--- title: "Morgan Stanley and J.P. Morgan Agree: Long-Term Agreements (LTA) Drive Revaluation of Memory Companies, Market Has Not Fully Priced It In" type: "News" locale: "en" url: "https://longbridge.com/en/news/286845100.md" description: "Memory chips are transitioning from cyclical commodities to strategic resources for AI infrastructure. Morgan Stanley and J.P. Morgan have successively stated that Long-Term Agreements (LTA) are rewriting the traditional cycle logic of \"price hikes - capacity expansion - price collapse\" into a new paradigm characterized by prepayments locking in volume, price floors, and visible profitability. Once the valuation framework shifts from PB to PE, industry leaders like SK Hynix could see a revaluation upside of over 38%, yet the market has not yet \"bought into\" this thesis" datetime: "2026-05-19T02:44:02.000Z" locales: - [zh-CN](https://longbridge.com/zh-CN/news/286845100.md) - [en](https://longbridge.com/en/news/286845100.md) - [zh-HK](https://longbridge.com/zh-HK/news/286845100.md) --- # Morgan Stanley and J.P. Morgan Agree: Long-Term Agreements (LTA) Drive Revaluation of Memory Companies, Market Has Not Fully Priced It In Memory chips are transforming from cyclical commodities into strategic resources for AI infrastructure—and the core mechanism driving this shift is known as the Long-Term Agreement (LTA). According to Zhuifeng Trading Desk, Morgan Stanley and J.P. Morgan recently released in-depth research reports on the global memory industry, both focusing on the same variable: Long-Term Agreements for memory supply, abbreviated as LTA. **In the past, the core narrative of the memory industry was price hikes, capacity expansion, oversupply, price declines, and subsequent clearing. Now, AI customers wish to lock in supply in advance, while memory manufacturers aim to secure orders and prices early, potentially dampening the amplitude of cyclical fluctuations.** Shawn Kim from Morgan Stanley's Technology Hardware team wrote in his research: "Memory has become a key bottleneck in AI infrastructure. LTAs signed by customers to ensure supply **are transforming a traditional cyclical business into one with guaranteed, high-margin, long-term revenue streams."** This statement summarizes the firm's core judgment: **If the market continues to value memory companies as traditional cyclical stocks, it may be underestimating the stability of future revenues and cash flows.** J.P. Morgan's Jay Kwon team offered a more direct assessment: **“LTAs pave the way for memory manufacturers toward a new valuation framework.”** Their logic is that buyers are not just afraid of price hikes, but more so of being unable to procure supply; sellers are not just looking to raise prices, but seek order visibility before committing to massive capital expenditures. LTAs embed the anxieties of both parties into contracts. What investors truly care about is not "how many years were signed," but whether the contracts are binding, whether cash is received in advance, and whether there is protection against downside price risks. If these terms are sufficiently robust, the profitability of memory companies will no longer fluctuate entirely with spot prices, **and the PB valuation framework may give way to the PE framework.** ## **LTAs Are Not New, But This Time Is Different** This is not the first time "long-term agreements" have appeared in the memory industry. During the DRAM shortage cycle of 2017, manufacturers also signed multi-year framework contracts. However, once demand cooled and prices fell by over 40%, customers still pressured for lower prices or delayed shipments, rendering the contracts virtually ineffective. The difference this time lies in the structure of the terms: - **Scale of Prepayments**: Some suppliers have already collected prepayments equivalent to 50% of the demand for 2027, reaching 100% for 2028. - **Pricing Mechanism**: Price ranges are established, including upper limits (collar/ceiling) and lower limit protections, rather than floating with market conditions. - **Cost of Default**: SanDisk's CEO stated directly on an earnings call, "Customers have effectively collateralized billions of dollars through various financial instruments. If they fail to meet their quarterly purchase obligations, this money immediately becomes ours." Kim Jae-jun, Executive Vice President of Samsung Electronics, was equally blunt during the 1Q26 earnings conference: "Unlike existing supply contracts based on mutual trust, these multi-year contracts involve a higher level of binding commitment." Kwak Noh-jeong, President of SK Hynix, stated at the shareholder meeting in March this year that the company is "comprehensively evaluating various structural solutions"—implying that simple penalty mechanisms are no longer sufficient, and exit paths for both parties need to be locked down at the contractual architecture level. The fundamental reason such agreements are difficult to tear up is the accounting nature of prepayments: once paid, they are recorded as capital expenditures or prepaid assets on the customer's balance sheet, making the cost of renegotiation far higher than the penalty for default. ## **Supply Gap Is the Root of Negotiating Leverage** Hyperscale Cloud Service Providers (CSPs) are willing to sign such contracts not out of goodwill, but out of necessity. The growth rate of AI inference demand far exceeds the physical expansion limits of DRAM capacity. J.P. Morgan estimates that **from 2026 to 2030, the cumulative wafer capacity shortage for the top three global DRAM manufacturers will be approximately 444,000 wafers per month (in equivalent monthly capacity).** From the demand side, the AI semiconductor market size is expected to grow by 50-60% by 2030, with TSMC recently raising its forecast for the total addressable market (TAM) for semiconductors from $1 trillion to $1.5 trillion. Meanwhile, DRAM supply is constrained by bottlenecks in EUV lithography machine capacity, and bit shipment growth may be suppressed to within 30% annually starting in 2027. The widening gap between these two trends represents the strongest leverage for memory manufacturers at the negotiating table. **In the past, the logic for memory manufacturers' capacity expansion was "Make-to-Stock"—building capacity first and waiting for demand. Now, to lock in volume, hyperscale cloud providers are pushing memory manufacturers to shift towards a "Make-to-Order" model, with prepayments essentially acting as guarantees for the manufacturers' capital expenditures.** ## **Sellers Seek Order Backstops, Not Blind Expansion** Memory manufacturers face their own challenges: **Capacity expansion is becoming increasingly expensive.** The cost per bit of new capacity is rising due to factors such as infrastructure costs, wafer equipment values, and commitments for backend packaging capital expenditures. Manufacturers are reluctant to expand capacity based solely on demand forecasts, as a scenario where AI demand falls short of expectations or customer inventories build up again would return the industry to the previous downward price cycle. **The role of LTAs is to transform "potential demand" into "more certain orders." With prepayments, volume commitments, and price protections, capacity expansion resembles fulfilling confirmed demand rather than betting on the next boom.** This also explains why a high proportion of LTAs may not necessarily stimulate aggressive capacity expansion by suppliers. In an environment of severe shortage, suppliers are more likely to prioritize maintaining high-quality contracts rather than filling capacity prematurely. Contract quality is more important than mere speed of expansion. Buyers will not accept conditions unconditionally. Potential protective measures include penalties for supplier non-delivery, bank or third-party guarantees, and even discussions on capital expenditure subsidies. However, capacity subsidies affect the flexibility of suppliers' production line operations, making them less favored arrangements in contract negotiations. ## **Valuation Framework Shift: Looking to TSMC for Reference** **For a long time, memory stocks have been priced using PB ratios—because earnings volatility was too high, making PE ratios meaningless. But once LTAs lock in volume and prices for the next 3-5 years, the predictability of earnings changes completely.** The reference point is TSMC. Before 2015, TSMC was also treated as a heavy-asset cyclical stock, valued using PB. After securing Apple's exclusive supply contract in 2014, demand visibility improved, and institutional investors began switching to a PE framework. Since then, TSMC's PE ratio has operated in the 10-30x range, with a mean of approximately 17x. Currently, the forward PE ratio for memory stocks (SK Hynix + Micron combined) is approximately 7.3x. The average over the past 11 years (excluding the 2022-2023 downturn cycle) is about 9x, with the gap relative to TSMC consistently maintained in the -50% to -80% range. Analysts believe that **for memory valuations to complete this framework shift, five conditions must be met: significant weakening of cyclicality, enduring pricing power, a continuously increasing proportion of LTA and HBM in revenue, disciplined capacity expansion generating substantial free cash flow, and deepened R&D collaboration enhancing differentiated moats.** None of these five conditions can be easily checked off—which is why J.P. Morgan states that "the investor path will be tortuous." ## **Morgan Stanley's Specific Calculations: How Much Is the Market Underpricing?** The firm constructed a sensitivity matrix, assuming HBM is 100% covered by LTAs, while commodity memory LTA coverage ranges from 50% to 80%. A PE of 6-12x is applied to the LTA portion, and 5x to the non-LTA portion. Conclusions: - **Most Conservative Scenario** (50% coverage + 6x LTA PE): The implied overall PE for Samsung and SK Hynix is approximately 5.5x, which is already higher than Samsung's current expected 2027 PE of 5.3x and SK Hynix's 5.0x. - **Neutral Scenario** (70% coverage + 10x LTA PE): The implied PE for both rises to approximately 8.5-8.6x. - **Optimistic Scenario** (80% coverage + 12x LTA PE): The implied PE for both rises to approximately 10.5-10.7x. In other words, the market currently prices earnings backed by LTAs in the same way as ordinary cyclical commodity earnings—if these contracts are truly irrevocable and cash-collateralized as companies claim, this pricing is incorrect. ## **Actual Impact of LTAs on Earnings Forecasts** **Taking SK Hynix as an example, Morgan Stanley raised its EPS forecasts for 2026-2028 by 6%, 3%, and 14% respectively, and increased its target price from 1.7 million KRW to 2.6 million KRW, representing a 38% upside from the current share price (1.88 million KRW).** Two key assumption changes underpin this adjustment: 1. **Commodity DRAM Pricing**: Previously forecasted to enter a downturn in 2028, now revised to remain stable in 1H28, as LTAs provide price floor protection. 2. **HBM Pricing**: Previously forecasted to decline by 5-10% annually from 2026-2028 (considering competitive pressure), now revised to increase by 15% annually (driven by product upgrades). This change has a particularly significant impact on 2028 earnings. The improvement in free cash flow is even more pronounced: SK Hynix's FCF forecasts for 2026-2028 are approximately 82.8 billion, 248.7 billion, and 298.8 billion KRW respectively (unit: billion KRW), with the FCF yield gradually climbing from near zero to double digits. The firm also cited cases involving Apple and Japanese shipping companies: About 70% of Apple's excess returns over the past decade came from buybacks and dividends rather than pure earnings growth; Japanese shipping companies continued to outperform the broader market despite a sharp drop in freight rates post-pandemic, due to significantly increased dividend payout ratios. The logic is the same—sustainable free cash flow can be converted into shareholder returns, changing the valuation system. Based on this logic, **J.P. Morgan raised Samsung Electronics' target price to 480,000 KRW (assigning an 8x PE to expected EPS for 2026-2027), sharply increased SK Hynix's target price from 1.8 million KRW to 3 million KRW, and doubled Kioxia's target price from 38,000 JPY to 80,000 JPY, making it the highest in the market.** ## LTAs Do Not Immune the Industry from Cycles; Lessons from 2017 Remain LTAs do not allow the memory industry to completely escape cycles. There have been failures in the past. During the DRAM shortage in 2017, the industry signed forward purchase agreements. However, when demand slowed and inventories rose, DRAM prices fell by over 40% within 2-3 quarters. Customers delayed deliveries, and contract commitments were adjusted via price resets to align with spot and market prices. That round of agreements did not truly lock out the cycle. This time, five factors must be watched. **First, whether AI monetization continues to materialize.** If AI capital expenditures enter a digestion period, customers' willingness to lock in supply will decline. **Second, whether technology changes memory demand.** Breakthroughs in model architecture, compression, KV cache, and tokenization may reduce the reliance on memory per generation of AI servers. **Third, whether supply suddenly expands.** If existing manufacturers significantly increase advanced memory supply, or new players enter, the supply-demand balance will shift again. **Fourth, there are risks at the contractual level.** If LTA terms are unfavorable to memory manufacturers, or if prepayments, guarantees, and penalty mechanisms are not strong enough, earnings stability will be overestimated. **Fifth, at the macro level, tariffs, sanctions, export controls on high-end AI chips,** and pressure on customer balance sheets may also affect contract fulfillment. Therefore, the core of LTAs is not "signing news," but cash flow. The evidence the market needs to see in the future is straightforward: whether substantial cash inflows appear on memory companies' balance sheets, along with corresponding deferred revenue obligations. Only when both money and obligations appear simultaneously will LTAs be more than just a narrative. ``` The above exciting content comes from [Zhuifeng Trading Desk](https://mp.weixin.qq.com/s/uua05g5qk-N2J7h91pyqxQ). For more detailed interpretations, including real-time analysis and frontline research, please join [**Zhuifeng Trading Desk ▪ Annual Membership**](https://wallstreetcn.com/shop/item/1000309) ``` ### Related Stocks - [XSD.US](https://longbridge.com/en/quote/XSD.US.md) - [SOXX.US](https://longbridge.com/en/quote/SOXX.US.md) - [SOXL.US](https://longbridge.com/en/quote/SOXL.US.md) - [512760.CN](https://longbridge.com/en/quote/512760.CN.md) - [07709.HK](https://longbridge.com/en/quote/07709.HK.md) - [512480.CN](https://longbridge.com/en/quote/512480.CN.md) - [PSI.US](https://longbridge.com/en/quote/PSI.US.md) - [SMH.US](https://longbridge.com/en/quote/SMH.US.md) - [DRAM.US](https://longbridge.com/en/quote/DRAM.US.md) - [MS.US](https://longbridge.com/en/quote/MS.US.md) - [JPM.US](https://longbridge.com/en/quote/JPM.US.md) - [SSNGY.US](https://longbridge.com/en/quote/SSNGY.US.md) - [TSM.US](https://longbridge.com/en/quote/TSM.US.md) - [AAPL.US](https://longbridge.com/en/quote/AAPL.US.md) - [MS-O.US](https://longbridge.com/en/quote/MS-O.US.md) - [MS-Q.US](https://longbridge.com/en/quote/MS-Q.US.md) - [MS-E.US](https://longbridge.com/en/quote/MS-E.US.md) - [MS-I.US](https://longbridge.com/en/quote/MS-I.US.md) - [MS-L.US](https://longbridge.com/en/quote/MS-L.US.md) - [MS-P.US](https://longbridge.com/en/quote/MS-P.US.md) - [MS-A.US](https://longbridge.com/en/quote/MS-A.US.md) - [MS-F.US](https://longbridge.com/en/quote/MS-F.US.md) - [MS-K.US](https://longbridge.com/en/quote/MS-K.US.md) - [JPM-M.US](https://longbridge.com/en/quote/JPM-M.US.md) - [JPM-C.US](https://longbridge.com/en/quote/JPM-C.US.md) - [JPM-D.US](https://longbridge.com/en/quote/JPM-D.US.md) - [JPM-L.US](https://longbridge.com/en/quote/JPM-L.US.md) - [8634.JP](https://longbridge.com/en/quote/8634.JP.md) - [JPM-K.US](https://longbridge.com/en/quote/JPM-K.US.md) - [JPM-J.US](https://longbridge.com/en/quote/JPM-J.US.md) - [SMSN.UK](https://longbridge.com/en/quote/SMSN.UK.md) ## Related News & Research - [Why Micron (MU) Stock Is Up Today and Why BofA Thinks It Can Still Go Higher](https://longbridge.com/en/news/286280489.md) - [AI boom puts SK Hynix on the cusp $1 trillion market value](https://longbridge.com/en/news/286347462.md) - [EXCLUSIVE-At Samsung, the global AI boom spurred a looming strike and deep divisions](https://longbridge.com/en/news/286544840.md) - [Forget Nvidia: New Memory-Focused ETF Soars 90% As CEO Calls Chips AI's 'Biggest Bottleneck'](https://longbridge.com/en/news/286257549.md) - [KLA approves 10-for-1 stock split and dividend hike](https://longbridge.com/en/news/286340025.md)