Can Arm (ARM.US), down 60% from its all-time high, stage a valuation comeback? AI agents are reshaping CPU demand, and Morgan Stanley remains cautiously bullish.
Morgan Stanley maintains its neutral "Equal-weight" rating stance and a price target of US$212. Relative to the November 13 closing price of US$190.53 used in the report, the potential upside is 11.3%; based on the US$175.05 price after the sharp drop on November 16, the upside widens to 21.1%.
Title context: Can Arm (ARM.US), down 60% from its all-time high, stage a valuation comeback? AI agents are reshaping CPU demand, and Morgan Stanley remains cautiously bullish.
Text:
Just as Arm Holdings plc (ARM.US), the owner of the ARM instruction set architecture, saw its share price plunge for two consecutive trading days after a brief rebound, Wall Street financial giant Morgan Stanley raised the chip giant's medium-term earnings forecast trajectory, but the analyst team maintained its "Equal-weight" neutral rating stance and US$212 price target. Relative to the November 13 closing price of US$190.53 used in the report, the potential upside is 11.3%; based on the US$175.05 price after the sharp decline on November 16, the upside widens to 21.1%.
Morgan Stanley's latest outlook on Arm can basically be summarized as follows: the AI data center buildout is in full swing and is accelerating the realization of royalty growth, while in-house CPUs are beginning to be tested on production capacity, actual deployment capability and return on capital; this means that Arm's specific investment opportunity depends both on the market expansion trajectory of NVIDIA Corporation's Vera CPU, which adopts the ARM architecture, and on the cloud AI business growth trajectory of the hyperscale cloud giants that use the ARM architecture to build their own data center CPUs, as well as on what valuation the market is willing to pay for Arm's transition from IP licensing to direct data center CPU chip sales.
Arm's share price has recently shown a high-volatility trajectory of "sharp decline, strong rebound, then rapid giveback," with both upside and downside elasticity markedly higher than the semiconductor sector. On November 5, Arm's share price plunged 13.08% to close at US$166.72; it then recovered along with AI computing stocks to US$203.36 by November 11, a cumulative rebound of about 22%, including a single-day surge of 12.05% on November 11, clearly exceeding the Philadelphia Semiconductor Index's gain of about 5.66% over the same period. But this counterattack did not last: the stock fell for three consecutive trading days on November 12, 13 and 16, with a cumulative drawdown of 13.92%, while the Philadelphia Semiconductor Index fell about 9.03% over the same period; in particular, on November 16, Arm dropped 8.12% to US$175.05, again significantly exceeding the Philadelphia Semiconductor Index's 5.28% decline.
With the Philadelphia Semiconductor Index still about 34.3% below its record closing high, Arm's recent share price moves show that the market's expectations for its AI data center business growth prospects are not yet sufficient to offset the severe contraction pressure on valuations across the AI computing-related semiconductor industry, driven by a series of negative events including the Democrats' landslide victory in the midterm elections leading to expectations of slower data center construction, and Jensen Huang being forced to step down as NVIDIA Corporation CEO. The short-term rebound has not yet translated into sustained relative outperformance.
With the Philadelphia Semiconductor Index deep in a bear market, Wall Street's growth focus for Arm appears to be shiftingwhether the incremental CPU demand brought by AI agents can support earnings and a valuation recovery. The most investment-relevant judgment in Morgan Stanley's latest research report covering Arm is that data center AI cloud royalties are providing quantifiable growth support, and whether in-house CPUs can convert revenue scale into sustained profit will determine how much valuation premium Arm can command.
Philadelphia Semiconductor Index sold off sharply again after rebound; market re-examines AI computing demand growth curve
From November 9 to 13, the Philadelphia Semiconductor Index and the Nasdaq 100, known as the "global tech stock barometer," rebounded about 8.05% and 3.53%, respectively; however, at Monday's US market close on November 16, the two fell again by 5.28% and 2.67%, so the US semiconductor sector remains in a technical bear market and the Nasdaq remains in a technical correction zone, with last week's rebound not yet changing the market's cautious stance.
Just as the Philadelphia Semiconductor Index fell into a bear market, as shown above, the share price of the chip design leaderArm, owner of the ARM instruction set architecturehas notably exhibited a high-volatility trajectory of "sharp decline, strong rebound, then rapid giveback," with both upside and downside elasticity markedly higher than the US semiconductor benchmark sector.
As of the US market close on November 16 Eastern Time, Arm was at US$175.05, down about 61.33% cumulatively from its all-time intraday high of US$452.70 set on June 18, 2026a decline of more than 60%. The Philadelphia Semiconductor Index last closed at 9,610.65 points, 34.3% below its record closing high of 14,634.72 points on June 22, and its drawdown on November 5 had reached 38.4%. On the same day (November 16 US market close), the Dow, S&P 500 and Nasdaq Composite fell 1.41%, 1.92% and 2.49%, respectively, showing that the selloff has spread from chips to broader risk assets.
Asian markets are still extending this wave of severe pressure on chip stocks. On November 16, Taiwan's weighted index, South Korea's KOSPI and Japan's Nikkei 225 fell 6.03%, 3.59% and 2.42%, respectively; on the 17th, Taiwan's weighted index fell another 2.31% to close at 40,092.11 points, a cumulative decline of 8.20% over two trading days, while South Korea's benchmark KOSPI index fell more than 2% intraday at one point. The combination of NVIDIA Corporation's leadership transition, export restrictions and uncertainty over financing for massive projects tied to large AI data centers is putting greater pressure on earnings expectation revisions in Asian markets, which are highly concentrated in wafer foundry, memory and server supply chains.
Asia's stock markets are home to a vast array of chip design, wafer manufacturing and memory chip leaders, as well as the most advanced 3D advanced packaging systems and chip testing systems, which belong to different semiconductor sub-markets yet share the procurement budgets and guaranteed financing terms of NVIDIA Corporation and a host of large AI cloud-related customers. When the market simultaneously reassesses customer spending, management continuity and cost of capital, geographic diversification can hardly fully offset the shock from a common source of demand.
NVIDIA Corporation's leadership change further raised the risk compensation the market demands. The company disclosed that its board accepted Jensen Huang's resignation on November 11 and that Chief Financial Officer Colette Kress would also serve as interim CEO.
Wall Street financial institutions' repricing of NVIDIA Corporation's future price target trajectory reflects the typical characteristic of "acknowledging the long-term growth rate of AI infrastructure orders while lowering the forward valuation premium": Goldman Sachs Group, Inc. cut its target price from US$300 to US$220, Citi from US$285 to US$225, and UBS Group AG from US$300 to US$240, reductions of about 26.7%, 21.1% and 20%, respectively, while retaining "Buy" ratings.
This typical cautiously bullish sentiment among the above Wall Street institutions is also reflected in Morgan Stanley's latest assessment of Arm's prospects and valuation: the bank raised data center royalty and medium-term earnings forecasts, but cut the target EV/EBIT multiple for the IP business from 49x to 41x, ultimately maintaining the US$212 price target and "Equal-weight" rating. For investors, the key change is that AI data center-related business growth must now pass scrutiny on both valuation and profit quality; whether Arm can achieve a sustained re-rating depends on the realization of royalty growth and whether in-house CPU expansion can maintain sufficient profitability.
Cloud royalties take over the growth baton; Arm's AI revenue is becoming concrete
As of November 16, the Philadelphia Semiconductor Index and Arm were about 34% and 60% below their respective all-time closing highs; meanwhile, tool calls, code execution, data processing and task scheduling in agent workflows are expanding the application space for CPUs.
Morgan Stanley expects data center-led Cloud AI (i.e., AI cloud business)-related royalties to grow about 86% and 68% in fiscal 2027 and 2028, respectively, driving an upward revision to Arm's medium-term earnings forecasts; however, direct CPU sales require bearing manufacturing costs, deployment constraints and competitive pricing, so the bank cut its fiscal 2031 CPU revenue forecast to close to US$10 billion, below management's target of about US$15 billion, and lowered the IP business valuation multiple, maintaining "Equal-weight" and a US$212 price target, implying about 21.1% upside from the November 16 closing price.
Arm's latest financial results show that data center royalties are growing significantly faster than the overall royalty business, with cloud computing becoming a more important source of revenue growth. In the second quarter of fiscal 2027 ended September 2026, the company's revenue reached US$1.443 billion, up 27% year-on-year; licensing and other revenue grew 50% to US$773 million, and royalty revenue grew 8% to US$670 million. Among these, data center royalties doubled year-on-year, reflecting that Arm's technology adoption in cloud computing is translating into actual revenue. In addition, adjusted operating profit grew 37% to US$638 million, with a margin of 44.2%; licensing revenue reflects customers' investment in starting designs and purchasing technology, while royalty revenue further verifies that chips using Arm technology are shipping.
Morgan Stanley therefore raised its data center-led Cloud AI royalty forecasts to US$686 million for fiscal 2027 and US$1.153 billion for fiscal 2028, corresponding to year-on-year growth of about 86% and 68%; the share of total royalty revenue rises from 14% in fiscal 2026 to 31% in fiscal 2028. The "raised 9% and 17%" in the report refers to the magnitude of forecast revisions relative to the old model. Growth sources include adoption of Amazon.com, Inc. Graviton, Alphabet Inc. Class C Axion, and NVIDIA Corporation Vera in integrated systems and potential standalone inference deployments; more cores, the Armv9 architecture and compute subsystems (CSS) increasing the IP value contained in each chip also support per-unit royalty growth.
From a technical mechanism perspective, AI agents expand the workload of the entire computing system, and the tasks handled by CPUs are increasing as well. GPUs and ASICs handle intensive model computation, while CPUs handle tool calls, code execution, data access, task scheduling and general business logic. As agent concurrency increases, what determines cluster output is not only the number of accelerators, but also whether CPUs, memory and networking can supply data in a timely manner and complete serial tasks. In power-constrained data centers, customers place greater emphasis on throughput per watt and total cost under actual workloads, which creates conditions for broader adoption of Arm server CPUs.
The mobile phone business remains an important pillar, but its growth pace is markedly slower. Morgan Stanley expects mobile phone royalties of about US$1.199 billion in fiscal 2027, up 3% year-on-year, rebounding to US$1.372 billion in fiscal 2028, up 14%, relying mainly on Apple Inc.'s stabilizing output, a higher mix of high-end products and increased licensing content, rather than a substantial expansion in global phone shipments. AI-related CSS and architecture license agreements continue to support new design wins; the report separately lists unconfirmed early-access licensing for v10 as a potential catalyst. Arm's revenue structure is thus gradually shifting from reliance on mobile phones to cloud computing providing higher growth.
Direct sales of data center CPUs open up revenue space and also change Arm's profit structure
Morgan Stanley's research report shows that the AGI CPUArm's self-developed and exclusively externally sold data center CPU productmoves Arm further from providing design technology toward selling chips, thereby expanding revenue scale and changing its cost structure as well.
The company expects the first commercial deliveries to occur in the fourth quarter of fiscal 2027, i.e., January-March 2027. Compared with high-margin IP licensing, chip sales require bearing wafer manufacturing, packaging and testing, supply assurance and inventory capital occupation; even with outsourced manufacturing, Arm must still be responsible for the entire product delivery process and actual commercial returns.
Morgan Stanley largely retained its fiscal 2028 CPU revenue forecast but cut its fiscal 2031 forecast by about 38% to close to US$10 billion, below management's target of about US$15 billion. Its constraints include deployable power, competitive pricing and product cost: the model assumes a gross margin of about 35% for first-generation products, with per-unit core revenue declining about 5% with each generation.
Even so, Morgan Stanley still assumes that by 2030, the fifth-generation product will have four times the deployed cores per gigawatt as the first generation. The US$10 billion long-term vision already incorporates significant product progress and market expansion, and whether it can be delivered on schedule remains an important valuation variable.
In Morgan Stanley's overall financial model, Arm's revenue for fiscal 2027-2029 is expected to be US$6.082 billion, US$8.840 billion and US$12.450 billion, respectively, with adjusted earnings per share of US$2.23, US$3.95 and US$5.66. Among these, the fiscal 2028 EPS forecast is about 13.8% higher than the old model and about 28.2% above the consensus estimate of US$3.08. However, as the share of chip sales increases, the group's gross margin is expected to fall from 97.5% in fiscal 2026 to 86.9% in fiscal 2028; the adjusted operating margin can still rise to 57.2% over the same period, relying on revenue growth diluting expenses such as R&D. What Morgan Stanley continues to focus on regarding Arm's long-term fundamental prospects is precisely whether this operating leverage can be sustained through the product expansion process.
Valuation treatment therefore remains restrained. Morgan Stanley uses a sum-of-the-parts valuation, cutting the enterprise value/operating profit (EV/EBIT) multiple for the IP business from 49x to 41x, maintaining 20x for the chip business, and discounting fiscal 2029 value at an 11% weighted average cost of capital. The earnings upgrade and the valuation multiple cut offset each other, ultimately maintaining the US$212 price target; based on a static calculation using the share count on the report's cover, this corresponds to a market capitalization of about US$217.8 billion. The US$370 bull case and US$65 bear case reflect the enormous impact of the degree to which the forward chip business is realized on valuation. Morgan Stanley has acknowledged stronger earnings growth but has not yet assigned a valuation for full realization of the chip business.
Related Articles

CITIC SEC: Cost reduction and profit growth; the economic inflection point for domestic standalone energy storage is approaching.

FIN STREET SEC (01476): Chairman Zhu Yanhui is under investigation and subject to retention measures

China Securities Co.,Ltd.: AI advanced packaging drives upgrading of glass substrate materials; equipment takes the lead in the early stage of industrialization.
CITIC SEC: Cost reduction and profit growth; the economic inflection point for domestic standalone energy storage is approaching.

FIN STREET SEC (01476): Chairman Zhu Yanhui is under investigation and subject to retention measures

China Securities Co.,Ltd.: AI advanced packaging drives upgrading of glass substrate materials; equipment takes the lead in the early stage of industrialization.






