Brokerages and research teams this week shifted ratings and forecasts across a set of companies whose businesses intersect with the AI investment cycle. The moves ranged from bullish re-ratings underpinned by cyclical recovery and AI-driven opportunity to more cautious notes raising doubts about software revenue durability as enterprise budgets reallocate toward AI infrastructure.
SanDisk - Bernstein lifts target, details an upside scenario
Bernstein increased its price target on SanDisk to $1,250 from $1,000 while keeping an Outperform rating and naming the stock its top short-term pick. Analyst Mark Newman also described a more aggressive, or "blue-sky," valuation scenario at $3,000, which he said would imply approximately 250% upside from current levels.
Newman argued the market is underestimating SanDisk’s earnings power. He noted the shares had fallen about 19% after Google revealed TurboQuant, a memory-compression algorithm that raised concerns about weakening NAND demand. Newman contends those reactions overstated the threat, saying TurboQuant chiefly addresses high-bandwidth memory (HBM) for AI inference and does not materially alter NAND’s broader role in storage and AI infrastructure.
On near-term results, Bernstein projects fiscal Q3 adjusted earnings per share of $14.18, a touch above consensus and above SanDisk’s own guidance range of $12–$14. The team sees Q4 adjusted EPS at $25.30 versus the Street’s $18.78, driven in part by an expected 40% sequential rise in average selling prices (ASPs).
Bernstein outlined two valuation paths. The base-case $1,250 target applies an 11x multiple to what the firm estimates as average earnings of $114 per share across fiscal years 2026–2029. The $3,000 blue-sky figure is rooted in a bull case where NAND prices rise 75% in each of the next two quarters rather than the 55% and 40% increases assumed in Bernstein’s base forecast; under that scenario, fiscal 2027 EPS would reach $224, which is then valued at a 13x peak-cycle multiple.
From a relative valuation perspective, Newman noted SanDisk trades at roughly 9x consensus forward earnings, below the 10–13x range seen in prior upcycles, and at about 0.5x the Philadelphia Semiconductor Index on a forward P/E basis, versus a historical 0.8–1.0x range. He argued this suggests the market either doubts the higher side of Bernstein’s estimates or expects a sharp collapse in NAND prices. The analyst added that NAND pricing only began a meaningful inflection six to seven months ago and that the industry remains structurally undersupplied. SanDisk is scheduled to report fiscal Q3 results on April 30.
Texas Instruments - Stifel upgrades to Buy on expected FCF inflection
Stifel upgraded Texas Instruments (TXN) to Buy from Hold and raised its price target to $250 from $215. The brokerage framed the move around an expected inflection in free cash flow after six years of elevated capital expenditures.
Analysts led by Tore Svanberg said TXN is positioned to take share during the next analog cycle and to return to robust free-cash-flow generation as capital intensity moderates. Stifel expects gross capex to decline from $2.39 billion in 2026 to $2.20 billion in 2027. That reduction underpins a projected rise in free cash flow per share from $8.13 in 2026 to $10.61 in 2028.
The analysts also highlighted government incentives under the CHIPS Act, including an investment tax credit recently increased to 35%, which Stifel models as delivering total benefits of about $1.55 billion in 2026.
Stifel flagged TXN’s pending acquisition of Silicon Laboratories as a meaningful contributor to earnings, projecting the deal could deliver over 10% EPS accretion by 2030 through revenue synergies and lower manufacturing costs as SLAB’s production shifts into TXN’s 300mm fabs.
Data center exposure is a central element of Stifel’s thesis. The team estimates data center revenue, currently near 9% of sales, could reach 20% by 2029, growing at a compound annual rate of 35%. They likened the emerging shortages in power management chips to earlier component constraints, positioning TXN’s 300mm domestic capacity expansion as a response to supply tightness for foundational analog and power-management components used in AI infrastructure buildouts.
On manufacturing, the transition to 300mm wafers is expected to drive up to 800 basis points of gross margin improvement over time, as the company targets more than 80% of revenue on that process by 2030. Stifel cautioned that near-term depreciation pressure may offset margin gains through 2026 before a reversal in 2027. The firm’s bull-case target for TXN is $286.
ServiceNow - UBS downgrades amid doubts on AI resilience
UBS moved ServiceNow to Neutral from Buy and cut its price target to $100 from $170. Analyst Karl Keirstead said he reassessed the company’s positioning against AI disruption and found the earlier view of ServiceNow’s resilience less certain. ServiceNow had been UBS’s sole Buy-rated application software pick on that premise.
UBS summarized three themes from customer conversations. First, while customers did not indicate a desire to replace ServiceNow as a system of record, some are exploring AI to custom-build workflow automation apps or to make service-ticket handling more agentic. Second, customer support was singled out as the area most susceptible to AI-driven headcount reductions, which creates seat-growth risk in the customer service management segment that represents roughly 10% of revenue. Third, UBS did not consistently hear customer enthusiasm for ServiceNow in the agent orchestration layer.
On budgets, Keirstead noted that more than half of enterprise customer discussions now include anecdotes of efforts to contain non-AI software spending as investments in AI infrastructure compete for dollars. Reflecting these inputs, UBS now expects thinner-than-normal earnings beats in the coming quarters, more limited upside to guidance for stable organic constant-currency subscription revenues growth of 19% in 2026, and trimmed cRPO growth expectations, with cRPO exiting 2026 at 16% c/c versus a prior 20% estimate.
Datadog - Guggenheim upgrades, cites Anthropic as a growth catalyst
Guggenheim Securities upgraded Datadog to Buy from Neutral, describing the monitoring and observability vendor as a primary beneficiary of AI-fueled growth in data volumes and IT complexity. Analysts Howard Ma and Joseph DiBartolomeo said prior reservations centered on Datadog’s exposure to OpenAI, which they believed might migrate off the platform.
However, Guggenheim’s checks indicate any migration is taking longer than expected, allowing Datadog more time to build alternative AI customer relationships. The analysts stated that an 8-figure deal announced on Datadog’s fourth-quarter earnings call — identified as the company’s largest new customer logo — is with Anthropic, and that this deal is part of a broader cohort of some 650 AI-native customers.
The firm now models 27% revenue growth for Datadog in 2026 to $4.36 billion, about 6% above consensus. That projection breaks down to core business growth of 24%, OpenAI-related revenue rising roughly 16% versus a prior view of a 20% decline, and revenue from other AI-native customers surging 220% year-over-year to add approximately $160 million.
Guggenheim expects growth to cool to 20% in 2027 as OpenAI’s contribution shrinks, before accelerating to 25% in 2028 as Anthropic and additional AI-native customers fill the gap. The analysts estimated Anthropic’s annual recurring revenue on Datadog could top $350 million by 2028 and suggested lower churn risk relative to OpenAI given the companies’ differing profiles.
On differentiation, Guggenheim pointed to Datadog’s purpose-built storage and query engines, specifically its Monocle metrics database and Husky event engine, as difficult-to-replicate capabilities that blunt concerns about commoditization from open-source telemetry tools. The firm also highlighted the Bits AI suite, which automates incident detection and response, as a promising monetization avenue and potential vehicle for outcome-based pricing.
GitLab - Guggenheim trims to Neutral, flags seat-model and AI threats
In a separate note, Guggenheim downgraded GitLab to Neutral from Buy and removed its $50 price target, citing greater AI-related risk and a lack of near-term catalysts. The focal concern is net revenue retention, which the firm notes has fallen from 133% in fiscal 2024 to 118% in fiscal 2026. Guggenheim expects net revenue retention to exit fiscal 2027 around 113%, beneath management’s expectation of approximately 115%.
The firm attributes the elongation in retention to budget reallocation toward AI coding tools and management’s acknowledgment that over 30% of annual recurring revenue is under pressure. Guggenheim parsed that stress as roughly 20% coming from price-sensitive small and mid-market customers and 12% from the public sector.
Guggenheim highlighted a structural vulnerability in GitLab’s seat-based model: incremental usage does not directly generate revenue under the current approach, meaning the company monetizes growth largely through tier upgrades. As AI coding assistants improve and developers lean on them to produce more code, the incentive for customers to expand GitLab spending could weaken even without headcount reductions.
The brokerage was skeptical of GitLab’s new Duo Agent Platform (DAP), a credit-based model, citing limited initial interest from customers. The analysts noted that about 70% of GitLab’s revenue derives from self-managed customers who would need to upgrade to access DAP. Management has guided for minimal DAP revenue contribution in fiscal 2027.
Investor tools and screening mention
The reporting also referenced an AI-driven stock-screening product that evaluates Texas Instruments alongside thousands of other companies using more than 100 financial metrics, noting the tool identifies stocks offering the best risk-reward based on current data. The mention included prior winners identified by the tool but did not alter the fundamental analysis conveyed by the research notes above.
Implications across sectors
Collectively, these analyst moves signal divergences in how sell-side research teams are parsing AI-related demand and the broader recovery in semiconductors and enterprise software spending. Semiconductor exposure and manufacturing strategy dominate the bullish narratives for SanDisk and Texas Instruments, where price-dynamics for NAND and a shift to 300mm capacity are central to upside cases. Software and platform businesses face a more nuanced set of dynamics: some application-layer vendors are being re-rated lower amid concerns that AI initiatives will compress traditional software seat and subscription economics, while observability and infrastructure-focused firms with deep technical moats and AI-native customers are being viewed more favorably.
The near-term path for semiconductor pricing and the pace at which enterprise budgets reallocate to AI infrastructure versus incumbent software will materially affect outcomes across these names. Several of the research notes explicitly link their scenarios to specific price or growth assumptions and to identifiable catalysts such as NAND price trajectories, the timing of corporate migrations, or the absorption of capex-driven margin headwinds.
What to watch next
- SanDisk fiscal Q3 results, which Bernstein noted are due on April 30, and any commentary on NAND pricing and ASP trends.
- Execution on Texas Instruments’ 300mm transition and the timing of capex normalization that Stifel projects will drive FCF accretion.
- Customer adoption patterns for Datadog among AI-native firms and any incremental details on Anthropic’s engagement or the broader 650-customer cohort.
- Evidence of seat-demand pressure or uptake of agentic products for ServiceNow and GitLab, including traction for initiatives such as GitLab’s Duo Agent Platform.
These points are grounded in the assumptions and customer-check themes highlighted by the analysts and reflect the material drivers they used to change ratings and targets this week.