Analysts this week made a string of consequential calls on firms tied to artificial intelligence, separating those they see as poised to win from companies facing nearer-term friction. Below are the key broker moves and the rationale the analysts provided.
Bank of America reinstates Microsoft at Buy with $500 target
Bank of America has resumed coverage of Microsoft with a Buy rating and a $500 price target, describing the company as "a primary beneficiary of AI monetization" and placing it "at the center of the AI supercycle." In a note led by analyst Tal Liani, the bank highlighted Microsoft's dual exposure to AI in both infrastructure and applications as the firm's chief advantage.
BofA emphasized Azure's role in supplying "the compute and data foundation for enterprise AI workloads," while also pointing to Microsoft software such as 365, Dynamics, GitHub and Windows that "embed into everyday tasks that drive attach and consumption." The $500 target — implying roughly 40% upside according to the broker's math — rests on a 24x estimated price-to-earnings multiple for calendar year 2027. That multiple is justified in BofA's view by projected revenue growth of 15% to 17% annually over the next three years, with Intelligent Cloud expected to expand 24% to 28% as AI workloads scale.
The research note also outlined several debates that could influence Microsoft's long-term outlook, including the sustainability of its AI-related backlog, the consequences of its partnership with OpenAI, and whether the current AI cycle is a transient bubble or a durable shift. On margins, BofA projects a marked rise in capital expenditures - from $44 billion in 2024 to about $143 billion by 2028 - while still expecting Microsoft to sustain operating margins above 46%.
JPMorgan downgrades SAP to Neutral amid decelerating cloud backlog
JPMorgan shifted SAP's rating from Overweight to Neutral, cutting its target to €175 from €260. The bank pointed to a softer near-term outlook driven by decelerating growth in SAP's current cloud backlog (CCB) and emerging strategic risks. The firm's prior optimism had been founded on expectations of accelerating revenue and meaningful margin expansion; JPMorgan now says "the picture for performance has shifted" as several headwinds come together.
Analysts led by Toby Ogg flagged that SAP's cloud backlog growth peaked in 2024 and has been moderating, with further deceleration anticipated into 2026 as the migrated-customer base matures. In an environment where the market now demands acceleration to counter software bear arguments, JPMorgan argued that ongoing deceleration is unlikely to support near-term stock performance.
The team also highlighted SAP's potential move toward a consumption or outcome-based model. While they acknowledged such a transition may be necessary, they warned it could introduce revenue volatility and "skew the traditional links between metrics that investors are currently wired up for," which would complicate forecasting. Competition intensifying in the AI agent layer - driven by large language model providers and increased peer spending - was cited as another pressure point that could weigh on margins and spur M&A activity.
"In aggregate, change is fast approaching and incumbents, including SAP, will need to invest and evolve to give themselves the best chance of remaining relevant as the AI cycle unfolds," the analysts wrote.
Bernstein reduces Qualcomm to Market Perform on memory cost and handset demand concerns
Bernstein lowered its rating on Qualcomm to Market Perform from Outperform and trimmed the price target to $140 from $175. Analyst Stacy Rasgon pointed to rising prices for mobile DRAM and NAND as a growing headwind, and warned of the potential for "double-digit unit declines this year" in smartphone shipments as vendors confront choices to raise prices, trim device specifications, or accept margin compression.
Rasgon indicated that consensus estimates look too optimistic, saying that even after Qualcomm's own guidance reductions, the numbers "now appear much too high." He identified downside risks from weaker volumes and an expected step-down in Apple-related revenue toward year-end.
The possible contraction of Apple's share of Qualcomm's modem business was singled out as a major concern. Rasgon suggested that Apple's portion of the modem business could fall from roughly 80% to about 20%, a headwind he believes the market underestimates. He also noted that Apple-related licensing contributes around $1.50 of EPS and that the current licensing arrangement expires in April 2027. While the analyst expects Qualcomm to prevail in any dispute, he cautioned the process could be volatile and harm sentiment.
Bernstein saw potential offsets such as a $20 billion buyback program and the prospect of a datacenter event, but judged those factors insufficient to materially change near-term sentiment. "We feel a bit bad for the company; they appear to be doing everything right, but are living in a bad neighborhood at the moment," Rasgon wrote. He added, "We suspect things get worse before they get better, and while the stock remains extremely cheap even on our numbers, that becomes less of a support when one can buy actual AI winners for under 15x on 'realistic' estimates." The analyst said the firm would be willing to revisit its view "if the neighborhood shows signs of gentrification."
Needham upgrades Arm Holdings to Buy, citing payoff from strategic bets
Needham & Company upgraded Arm Holdings to Buy with a $200 price target, arguing that a set of high-stakes strategic moves are beginning to pay off as demand for AI-focused CPUs grows. After remaining on the sidelines for two and a half years, the firm said it now sees a turning point: "We have been on the sidelines on ARM for 2.5 years and now see a series of their high-stake bets ... are working," the analysts wrote.
Needham described these steps as "bold moves" that could disrupt the industry landscape but asserted they are now transforming Arm for the better. Key drivers cited include Arm's push to higher royalty rates, expansion into compute subsystems, and progressing toward its own silicon. The analysts argued that "ARM has become a credible AI play, right around the time when the company has better structured itself to capture greater value from AI," pointing to the rise of agentic AI and increasing CPU workloads inside data centers.
Central to Needham's thesis is Arm's AGI CPU, the firm's first Arm-designed data center processor developed with Meta and targeted at agentic AI workloads. The analysts said this initiative "perfectly epitomizes" the move-fast approach and that Arm's entrance into silicon with Meta - once seen as a risky gamble - now shows clear momentum.
BofA raises Dell and Sandisk targets after Asia supply-chain checks
In a separate research piece, Bank of America raised price targets for both Dell and Sandisk following an Asia supply-chain review that pointed to robust AI server demand and a tightening NAND market. Analyst Wamsi Mohan reported checks indicating "demand for AI servers remains strong," with near-term revenue driven largely by GB300-based systems and expectations for additional VR-rack shipments in the second half of the year.
On Dell, BofA came away favorably and suggested Dell's $50 billion fiscal 2027 AI-server revenue guide "could end up conservative." The bank raised its estimate for Dell's AI-server revenue in the first quarter to $15 billion from $13 billion and increased its full-year AI-server forecast to $60 billion from $50 billion. Dell's price target was lifted to $172 from $155.
For Sandisk, BofA pointed to a "strong NAND outlook" and tightening supply, and noted that pricing "continues to be strong" while customers increasingly seek long-term agreements. The bank modeled a 63% quarter-over-quarter average selling price jump for the March quarter and raised its June-quarter forecast to 20% growth. Sandisk's price target was raised to $900 from $850.
On the broader industry picture, BofA also highlighted persistent tightness in global optical fiber supply and described PC demand as mixed. The bank warned that pull-forward buying tied to rising memory prices could leave the second half of the year weaker, and flagged potential "incremental downside" to industry PC demand.
What these moves collectively signal
Across these calls, analysts drew a distinction between companies that they believe are structurally positioned to monetize AI at scale and those facing nearer-term cyclical or structural headwinds. Bank of America's stance on Microsoft and its upgraded views on Dell and Sandisk reflect confidence in infrastructure and components that support AI workloads. Conversely, JPMorgan's caution on SAP emphasizes the complications that can follow as software incumbents shift models and face intensifying competition at the agent layer. Bernstein's view on Qualcomm highlights how commodity cost dynamics and OEM relationships can materially affect chipmakers' fortunes, even amid broader AI demand.
The week’s analyst activity underscores the range of considerations investors must weigh in the AI era - from capex, margins and revenue backlog durability to supply-chain tightness, licensing arrangements, and the pace at which new CPU and silicon plays become commercially meaningful.
Bottom line
The batch of analyst updates this week presents a mixed picture. Some firms are being rewarded for perceived proximity to scalable AI monetization, while others are being marked down due to slowing backlogs, rising component costs, and evolving business models. Investors tracking AI-related equities will likely continue to see a bifurcated landscape as demand for compute and the structure of software monetization evolve.