Overview
Analysts have realigned recommendations and forecasts across companies tied to artificial intelligence this week, producing divergent signals: fresh buys on memory suppliers, a reaffirmation of the strategic importance of memory in AI stacks, and downgrades for select software and freelance platforms that analysts say face near-term headwinds from rapid AI adoption. Separately, a macro research house flagged the possibility of a sharp repricing of equity market expectations beyond 2026, creating a longer-term caution even as it expects the AI-led rally to pick back up in the nearer term.
UBS starts SK Hynix ADR at Buy, assigns $204 target
UBS opened coverage on SK Hynix’s American Depositary Receipts with a Buy rating and a $204 price target, arguing that current market pricing fails to reflect sustainably higher memory profitability. Analyst Nicolas Gaudois pointed to the ADR’s recent moves - a decline of 33% from a July 14 peak and trading about 13% below its offering price of $149 as of Thursday - and contrasted those levels with UBS’s long-term profitability view.
UBS said the market is pricing in a long-term return on equity similar to the pre-AI average of 17.7%, while the bank’s own 2027-2031 average ROE forecast is 40.2%. The thesis rests on an industry the analyst describes as having "drastically changed," with agentic AI expected to materially lift memory bit demand through 2027. UBS projects DRAM bit demand growth to accelerate to 36% year-over-year in 2027, up from 22% in 2026, and forecasts NAND bit demand growth to rise to 23% in 2027 from 20% in 2026.
On corporate results, UBS highlighted that revised long-term supply agreements are being signed sooner than it anticipated, with 10 such contracts already in place and additional agreements under negotiation. Gaudois noted that while those contracts may limit immediate pricing upside, they should be margin-accretive over the longer term. He also underlined that DRAM average selling prices rose 30% quarter-over-quarter, with that increase partially tempered by a relatively higher mobile product mix and the fact that HBM4 shipments began only late in the reporting period.
UBS concluded that current share prices do not fully incorporate expectations for higher profitability, improved free cash flow generation, and enhanced shareholder returns.
Capital Economics: AI rally should resume, but S&P 500 may weaken by end-2027
Macro strategist Capital Economics told clients that the recent selloff among AI-related technology leaders appears driven more by fear than clear-cut evidence and expects the rally to resume in the near term. John Higgins, Chief Economic Adviser, outlined several sources of pressure that have weighed on sentiment.
One such pressure is the potential for greater memory chip supply out of China following new listings such as CXMT, which the note said listed on Monday. Capital Economics suggested that while this new capacity may not significantly alter supply of high-bandwidth memory prioritized for AI workloads, it could ease shortages in mainstream memory chips and, if prices decline, undermine revenue at large U.S. and Korean memory manufacturers.
The firm also highlighted developments in China’s AI capabilities as another source of investor concern. Reports that Chinese entities are producing their own deep ultraviolet lithography machines took a toll on ASML’s shares, even though ASML’s extreme ultraviolet systems are substantially more advanced. Similarly, the emergence of lower-cost Chinese labs training near-frontier models has sparked questions about whether U.S. hyperscalers will be able to monetize their AI investments.
Capital Economics also said reports that Nvidia might provide roughly $250 billion in financing guarantees for OpenAI have reopened worries about complex financing arrangements in the AI rollout.
Despite these headwinds, the research house retained an end-2026 S&P 500 forecast of 8,250, indicating an expectation that the "AI train" will get back on track in the nearer term. Over a longer horizon, however, Capital Economics warned that earnings expectations that currently look "exceptionally optimistic" could be scaled back, projecting the index to decline to 6,500 by the end of 2027.
Bernstein views memory partnerships as confirmation of secular importance
Bernstein reiterated a constructive stance on memory chipmakers following a string of major partnership announcements unveiled at a South Korean government AI Summit in San Francisco. The briefing emphasized agreements between Samsung, SK Hynix, Nvidia and Broadcom that span memory supply, foundry services and the buildout of AI data center capacity.
Analysts led by Mark Li highlighted an agreement between SK Hynix and Nvidia documented in letters of intent that the note described as exceeding $500 billion. That partnership reportedly covers memory supply and plans for a 2GW Vera Rubin DSX AI Factory destined for SK Telecom in 2027. Bernstein’s team also said SK Group is pursuing an additional $250 billion in memory supply collaborations over the next five years.
Separately, Samsung and Broadcom signed a memorandum of understanding covering $200 billion of memory, including HBM, as well as foundry services through 2030 focused on 2-nanometer and smaller process nodes and advanced packaging.
Bernstein’s analysts argued the announcements reinforce memory’s central role in AI infrastructure and noted consensus expectations for roughly $1.3 trillion in annual memory revenue in both 2027 and 2028. They portrayed the recent sector pullback as a "good entry point" and reiterated a preference for memory over logic semiconductors, while calling out a longer-term competitive threat in NAND from China. The firm retained an Underperform rating on Kioxia, saying that the competitive risk looks more material in NAND than in DRAM, where China would likely struggle without access to EUV lithography.
TD Cowen trims Intuit to Hold; flags extended uncertainty
TD Cowen downgraded Intuit to Hold from Buy, with analyst Jared Levine characterizing the company’s near-term catalyst path as skewed more negative than positive and likely to pressure the stock until at least Intuit’s 3Q27 earnings release in May. Levine highlighted a difficult set-up for the company’s fourth-quarter results, an investor day in September where no major new announcements are anticipated, and pervasive concerns about AI-related risk and competitive intensity.
For the fourth-quarter report due in August, Levine expects Intuit to guide fiscal 2027 revenue below consensus and below its long-term target of 10%-plus growth. He also anticipates that Intuit will reduce long-term growth targets for its Global Business Solutions and TurboTax segments. While Levine does not fully endorse the most pessimistic "existential AI" bear case, he said significant execution risk remains tied to Intuit’s AI product strategy and that competition from free do-it-yourself tax solutions could erode the durability of double-digit revenue growth.
TD Cowen lowered its price target to $304 from $504, using a 10 times calendar 2027 earnings multiple, and trimmed fiscal 2027 and 2028 revenue estimates to 1.3% and 2.2% below Street consensus, respectively. The firm noted that sustained fund outflows from the AI trade represent the principal risk to its neutral view.
Analysts pull back on Fiverr amid accelerating AI pressure
Fiverr, the online freelance marketplace, saw multiple broker downgrades after the company cut full-year guidance in response to what analysts described as accelerating pressure from AI adoption. Oppenheimer moved Fiverr to Perform from Outperform and Goldman Sachs reduced its rating to Neutral from Buy.
Oppenheimer’s Jason Helfstein said an increase in adoption of large language models "noticeably increased intra-quarter," prompting Fiverr to cut full-year revenue guidance by 9% and EBITDA guidance by 21%. The company reported second-quarter revenue that fell 10% year-over-year, deeper than the 2% decline reported in the prior quarter, and trailing-twelve-month buyers declined 22%.
Oppenheimer noted that while Fiverr is shifting toward higher-value services, 85% of its gross merchandise value continues to be comprised of lower-cost offerings, and that headwinds appear to be accelerating. The broker now models full-year 2026 and 2027 revenue declines of 16% and 38%, respectively, and disclosed a "low degree of confidence" in its projections. Management indicated the turnaround may take roughly six more quarters. Oppenheimer added that it is likely to wait for revenue stabilization before altering its stance, even though the stock currently trades near cash on an enterprise value of approximately $28 million.
Goldman’s Eric Sheridan described the freelance market as exhibiting "dueling themes" - compression among high-volume, lower-value work alongside a migration to higher-end, AI-enabled projects. Goldman lowered its price target for Fiverr to $13 from $26 and highlighted that annual active buyers fell 22% while spend per buyer increased 16%. The bank said it needs clearer visibility on exit velocity, normalized growth, and a stabilized margin structure following the industry transition before it will adopt a more positive view, even as it acknowledged Fiverr remains exposed to longer-term secular trends in flexible work.
Implications for markets and sectors
These analyst actions underscore two broad themes. First, memory and AI infrastructure remain central to many bullish analyst views, with substantial partnership commitments and rising demand forecasts cited as the basis for upgraded or constructive stances. Second, software and gig-economy companies that face direct disruption from AI tools are encountering more skeptical near-term analyst views, with ratings and forecasts trimmed to reflect execution risk and rapid shifts in customer behavior.
On the macro front, the mix of near-term optimism about an AI-driven rally alongside warnings of a substantial rerating for the S&P 500 by late 2027 highlights an uneven risk environment for equity investors: pockets of outsized opportunity in infrastructure-related names, paired with broader valuation and earnings concerns over a longer horizon.
Looking ahead
Investors tracking the AI theme will be watching a set of interrelated developments: the pace and terms of long-term memory supply agreements, the trajectory of DRAM and NAND bit-demand growth into 2027, and corporate execution on AI product road maps at software and marketplace businesses. For portfolio construction, the week’s analyst moves reinforce the importance of distinguishing between firms that underpin AI infrastructure and those whose business models face direct substitution or margin pressure from AI capabilities.
Note: All numerical targets, estimates and company-specific details referenced in this article reflect the analyst notes and company disclosures as described above.