discount-driven clearing | Falling | Short-term growth | Under pressure | Still destocking |
| Production cuts | Uncertain | May decline | Improves later | Supply contraction |
| Inventory write-down | Not necessarily better | Not necessarily better | Current-period damage | Accounting adjustment |
Historically, memory companies often record inventory write-downs near the bottom of the cycle. In fiscal 2023, Micron wrote down work-in-process and finished goods inventories to net realizable value because of weak pricing, recognizing about $1.83 billion in inventory write-downs for the full year. That shows a decline in book inventory does not necessarily mean demand has truly recovered. When reading earnings reports, you should not equate “lower inventory” with “a new upcycle has begun.”
A better approach is to place inventory within the income statement and the pricing cycle. For example, Micron reported revenue of $41.456 billion and a non-GAAP gross margin of 84.9% in its third quarter of fiscal 2026, while management emphasized that DRAM and NAND industry demand was materially exceeding supply. This set of signals is much stronger than inventory decline alone, because it shows the company is not merely selling product; it is selling at higher prices and with a better product mix.
You also need to watch for the situation of “low inventory but high valuation.” Inventory usually falls during recovery and strong-cycle phases, but the stock market often prices in expectations early. If investors have already discounted several quarters of ASP increases, AI memory demand, HBM orders, and gross margin expansion, further inventory decline may not bring the same upside to share prices.
Summary: Falling inventory is a necessary signal of improvement in the memory cycle, but not a sufficient one. The more important signal is the combination of falling inventory, rising ASP, stable bit shipments, and improving gross margin. If inventory falls mainly because of discount-driven clearing, accounting write-downs, or production cuts, the cycle may still be near the bottom or in the early repair phase. You should always analyze inventory alongside pricing, shipments, gross margin, and management guidance, rather than making a judgment based only on inventory value.

When assessing inventory pressure at a memory company, days of inventory is usually more important than inventory value. Inventory value can be affected by production cost, product mix, currency effects, advanced process spending, and a rising share of high-value products. A company’s inventory value can increase without necessarily implying deterioration; if revenue and cost of goods sold are growing faster, days of inventory may actually fall.
The two basic formulas are:
Inventory value answers the question: “How much inventory is on the balance sheet?” Days of inventory answers: “How many days of sales can this inventory roughly support?” For memory companies, the second question is closer to the real supply-demand pressure.
In Micron’s third-quarter fiscal 2026 earnings materials, the company disclosed ending inventory of $8.6 billion and days of inventory of 120 days, while noting that DRAM inventory was very tight and below 120 days. This detail matters. If you only look at the $8.6 billion inventory figure, the inventory level may still appear large. But once you combine it with revenue growth, cost of goods sold, and days of inventory, it becomes clearer that DRAM supply is already tight.
Inventory should also be separated into finished goods, work in process, and raw materials:
| Inventory Category | Increase May Mean | Decline May Mean | What Investors Should Watch |
|---|---|---|---|
| Finished goods | Weak sell-through or pre-build | Better sales or tight supply | Whether products are sold at discounts |
| Work in process | Expansion or longer process cycle | Slower production pace | Future supply release |
| Raw materials | Pre-purchasing supply | Reduced production or procurement | Utilization rate |
For DRAM and NAND companies, work-in-process inventory should not automatically be seen as a negative signal. Advanced process nodes, HBM packaging, high-capacity DDR5, and enterprise SSDs all have longer production cycles. When high-value products account for a larger share, work-in-process and materials inventory may rise. By contrast, if finished goods inventory keeps increasing while ASP declines and shipments are weak, demand-side pressure is more likely.
Samsung Electronics offers another useful example. In its first-quarter 2026 results, Samsung reported Device Solutions revenue of KRW 81.7 trillion and operating profit of KRW 53.7 trillion. The Memory business delivered record quarterly sales and profit, supported by high-value AI demand, limited supply, and industry price increases. Even if absolute inventory value is affected by business scale and product mix, inventory total alone cannot fully explain the cycle position.
Another often-missed distinction is that supplier inventory and customer inventory are not the same thing. A chipmaker’s inventory may fall because customers are buying again. Customer inventory may fall because end demand is finally consuming old stock. Cloud inventory may fall because server orders are restarting. In many cases, downstream customers finish destocking first, and only then does inventory pressure at memory suppliers ease more visibly.
Summary: When analyzing inventory, the priority should be days of inventory, finished goods inventory, customer inventory, and then absolute inventory value. Inventory value can be distorted by product prices, cost structure, and high-end product mix, while days of inventory better reflects turnover efficiency. Falling finished goods inventory alongside rising ASP is usually a healthy signal. Falling finished goods inventory while ASP continues to decline may simply mean discount-driven destocking. You should avoid judging the cycle by balance-sheet inventory alone and instead combine inventory mix with management commentary on customer inventory.

Whether falling inventory becomes a strong recovery signal depends on whether ASP rises at the same time. ASP means average selling price, and it directly affects memory company revenue and gross margin. If suppliers sell down inventory by cutting prices, inventory can fall without improving profitability. If ASP rises, shipments remain stable, and product mix improves at the same time, destocking can translate into real earnings leverage.
Memory company revenue can be simplified as:
Revenue change ≈ Bit shipment change + ASP change + Product mix change
You can use the table below to judge the quality of destocking:
| ASP Trend | Bit Shipments | Typical Situation | Revenue and Profit Implication |
|---|---|---|---|
| ASP up | Shipments up | Strong recovery | Revenue and profit improve together |
| ASP up | Shipments down | Tight supply | Profit improves, but demand needs validation |
| ASP down | Shipments up | Discount-driven destocking | Revenue may not be strong; gross margin pressured |
| ASP down | Shipments down | Downcycle | Cycle still deteriorating |
Micron’s latest numbers are a useful example. In its third-quarter prepared remarks, Micron said DRAM revenue increased 67% quarter over quarter, with bit shipments up in the low single-digit range and prices up in the low 60% range. NAND revenue increased 99% quarter over quarter, with bit shipments up in the mid-single-digit range and prices up in the mid-80% range. This shows that revenue growth was driven mainly by sharp ASP increases and product mix improvement, not simply by shipping more bits.
ASP should also be analyzed by layer. Spot prices, contract prices, and company ASP are not identical:
| Pricing Indicator | What It Reflects | Advantage | Limitation |
|---|---|---|---|
| Spot price | Short-term marginal transactions | Moves quickly | Limited representativeness |
| Contract price | Large-customer procurement price | Closer to mainstream orders | Adjusts more slowly |
| Company ASP | Actual sales mix | Closest to reported earnings | Affected by product mix |
TrendForce’s DRAM price trend indicated that DRAM contract prices continued to rise in June 2026, with low inventory and strong restocking demand pushing price expectations higher. This signal suggests that the price increase is not only a product-mix effect inside one company’s earnings report, but also reflects tight industry supply-demand conditions.
Still, you should not apply one pricing logic to DRAM, NAND, and HBM. Traditional DRAM is more affected by PC, smartphone, and server memory demand. NAND is also influenced by enterprise SSDs, consumer storage, QLC SSD adoption, and supplier capacity discipline. HBM has a different profile, with customer qualification cycles, advanced packaging capacity, AI GPU platform dependence, and long-term order characteristics. Strong HBM pricing does not mean every DRAM product is strong; strong enterprise SSD demand does not mean consumer NAND has fully recovered.
For ordinary investors, tracking ASP does not require monitoring every product quote. But you should at least follow three things: management’s quarter-over-quarter pricing commentary for DRAM and NAND, third-party contract price commentary, and the company’s next-quarter gross margin guidance. If all three point in the same direction, the ASP trend is more credible. If only spot prices rise while company gross margin does not follow, caution is needed.
Summary: ASP is the bridge between inventory and profit. Falling inventory only tells you that supply-demand pressure may be easing; rising ASP tells you that suppliers are regaining pricing power. A healthy memory recovery usually shows up as customer restocking, lower days of inventory, rising DRAM and NAND ASP, stable bit shipments, and higher gross margin. If inventory falls while ASP keeps declining, the company may still be clearing inventory at the expense of profitability. You should analyze pricing and shipment volume together, not inventory direction alone.
Gross margin is the final income-statement confirmation of changes in inventory and ASP. For memory companies, rising gross margin usually means higher pricing, better utilization, lower unit cost, or a larger share of high-end products. But gross margin can also be affected by inventory write-downs, the sale of previously written-down inventory, and a higher HBM mix. So you should not only look at one quarter’s gross margin level; you should judge whether the improvement is sustainable.
Memory company gross margin is mainly driven by five factors:
| Gross Margin Driver | Sustainability | Indicators to Watch |
|---|---|---|
| ASP increases | Medium to high | DRAM/NAND contract prices |
| Process cost reduction | High | Cost per bit, yield |
| Utilization recovery | Medium | Production cuts, utilization rate |
| Higher high-end product mix | High | HBM, server DRAM, enterprise SSD |
| Inventory accounting impact | Low | Write-downs, sale of written-down inventory |
Micron reported a non-GAAP gross margin of 84.9% in its third quarter of fiscal 2026 and guided for a fourth-quarter gross margin of about 86%. This level of gross margin shows that the industry is clearly no longer at the bottom of the cycle; it is closer to a high-profit stage. The drivers are not only higher prices, but also AI data center demand, HBM, high-capacity DDR, enterprise SSDs, and favorable product mix.
Inventory write-downs can complicate gross margin analysis. In Micron’s fiscal 2023 full-year results, the company explained that write-downs to net realizable value first increased cost of goods sold in the period of recognition. Later, when previously written-down inventory was sold, it could reduce subsequent cost of goods sold. In other words, some quarters of gross margin improvement may partly reflect accounting timing, not only true pricing power recovery.
High-end product mix can also change overall gross margin. In its first-quarter 2026 results, SK hynix reported quarterly revenue of KRW 52.5763 trillion and operating profit of KRW 37.6103 trillion, both record highs, driven mainly by increased sales of high-value products supported by AI demand. For companies like this, HBM and server products may lift overall profitability significantly, even if some consumer DRAM or NAND categories are not equally strong.
Samsung also needs to be broken down by business. Samsung’s 1Q 2026 financial data showed a consolidated gross margin of 61.2% and an operating margin of 42.8%. But Samsung spans memory, foundry, smartphones, displays, and consumer electronics. Group-level gross margin cannot be treated as memory gross margin. You should prioritize the DS segment, Memory business commentary, ASP commentary, and the share of high-value products.
When using gross margin to judge the cycle, ask three questions:
If the answer is yes to all three, the cycle is likely still moving upward. If gross margin is already very high, ASP growth is slowing, days of inventory are rising, and capital expenditure is accelerating, you should begin watching for cycle-top risk.
Summary: Gross margin is the financial confirmation signal for the memory cycle, not a standalone leading indicator. It tells you whether falling inventory and rising ASP are truly turning into profit, but it can also be influenced by inventory write-downs, high-end product mix, and utilization rates. High gross margin is not itself a problem; the risk is that the market may have already priced in the continuation of high margins while future price increases, customer procurement, and capital expenditure begin to change at the margin. You should use gross margin together with ASP, days of inventory, and product mix.
As of mid-2026, the memory industry no longer looks like it is at a cycle bottom. It looks closer to a high-profit stage after a strong upturn. Supplier inventory is low, DRAM and NAND ASP have risen sharply, AI data center demand is strong, and companies such as Micron and SK hynix have reported much higher profitability. The more important question now is not whether inventory can be cleared, but whether price increases can continue, when new capacity will arrive, and whether rising capital expenditure is already planting the seeds for the next supply cycle.
You can divide the memory cycle into four stages:
| Cycle Stage | Inventory | ASP | Gross Margin | Capital Expenditure | Investment Implication |
|---|---|---|---|---|---|
| Early downcycle | Rises quickly | Starts falling | Declines | Still high | Earnings expectations cut |
| Cycle bottom | High or peaking | Declines slow | Low | Cut sharply | Watch for price stabilization |
| Recovery | Keeps falling | Turns upward | Improves quickly | Recovers cautiously | Earnings leverage releases |
| High-cycle phase | Low or tight | High and rising | High profit | Expands meaningfully | Watch for marginal changes |
Many current data points look closer to the stage between late recovery and high-cycle strength. Micron has said DRAM and NAND supply-demand tightness is expected to persist beyond 2027 and that AI-driven memory and storage demand is changing the industry’s structure. The company has also signed 16 strategic customer agreements involving long-term supply, pricing ranges, and customer funding commitments. These strategic customer agreements improve revenue visibility and may reduce some of the extreme volatility traditionally associated with the memory cycle.
But high-cycle phases have their own risks. First, the slope of price increases may slow. Second, customers may reduce memory configurations, delay orders, or seek alternatives if memory prices become too high. Third, high profitability may encourage suppliers to increase capital expenditure, eventually bringing new supply back to the market. Fourth, HBM, DDR5, and enterprise SSD demand may diverge, and overall gross margin may hide changes in individual product categories.
If you track U.S.-listed memory companies, you can use Biya U.S. stock information search to place Micron, Western Digital, Seagate, Pure Storage, NetApp, and other companies into the same framework. For chip manufacturers, watch inventory, ASP, and gross margin. For HDD companies, watch nearline demand, cloud long-term agreements, and utilization. For enterprise storage system companies, focus more on orders, ARR, service revenue, and software-driven profitability.
When cycle analysis turns into trading decisions, you should not only look at stock price moves. Memory stocks can be highly volatile around earnings, and actual trading costs may affect your position sizing and rebalancing experience. U.S. stock trading costs usually include more than commissions; they may also include platform fees, external agency fees, transaction activity fees, and other charges. If the service is available in your region and you meet the applicable requirements, Biya U.S. stock trading fees state that U.S. stock trading commission is $0, while platform fees, external agency fees, and other charges are subject to the fee schedule and order page. The fee structure is not investment advice. Before trading, you still need to consider order types, volatility risk, and your own risk tolerance.
You can also monitor these potential inflection signals:
Summary: The inventory–ASP–gross margin framework helps you judge where the memory cycle stands. Inventory reflects supply-demand pressure, ASP reflects pricing power, and gross margin verifies earnings quality. In mid-2026, the key issue in the memory industry has shifted from “Can inventory fall?” to “Can high prices and high profitability last?” When inventory starts to rise again, ASP growth slows, gross margin guidance stops improving, and capital expenditure accelerates, you should become more alert to cycle-top risk and valuation pullbacks.
If you continue to track the memory supply chain, the priority is not to chase one quarter of earnings surprise, but to build a quarterly review sheet: days of inventory, DRAM ASP, NAND ASP, HBM progress, gross margin, capital expenditure, and next-quarter guidance. When using Biya to follow relevant U.S. and Hong Kong stocks, you should also include trading costs, order-page fee disclosures, and market volatility in your decision process. Service availability depends on the user’s location, identity verification results, platform rules, and applicable laws and regulations. Public market information and fee structures are for research reference only and do not constitute investment advice. For mobile tracking of prices and trading rules, you can also use the Biya app.
It usually means cost of goods sold or revenue is growing faster than inventory, so it does not necessarily indicate worsening inventory pressure. For memory companies, advanced processes, HBM, enterprise SSDs, and a higher share of high-value products can all increase inventory value. You should also review finished goods inventory, ASP, bit shipments, and management commentary on customer inventory.
Customer inventory decline usually reflects demand recovery earlier, because downstream customers first consume old inventory and then resume purchases from chipmakers. Chipmaker inventory decline is often the next stage of that recovery. But the timing differs across end markets, so AI servers, PCs, smartphones, and automotive electronics should not be analyzed as one single cycle.
Contract prices are better for judging the main memory cycle trend, while spot prices are more useful for short-term marginal changes. Large DRAM customers typically buy through contract pricing, which better reflects supplier pricing power. Spot prices move quickly, but trading volume and representativeness are limited, so they should be verified against company ASP and gross margin.
Yes, that can happen. HBM has high unit value, tight supply, and long customer qualification cycles, so it can significantly lift overall revenue and gross margin even if some consumer DRAM or NAND categories remain weak. You should separately review HBM revenue share, traditional DRAM ASP, NAND pricing, bit shipments, and segment profitability.
No. Inventory write-downs may temporarily reduce cost of goods sold when previously written-down inventory is sold, helping gross margin in later quarters. But sustainable improvement still depends on ASP, demand, utilization, and cost structure. If prices continue to fall, the company may still face profit pressure.
Rapid capital expenditure growth is one potential cycle-top signal, but it is not enough by itself to prove the cycle is ending. You also need to consider when new capacity comes online, whether HBM is absorbing traditional DRAM capacity, customer long-term agreements, technology migration difficulty, and demand growth. If supply expands faster than real demand, the next round of inventory pressure may return.
*本文仅供参考,不构成 BiyaPay 或其子公司及其关联公司的法律,税务或其他专业建议,也不能替代财务顾问或任何其他专业人士的建议。
我们不以任何明示或暗示的形式陈述,保证或担保该出版物中内容的准确性,完整性或时效性。



