Morgan Stanley Asia Research: Clients' biggest anxiety is not being able to buy enough Nvidia chips; storage shortages are "one of the most severe in 30 years."
As debates continue in the market over the sustainability of AI capital expenditure, a recent Morgan Stanley Asia survey of the Asian semiconductor industry chain reveals the supply-demand extremes under the AI boom.
According to Chase Wind Trading Desk, Morgan Stanley has just released a report based on its field research in Asia, stating that Nvidia’s market dominance is even stronger than generally recognized. The greatest anxiety of clients over the next 12 months is “not being able to obtain enough Nvidia products”, especially the Vera Rubin chips;
Secondly, memory chip shortages have reached “one of the worst levels in 30 years,” with a buying frenzy from cloud computing clients depleting products available for PC and server OEM manufacturers;
Additionally, Morgan Stanley noted that Broadcom’s supply chain prospects for Google-designed TPUs have been revised upwards, but this growth primarily comes from substitution for other clients’ orders.
Based on the survey, Morgan Stanley raised its target prices and profit expectations for Nvidia and Broadcom, asserting that AI intensity is testing the limits of the entire semiconductor ecosystem, from front-end wafers to back-end packaging and memory, all facing supply constraints.
Nvidia’s Moat: Clients’ “Procurement Anxiety”
Despite a market narrative about the rise of competitors, supply chain realities remain harsh. Morgan Stanley’s research shows that Nvidia’s dominance remains unshaken, and client anxiety about shortages is intensifying. Morgan Stanley states:
In the coming 12 months, clients’ greatest anxiety is whether they can source enough Nvidia products... While everyone wants alternatives and some of these work economically for specific applications, what we’ve confidently learned from industry contacts is: for most use cases, Nvidia still offers the best economic returns.
The data comparison is persuasive: Nvidia just completed $51 billion in quarterly data center revenue — about 14 times Google’s TPU revenue. Its sequential quarterly growth of $10 billion alone equals three times TPU’s revenue. Even in “TPU-heavy” workloads, Nvidia contributes a significant share, possibly representing most of data center revenue.
Based on industrial chain verification, Morgan Stanley raised Nvidia’s target price from $235 to $250, while also increasing profit expectations. Although its valuation multiple was lowered from 27x to about 26x CY27 P/E (still above the semiconductor sector but below fellow AI giant Broadcom), based on the new MW EPS forecast of $9.57 (above the previous $8.66), the new target price is $250.
The Undercurrent in Custom Chips: Broadcom Benefits, Meta Strategy Shifts
Beyond Nvidia, the custom chip (ASIC) market is anything but smooth sailing. Research shows that the supply chain outlook for Google’s Tensor Processing Unit (TPU)—designed and sold by Broadcom—is being revised upward, and this positive signal appeared even before Gemini’s favorable reviews. Morgan Stanley heard about the upward revision in TPU orders from analog companies, memory companies, and ODM partners; the greatest increase in expectations is concentrated in CY27 and beyond.
However, Morgan Stanley issued a key warning: Some TPU growth is actually replacing orders previously expected for other Broadcom ASIC clients. Specifically, Meta’s MTIA chip mass production plans (originally slated for 2H26) have been delayed and partially substituted by Meta’s use of TPUs. Industry insiders believe Meta and OpenAI’s use of TPUs is partly to familiarize with ASICs before eventually migrating to internal ASIC solutions.
A more important long-term risk is that Google is developing its own TPU variant with MediaTek, which could threaten Broadcom. The research found Google is pushing for mass production of a chip next year, though it still has functional issues to resolve, showing an urgency for alternative solutions.
Based on these findings, Morgan Stanley raised Broadcom’s target price from $409 to $443, kept the 41x valuation but factored in higher expectations, based on CY27 MW EPS (including equity incentives) of $10.79. Morgan Stanley revised Broadcom’s FY2026 ASIC revenue forecast to $27.21 billion and FY2027 to $59.475 billion, significantly higher than previous estimates.
Memory Chip Crisis: Shortage at Worst Level in 30 Years
This is the most alarming part of the report. If AI computing power shortages were expected, then a comprehensive shortage of memory chips may be next market’s biggest bottleneck.
Morgan Stanley’s report states:
In the 30 years we've tracked the memory cycle, we've never seen such a severe situation. While the largest cloud buyers are in a gold-rush buying frenzy, the rest of the world seems to be facing a product famine… Once current inventory is depleted, there appears to be no capability to replenish at any price.
Morgan Stanley noted that as the largest cloud buyers are in a buying spree, the rest of the world seems to be facing product starvation. All markets could face production stoppages, showing that this is a true supply crunch – not artificially induced. The tightness in DDR4 persists and is beginning to impact the auto market in the coming months.
For NAND, enterprise market (especially QLC-based products) is feeling the squeeze most, but the consumer market is also tight to varying degrees. For high-bandwidth memory (HBM), although 1Q long-term contract renewals face some downward price pressure, profits remain high. The current dynamic is that DDR5 profit drives HBM prices up, not the other way around. As for HBM4, Morgan Stanley believes Micron will enter Nvidia's supply chain, albeit with a relatively small initial share.
~~~~~~~~~~~~~~~~~~~~~~~~
The above highlights come from Chase Wind Trading Desk.
For more in-depth interpretations, including real-time insights and frontline research, please join [Chase Wind Trading Desk • Annual Membership]
Risk Disclosure and DisclaimerThe market involves risks; investment requires caution. This article does not constitute personal investment advice, nor does it take into account individual users’ specific investment objectives, financial situations, or needs. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investing accordingly is at your own risk.