Good morning. Futures take a back seat to a stock-picker’s AI tape: INTC -20% AH is the outlier, and the Nasdaq looks set to trade as a referendum on hyperscaler returns rather than a simple risk-on session. AMZN gave bulls the headline they wanted—AWS revenue $46B, +39% Y/Y, with >40% growth projected for 2027 and 2028—but FCF flips to -$63B this year and -$79B in 2027, so capex payback now matters as much as acceleration. GOOGL brings top-tier Gemini momentum, five-year-high query growth and cloud upside, but Search clicks slowed to +8% with pricing +4%; META’s ad business overtaking Search next quarter (if that plays out) would sharpen the market-share debate. INTC’s below-consensus guide comes with a Kiryat Gat fab explosion that cut the Q4 revenue outlook by ~$300M and gross margin by ~200 bps, with a similar revenue hit expected in Q1. Asia read-through stays firmly pro-AI hardware: China optical checks show 2027 CSP demand converting into firm orders/RPO with no cuts, while TSM targets 6.4Tbps COUPE by 2027. The macro input is supply, not rates—high-end MLCC tightness should peak in 1H27, and advanced wafers, HBM/DRAM, substrates and CoWoS remain constrained. Theme one: hyperscaler growth still works, but the market will pay for returns, not just capex; AMZN’s >40% AWS guide and worsening FCF capture that tension perfectly. Theme two: AI compute broadens beyond GPUs—INTC sees industry server CPU units growing >30% in both 2026 and 2027, a helpful AMD setup, while custom ASICs moving toward nearly 2:1 versus GPUs by 2028 keeps the longer-term NVDA share debate alive. Theme three: own the bottlenecks, not every memory price—LRCX/RMBS/SNDK offer different exposures, but unverified distributor-inventory chatter could pressure pricing. Theme four: RBLX’s soft Q4 guide (2% below Street, roughly a $300M delta) argues against blindly buying every AI-enabled platform dip, even with AI games +50% in the quarter. We’ll hit up AMZN, GOOGL and INTC first, then get to semicap, memory and optics...
Verdict: GOOD QUARTER, MESSIER ALGORITHM. The Street cut numbers, not ratings. AI demand keeps improving, but accounting changes and a softer Q4 cRPO guide make the near-term growth story harder to underwrite. R/r looks better after the target resets; PMs still need proof that model noise, not demand, drives the deceleration.
NOW delivered Q3 cRPO +21% CC, one point above expectations, and SUBSCRIPTION REVENUE +20.5% CC, roughly 50bp above consensus. Operating income beat by 4.9%. Fine headline. Less fine versus the real bogey.
TD Cowen called the cRPO beat softer than both the prior quarter and buy-side expectations. BMO put reported and guided cRPO roughly 50bp below its upside case. Then management guided Q4 cRPO TO +18.5% CC VS +19.5% CONSENSUS, with on-premise and termination-for-convenience deals adding pressure.
"The model noise and lower outlook may revive some AI disruption concerns. The firm remains constructive on medium-term growth drivers against a lower bar." - TD Cowen
That captures the print. No demand collapse, but not the clean delivery PMs expect from a premium workflow compounder.
Post-print targets now run $140-$201, with the actionable cluster at $160-$175. Five firms cut targets, mostly into that range, while the AI bulls held at $175-$201. Freedom Broker went the other way, raising its low-end target to $140 from $120 because it thinks the shares already discount most near-term risks.
The consensus remains Buy/Outperform in name, “show me” in substance. Nobody abandoned the structural AI thesis; several firms simply refused to pay the old multiple for a less predictable cRPO algorithm.
Cantor remains the high-end outlier at $201. Its target uses 11x CY2027 revenue and 30x FCF, versus three-year NTM medians of 13x and 40x. Freedom supplies the cleanest bull framing on the OpenAI relationship:
"ServiceNow’s expanded strategic partnership with OpenAI demonstrates the company is cementing its role as an execution and audit layer for enterprise AI rather than being displaced by AI developers." - Freedom Broker
Bull: AI monetization keeps moving from slideware to production. NOW raised its FY26 AI ACV target to MORE THAN $1.7B, its second increase from the original $1.0B bogey, while production agentic-AI consumption jumped more than 13x y/y. The company signed 19 deals above $5M in net-new ACV, up more than 135% y/y, and 19 of its top 20 transactions included at least six products. Management also raised FY26 FCF margin guidance to 36% FROM 35% and backed the high end of its FY30 revenue target of $32B.
Bear: The near-term algorithm now points below 20% cRPO growth, and the quality of the beat deteriorated versus buy-side whispers. AI deployment support moves revenue from subscriptions into professional services, T4C backlog no longer counts toward cRPO, and acquired-company and federal contracts create additional noise. Add macro deal delays and potential gross-margin dilution from AI infrastructure, and bears can argue that AI makes the model heavier and less transparent before it makes growth faster.
The AI narrative itself isn’t new. The second ACV raise, 13x production consumption and the accounting changes are. Together, they strengthen the medium-term adoption case while weakening cRPO as a clean near-term proxy.
Management’s explanation makes sense analytically. It also creates an obvious credibility test: companies rarely win multiple expansion by de-emphasizing the metric that just disappointed.
"cRPO might not always be a reliable metric for assessing business momentum due to various influencing factors." - CFO Gina Mastantuono, via StoneX
The next quarter therefore matters less for another AI ACV raise and more for whether subscription revenue and cRPO start corroborating the AI bookings story.
Constructive for the enterprise application-layer AI thesis: agents can sit on top of workflow platforms rather than replace them, and large customers still buy broad multi-product suites. That helps the strategic debate around CRM and other systems of action.
Less constructive for near-term software modeling. AI ACV does not map one-for-one into subscription revenue or backlog when deployment shifts into services and contract exclusions reshape cRPO. For NOW, the setup is simple: lower bar, better AI evidence, higher burden of proof. If cRPO stabilizes while AI consumption keeps compounding, the target cuts will look overly cautious. If cRPO keeps sliding, PMs will treat the accounting explanations as narrative cover.
The print earns the benefit of the doubt, even for a consensus long near the highs. Microsoft delivered the two things PMs needed: AZURE ACCELERATED TO +48%, and management guided FQ2 to +49-50% IN CONSTANT CURRENCY versus a roughly 46% Street bogey. AI demand no longer carries the whole debate. Cash conversion does. Stay constructive, but don’t treat the capex relief as permanent.
Five fresh raises move targets from a prior $600-675 range to $630-700, while the broader post-print range runs $595-700. UBS leads at $700; Scotiabank and TD Cowen sit at $650, Cantor and StoneX at $630, with Guggenheim and Oppenheimer at $595 and $625. Ratings remain firmly Buy/Overweight/Outperform, and Scotiabank adds MSFT to its Top Pick list. No one wants to fight this print.
The fundamentals backed the target creep. Revenue grew 19% CC TO $92.3B, adjusted EPS rose 18.4% TO $4.89, and FQ2 revenue guidance of $96.0-97.1B cleared the roughly $95.6B consensus. That matters with the stock around $526-531, just below its roughly $551 high, and already a consensus long.
“Microsoft is transitioning from proving AI demand toward proving AI yield.” - StoneX
Bull: Azure accelerated despite delayed third-party GPU-as-a-Service capacity, so Microsoft produced the beat before every promised shovel arrived. Engineering gains across existing CPU and GPU fleets increased usable capacity, while Foundry, OpenAI, Copilot and Azure give the spending several monetization paths. Management still expects positive FCF in FY27, and UBS sees MSFT as defensive relative to misses elsewhere. Scotiabank even sketches a 54% AZURE UPSIDE SCENARIO for FQ2, though that’s a scenario, not the new base case.
Bear: The capex reduction looks like timing, not discipline. Third-party delays lowered near-term spending—Raymond James puts the revised CY26 outlook around $170B—but Microsoft plans faster FY27 investment, more leasing and potentially debt funding. Scotiabank flags possible NEGATIVE FY28 FCF, while Cantor watches Azure mix, component inflation and infrastructure returns. Positive FY27 FCF helps; it doesn’t settle the longer-duration return question, especially for a crowded long near the highs.
The incremental signal isn’t “AI demand remains strong.” PMs knew that. It’s the rate of change plus the guide: Azure accelerated roughly 500 bps, and the 49-50% FQ2 outlook implies Microsoft can sustain something close to that pace rather than fade immediately.
The financing discussion also moved forward. Management’s FY27 FCF commitment reassured Cantor, but Scotiabank’s debt-market and FY28 warnings put a firmer clock on the cash-conversion debate. That push-pull—not another generic OpenAI headline—drives the next leg of the argument.
“Strong growth acceleration and margin upside warrant a continued rally in the shares.” - TD Cowen
Constructive for GOOGL, ORCL and the broader AI-capacity complex: hyperscaler demand still supports heavy infrastructure spending, and capacity delays don’t equal demand destruction. But Microsoft’s fleet-efficiency gains make the read-through less clean for a simple “more growth means proportionately more hardware” model.
The bigger thematic shift: AI yield now matters as much as AI spend. The winners need to turn capex into accelerating cloud revenue without letting leasing, debt and depreciation consume the equity story. MSFT just moved ahead on that scorecard; it hasn’t finished the race.
THE PRINT BROKE THE NEAR-TERM NUMBERS, NOT THE STREET’S 2027 AI THESIS. Seven firms cut PTs into a $300-$435 RANGE, down from $350-$480, but six kept Buy/Overweight/Outperform ratings; DA Davidson remains the lone Neutral. That’s the consensus: constructive long term, cautious near term, and increasingly dependent on 2027 search monetization and TPU profits. With GOOGL DOWN ~9% AFTER HOURS TO ~$285, just 7% above the 52-week low, this now looks like an estimate-reset and positioning cleanup — not a clean fundamental all-clear.
Revenue reached $128B, +25% CC, roughly meeting or slightly exceeding expectations, but EBIT MISSED THE STREET BY 7% and operating income landed at $40.6B. Search did the real damage: +12% VS +15% CONSENSUS AND ~14% BUYSIDE, with ex-FX growth of 13% marking a three-point deceleration on a two-point tougher comp. FX explains part of it; European retail softness and weak AI Mode monetization explain the rest.
Cloud supplied the counterweight. CLOUD +103% Y/Y, with core GCP and TPU sales driving acceleration, backlog expanding and margins improving. Management kept the 2026 capex outlook and still expects a significant 2027 increase.
Bull: Bulls think the market conflates a monetization lag with demand destruction. Query growth hit a five-year high, Gemini Carbon reached state-of-the-art performance on several benchmarks, and Cloud’s 103% growth validates the full-stack AI strategy. BMO argues investors give little to no credit for TPU EBIT entering the model next year, while Truist expects material TPU sales from FY2027 onward; at Cantor’s 21X FY2027 EPS, the r/r works if search stabilizes and Cloud profit catches up with spending.
"Underlying search engagement is accelerating due to AI but is currently undermonetizing." - BMO
Bear: Bears see both a core-business problem and an AI-returns problem. Search still missed buyside expectations ex-FX, AI Mode monetized below plan, and TPU hardware costs plus higher R&D drove the earnings shortfall. BMO also flagged worse cash burn and $27B REMAINING ON THE ATM, while Truist expects TPU inventory and third-party Cloud capacity to pressure margins in Q4 and FY2027.
"TPU systems inventory build-up and third-party capacity utilization for Cloud are likely to pressure near-term margins in the fourth quarter and fiscal year 2027." - Truist
New: The search miss includes an AI monetization issue, not just FX and macro. The full-stack AI cost curve also hit the P&L before TPU revenue could scale — that pushes the proof point out and makes 2027 the real bogey.
Mostly confirmation: Cloud demand, query engagement and infrastructure spending remain strong. Management largely reiterated the capex framework, so the incremental information cuts against timing and returns, not demand. That explains why analysts cut estimates aggressively but mostly kept their ratings.
For MSFT and the AI infrastructure complex, the demand signal stays strong: Raymond James said both companies confirmed accelerating AI/Cloud growth and continued infrastructure investment. But GOOGL’s margin print warns PMs not to treat Cloud acceleration as automatic EBIT drop-through.
For digital ads, don’t extrapolate GOOGL’s 12% search growth into a clean read on the whole tape. FX, European retail and company-specific AI Mode monetization all contributed. The broader AI debate now shifts from “can engagement survive?” to “when does engagement turn into revenue — and what margin does it carry?”
Clean beat-and-raise. Real category momentum. Less obvious r/r after the run. eBay beat every headline metric and guided Q4 revenue, GMV and EPS above consensus, giving bulls a credible durability argument into 2027. The problem: shares already trade at $122.08, just below the $124.59 high, after gaining 37% YTD. PMs now need to underwrite the multiple, not just another solid print.
GMV HIT $23.2B VS. $22.3B CONSENSUS, with organic FX-neutral growth reaching 13%—four points above the high end of guidance. Revenue of $3.21B beat $3.11B consensus, non-GAAP operating income of $843M crushed both the $794M bogey and the $770–800M guide, and EPS of $1.51 cleared $1.41.
The mix matters more than the headline beat. Focus Categories, re-commerce and C2C grew 20% Y/Y and now drive 72% OF GMV; organic U.S. GMV accelerated to 22%. Live reached a multibillion-dollar annualized GMV run rate while growing at triple digits, and Depop, Vehicles and Magical Listings add separate supply and demand vectors. The reduction in force also freed more than $200M, giving management room to fund Depop without sacrificing the operating story.
PTs now span $120–$150, with the bullish group converging around $140 and BMO at the top. Piper flowed the better results through its DCF, while StoneX raised estimates across all periods. Bernstein took the more interesting line: it lifted its target to $125 but kept Market Perform.
“eBay appears positioned to deliver business momentum into 2027, though the firm questioned the appropriate valuation multiple for the stock.”
That’s the debate. These hikes mostly reflect estimate catch-up after a broad beat—not unanimous support for a sustained re-rate.
Bull: The core marketplace no longer looks ex-growth: 13% organic GMV, 20% growth across the strategic categories and several emerging vectors suggest eBay can carry momentum into 2027 while self-funding investment.
Bear: At 25.6x earnings and a 3.3 PEG, near the 52-week high, the stock already prices in a meaningful execution premium. Live’s run-rate disclosure sounds good, but we still can’t read too much into its durability or contribution from one print.
Not a short on these numbers. Not an obvious chase either. The easy “dead marketplace” narrative has gone; from here, upside needs sustained category growth and evidence that Live and Depop can broaden the earnings algorithm.
Verdict: this is no longer an AI-demand debate. It’s a monetization-and-cash-conversion debate. GOOG FELL 9–10% AFTER HOURS as weak Search monetization, a huge EPS miss and worsening free cash flow swamped another strong Cloud print. The Street kept its bullish ratings but cut targets and multiples. That tells you the problem sits in the earnings algorithm, not the AI narrative.
Headline revenue looked fine at $127.6B versus $127.22B expected, but TD Cowen’s consensus set showed a 0.3% total miss, with Search and YouTube each roughly 2% light. The bigger damage came below the top line: ADJUSTED EPS PRINTED $2.12 VERSUS $3.02 EXPECTED, while operating income missed consensus by 6%.
Cloud beat consensus by 6.9%, although it still missed TD’s own bogey by 3.5%. CLOUD BACKLOG HIT $561B, UP 9% Q/Q, and management reiterated 2026 capex guidance after Q3 spending landed about 8% below TD’s estimate.
Search supplied the more uncomfortable signal. Query growth reached a five-year high, yet SEARCH REVENUE GROWTH SLOWED TO 12% FROM 17% because AI Mode monetizes below traditional Search. Engagement isn’t the issue. Revenue per query is.
Meanwhile, FREE CASH FLOW WORSENED TO NEGATIVE $9.1B from negative $5.9B last quarter. That makes the Cloud backlog less reassuring: demand looks enormous, but the equity needs margins and per-share cash flow, not just more AI workload commitments.
Analysts cut targets across the group, leaving a $350–460 range, while Guggenheim reiterated its Buy and $450 target on Cloud strength. Nobody here abandoned a constructive rating. But the cuts carried real thesis weight: TD trimmed long-term estimates, while others lowered Search, margin and FCF forecasts, partly offsetting those reductions with higher Cloud revenue. Susquehanna even raised 2026–27 revenue estimates while cutting its target on margin pressure — topline growth alone no longer clears the bar.
Scotiabank delivered the cleanest reset, cutting its valuation multiple to 24x 2027 GAAP EPS from 30x. That’s a de-rating, not a rounding error. The firm also flagged TPU’s growing contribution as a drag on the quality of Cloud upside.
“Investors need evidence that AI usage can convert into search revenue per thousand impressions, cloud margins and per-share free cash flow.” — Scotiabank
Exactly. Gemini 4 Carbon may strengthen competitive positioning, but positioning doesn’t pay for the capex.
Bull: AI demand looks undeniable across Cloud and Search, with a $561B backlog, consensus-beating Cloud revenue and five-year-high query growth. If GOOG closes the AI Mode monetization gap while Cloud margins scale, today’s earnings compression becomes the setup for substantial estimate recovery.
Bear: The bear case doesn’t require demand to collapse. AI costs and lower-quality revenue mix simply need to outrun monetization, driving more FCF burn, estimate cuts and multiple compression — with the stock already moving closer to its $268.43 52-week low.
PM takeaway: Don’t use backlog as the all-clear. The next real tells are Search revenue per thousand impressions, Cloud margins and per-share FCF. Until those turn together, GOOG remains a show-me story despite the Street’s still-bullish labels.
VRT just bought back the execution narrative. This was an execution-relief print first, an AI-demand print second. BMO, Wells Fargo and Bernstein all kept positive ratings and nudged PTs higher; the cluster now runs $331-$376 after raises of just $2-$8. The modest target moves matter less than the shift in thinking: Q3 answered the execution concerns that had built since Q2.
Q3 REVENUE HIT $4.31B, +61% Y/Y, versus $3.75B consensus and roughly $563M above the guidance midpoint. Organic growth reached 55%, with all three regions ahead of guidance. Adjusted EPS of $2.21 beat the $1.82 bogey and guidance by about $0.41, while adjusted operating margin of 25.2% came in 70bp above guide.
VRT also raised its full-year revenue outlook by $650M AT THE MIDPOINT, even after accounting for $100-$150M of revenue pushed from Q2. Orders supplied the real kicker:
“Data points suggest demand continues to accelerate for the company’s products.” — BMO, with management highlighting ORDERS GROWTH ABOVE 100% Y/Y
The estimate debate now shifts to 2027. Bernstein sees investors moving 2026 EPS toward roughly $7.30, with 2027 numbers approaching or exceeding $11.00, though some remain closer to $10.50. That spread matters: the bull case needs follow-through, not merely another round of PT tweaks.
Bull: Orders more than doubled, organic growth hit 55%, and every region beat guidance — broad acceleration, not a one-region story. Wells likes the r/r now that VRT has relieved the Q2 execution overhang, while BMO sees a favorable 2027 setup.
Bear: A 46X P/E and recent weakness across AI-infrastructure plays leave little slack if order conversion or margins wobble. The print also raises the 2027 bogey; EPS estimates already clustering around $10.50-$11.00 make “good” results less useful without continued upside.
Net: cleaner story, better execution evidence, higher bar. Watch order follow-through and whether 2027 EPS keeps migrating toward $11.
The print strengthened the story, but the stock already discounts plenty of it. I wouldn’t chase TENB at $48.72 after +107% YTD and +127% over six months. The next leg needs sustained growth acceleration—not another clean beat with a cautious guide. Shares sit just below the $50.21 52-week high and fell after hours as the slower-growth Q4 outlook trumped the Q3 upside.
Tenable beat on REVENUE ($276.4M VS. $271.6M CONSENSUS) and NON-GAAP EPS ($0.63 VS. $0.51), then raised FY26 guidance for revenue, operating income and FCF. Underneath, calculated current billings and net dollar expansion flashed early acceleration. Tenable One drove a RECORD 59% OF NEW BUSINESS, with larger deals and higher ASPs supporting the exposure-management narrative.
Analysts clustered new PTs at $45–60. Stephens delivered the loudest reset—$48 from $34—but kept Equal Weight and anchored its target to 5x FY27 revenue. Scotiabank went $60 from $50 and stayed Sector Outperform; the rest landed at $45–53. That mix matters: the hikes validate execution, but the neutral ratings and current price say the easy re-rate already happened.
“More sustained and material growth acceleration will be required to further increase investor confidence and drive significant upside in the stock.” — Stephens
That’s the right frame. Calculated billings grew 9%, while Scotiabank said billings improved but missed elevated investor bogeys. Bulls got proof of better execution; bears got confirmation that expectations still run ahead of reported growth.
Bull: Tenable One adoption across new and existing customers lifted NRR, deal sizes and ASPs. Hexa (GA since May) gives TENB a credible route into AI-driven exposure-management and automated-remediation budgets, while 78% gross margins leave plenty of operating leverage if growth reaccelerates.
Bear: Revenue grew 9.5%, and the cautious Q4 outlook points to slower growth despite the AI excitement. Scotia’s CISO checks also suggest PHP2Shell’s elevated interest won’t drive outsized customer spending—product buzz still needs to become billings.
My read: better company momentum, trickier r/r. Keep TENB in the prove-it bucket. Another quarter of cCB and NDE acceleration could underwrite Scotia’s $60; without it, this looks like a high-expectation execution story rather than a fresh AI discovery.
Funding clarity bought the reprieve; execution still has to earn the re-rating. ORCL sits at $144.54, DOWN 47% over the past year but UP 15% in the past week, including an 11% post-event pop. Investor Day answered the immediate “who funds the AI build?” question. It did not make the capital intensity disappear.
Management capped incremental capital needs at $20B, split evenly between debt and equity, with fundraising potentially starting this calendar year. Customer prepayments, vendor financing and bring-your-own-hardware arrangements should cover half of the $220B capex requirement through FY28. Management also said it is not considering off-balance-sheet debt structures. Oracle reiterated its $225B+ FY30 revenue target and FY30 EPS guidance.
That kept the bullish analyst cluster intact: Guggenheim stays Buy/$400 and calls ORCL its Best Idea, the other Outperform targets bunch at $215-$325, and JMP upgrades on the enterprise AI stack. Scotia supplied the counterpoint, trimming to $205 from $215 while keeping Sector Outperform. The common thread: funding visibility matters more near term than another demand datapoint.
“Oracle should turn free cash flow profitable in the second half of fiscal year 2028 and be comfortably profitable in fiscal year 2029.” > > — Guggenheim
Bull: At 16x CY27 EPS on Scotia’s numbers, PMs get an independent GPU-as-a-service provider with 22% LTM revenue growth, a $225B+ FY30 target, and a financing plan that leans heavily on customers and vendors. If FCF inflects in 2H FY28, the de-rating looks overdone.
Bear: $20B still means $10B of debt and $10B of potential equity dilution, while Guggenheim sees roughly another $150B of gross capex beyond the FY28 plan to support FY30 targets. The FCF bridge stays back-half loaded; slippage in prepayments, utilization or delivery puts the multiple right back under pressure.
My read: The event moved the debate from financing existentialism to execution. Better r/r after a 47% drawdown, but not a clean balance-sheet story. Next proof points: actual funding, capacity coming online, and progress toward that 2H FY28 FCF bogey.
GEV is now an expectations story, not an execution story. Four firms trimmed PTs into a $1,100-$1,250 range while keeping positive ratings; Oppenheimer went the other way, lifting its target to $1,375. That reads like valuation discipline and a peak-orders debate — not a fundamental rethink.
Q3 gave bulls plenty to work with: ADJUSTED EPS OF $4.81 VS $4.00 EXPECTED, revenue of $12.64B VS $12B, and POWER MARGINS OF 21.4% in the seasonally lightest services quarter. GEV added 11GW of gas contracts at pricing more than 20% above Q4’25 levels, taking total commitments to 122GW, then raised its FY26 targets.
BMO’s broader point matters: GEV can fund capacity, pursue the $9B Trench acquisition, and double its buyback authorization to $20B simultaneously. That supports the “backlog becoming earnings” narrative rather than forcing PMs to choose between growth and capital returns.
Jefferies still sees 36% UPSIDE TO FY30 EBITDA, with a spring-2027 refresh providing the next expectation reset. Its longer-term pitch combines electrification synergies with gas-services economics that the market has not fully priced:
“Electrification upside should become clearer with Prolec/Trench synergies, while the depth of longer-term gas services accretion remains underappreciated but further out.”
Bull: Gas backlog conversion, better pricing, and grid M&A create several earnings drivers at once. The services tail and Prolec/Trench synergies give estimates room to keep moving higher beyond the current gas-order cycle.
Bear: Valuation leaves little room for an orders air pocket, and TD Cowen’s concerns keep the peak-orders debate alive. Jefferies’ roughly 45-50GW FY27 LEVEL-SET makes duration — not another strong Q3 — the next proof point.
My read: The PT cuts do not crack the thesis, but they do change the r/r conversation. Watch order elongation and services accretion; those now matter more than another headline EPS beat.
Good print, steady outlook, full multiple. Stifel raised its PT to $305 from $285 but kept Hold; UBS stayed Neutral at $310. Same message from both: execution looks fine, valuation does the heavy lifting now. With shares at $284.56, those targets leave only ~7–9% upside — the r/r now depends on estimate revisions, not another re-rating.
ADP delivered adjusted EPS of $2.82 versus $2.77 consensus and revenue of $5.5B versus $5.48B. REVENUE +6%, EPS +14%, MARGINS +130 BPS Y/Y — clean execution, modest beat. Shares still dipped premarket, which tells you the bogey sat higher than the headline numbers.
FY27 guidance calls for roughly 6% REVENUE GROWTH AND 10–12% EPS GROWTH, with ADP reiterating 70–90 bps of margin expansion. Rates and balances provide some help, while interest income supports the outlook. Don’t read the EPS algorithm as pure demand acceleration.
Employer Services, roughly 67% of revenue, grew 6% organic constant currency with margins up 130 bps. Upmarket momentum, integrated-solution cross-selling and the Lyric pipeline — including expansion in the UK, Germany and France — give bulls enough evidence that the core engine remains healthy.
PEO looks less clean. Worksite employees grew just 2%; revenue should grow 5–7%, or 3–5% excluding zero-margin pass-throughs. Margins stayed flat as pass-through mix, workers’ comp costs and selling expenses offset progress.
Bull: The “low hire/low fire” backdrop remains stable, and ADP keeps converting modest growth into strong EPS through margins, execution and mix. This remains a high-quality defensive compounder.
Bear: UBS expects growth to decelerate in FY27, PEO margins lack traction, and ADP has re-rated from parity with the equal-weight S&P 500 to a 45–50% EARNINGS-MULTIPLE PREMIUM in six months. That leaves little room for merely fine.
“The multiple appears full and stock movement is now likely tied more closely to estimate revisions.” — Stifel
That’s the right frame. Solid hold; not a table-pounding fresh buy after the re-rate.
MRVL -9% AH as the market votes “great asset, painful check.” Bloomberg says Marvell is in advanced talks to buy Lumentum, potentially announcing a deal next week; LITE +11%, with an EV just above $8B BEFORE ANY PREMIUM, and Piper flags a significant equity/debt funding package while keeping MRVL at Overweight/$400.
Strategically, I get it: high-powered lasers push Marvell deeper into the AI optical bottleneck, where faster data transfer and lower power consumption matter more every generation (LITE says AI-server optical components are sold out through early 2029). Near term, though, the r/r hinges on financing, dilution and integration—not the AI narrative; terms aren’t final, and U.S./China approvals add another gate.
“Optical networking is becoming significantly more important in AI to speed up data transfers and reduce power consumption.” — Piper Sandler
SNDK delivered the kind of print that keeps the AI-memory squeeze narrative intact: $11B REVENUE, NEARLY 5X Y/Y and ~$400M ABOVE ESTIMATES, with adjusted EPS of $49.51 vs. $46.22 consensus and next-quarter EPS guidance of $55-$57 vs. $53. Data center did the heavy lifting (+61% TO $4.8B; 85% GROSS MARGINS), more than offsetting CONSUMER -29%, while the $5B buyback and roughly 6x annualized earnings keep the r/r pointed higher if hyperscaler AI spending and NAND tightness hold. Bear case is simple: this is now a $194.6B MARKET CAP riding supply shortages and explosive pricing, so any crack in data-center demand or NAND pricing hits fast—not sure consumer weakness matters much while the AI side compounds like this.
INTC LOOKS RANGEBOUND, NOT LIKE A CLEAN AI WINNER: shares dropped ~20% after hours as a well-below-consensus Q4 revenue guide and PC weakness through mid-2027 drowned out DATA CENTER +78%, its strongest growth in 15 years. Canaccord blames the PC business rather than AI demand, but after a 150% YTD run, PMs need more than hyperscaler spend flowing through to like the r/r here.
“Firmly a battleground stock and not a clear AI winner.”
THE BAR SITS ABOVE THE STREET. Jefferies stays Buy/$650 into the NOV. 3 print, modeling Q3/Q4 revenue at $13.5B/$17B and looking for 2027 data-center revenue to move toward ~$75B vs $70B consensus; buyside bogeys already run $80–90B after AMD +171% YTD TO A $923B MARKET CAP, so a routine beat/raise won’t cut it. Server CPU upside, Venice/MI450 and Meta’s initial Helios systems keep the bull case intact, but supply commentary matters as much as demand:
“Supply constraints at TSMC’s N2 process node may force AMD to prioritize between its Venice CPU and MI450 GPU products.”
FTNT -25% AFTER HOURS LOOKS LIKE AN EXPECTATIONS RESET, NOT A DEMAND CLIFF: BILLINGS +29%, PRODUCT REVENUE +37%, and $1M+ enterprise deals +51%, but the print missed elevated buy-side bogeys on service revenue and gross margin while product growth broke its five-quarter acceleration streak. Stifel cut its PT to $140 from $175 and stayed at Hold; the broader tape skews Neutral as service growth, fading pricing/inventory advantages, and steeper product deceleration drive the resets (BTIG remains the outlier Buy at $186). Bulls get raised FY26 billings, revenue, and operating-margin guidance; bears get lower service-revenue guidance and a more normalized product quarter after a strong 1H. At $134.72, the r/r hinges on whether this marks a one-quarter air pocket or the start of a slower services compounding story—the headline growth rates alone don’t settle that.
PT CUTS, NOT A THESIS BREAK. Cantor slashed its target to $225 FROM $325 but stayed Overweight after KLAC’s beat-and-raise; targets across Cantor, Evercore, Stifel and Needham now sit at $185-$230, with positive ratings intact. The print backed the bulls—$1.22 EPS ON $4.14B REVENUE vs. $1.18/$4.04B bogeys—while 12-MONTH AVERAGE LEAD TIMES (longer for Gen 4/5) suggest supply, not demand, remains the constraint. KLAC also raised its CY27 WFE outlook to >$200B FROM >$190B and expects to outgrow WFE (>32%, per Cantor), making this a valuation/execution debate against an accelerating AI and process-control cycle—not a demand scare.
“Customer conversations now extend multiple years, providing confidence in multi-year growth.”
Bernstein is sticking with OUTPERFORM / $370 PT, but AAPL heads into the Nov. 2 print with a units problem, not a share problem. Global iPhone shipment growth slowed to +2% Y/Y from +5% last quarter, while SEPTEMBER UNITS FELL 9% and snapped an 18-month growth streak after Apple pushed the base iPhone 18 launch to spring 2027; Bernstein now sees iPhone sales below consensus, partly offset by Services tracking ~1% above. Bulls can call this launch timing—share still rose to 21% from 20% Q/Q and hit 24% in September—but at 39.5x earnings, PMs need CEO John Ternus’s first call to show the delay pushes demand rather than loses it.
The print looks better than the stock. TMUS beat Q3 with $3.22 ADJ. EPS VS $2.89 EXPECTED and $23.58B REVENUE, raised full-year service revenue, EBITDA, FCF and postpaid ARPA targets, yet trades at $149.17, near its 52-week low. StoneX reiterated Buy with a $280 PT, arguing capital-efficient AI can improve growth, network management and customer interactions; the wider analyst tape remains constructive but less enthusiastic, with PTs spanning $214-$280 after a small raise and a trim on mixed estimates. The r/r now looks more about multiple repair than another beat-and-raise—provided AI delivers real efficiencies and SpaceX stays more headline than competitive threat.
StoneX on SpaceX competition: fears look “exaggerated if not entirely erroneous.”
Mizuho’s PT CUT TO $70 FROM $80 (Outperform intact) reads as valuation discipline, not a demand crack, after STM ripped 98% over the past year and 10% in the last week to a 98x P/E. Bulls still own the fundamental tape: December revenue and gross-margin guidance beat consensus, while 2026 AI DATA-CENTER REVENUE SHOULD GROW >6X TO $1.3B, backed by book-to-bill above 2 and backlog above $20B. Bear case is expectations, not orders—great setup, but much less forgiving r/r at this multiple.
SIMO’s print shifts the story from NAND-cycle recovery to enterprise mix and operating leverage — and B. Riley now underwrites much more than a one-quarter beat. The firm lifted its PT to $310 FROM $288 and raised 2026-28 estimates 20%-23% after Q3 EPS of $4.35 VS. $3.30 and revenue of $560.2M, +131% Y/Y, with the stock already +139% YTD AT $231.69. The 2027 bogey now calls for 40%+ SALES GROWTH, enterprise above 50% OF SALES, low-to-mid-50% gross margins and 30% OPERATING MARGINS, backed by secured TSMC wafer supply and ramps across MonTitan, Ferri Auto and DPU boot drives; near-CPU NAND adds a 2H27 option against an estimated $10B TAM. R/r gets less forgiving after that run — PMs now need clean enterprise execution and supply conversion, not just another controller beat — but estimate momentum still points up.
The setup looks skewed to the upside into next week’s Q3 print, but HLIT needs backlog conversion—not another broadband promise—to close the valuation gap. Rosenblatt reiterated Buy and a $20 PT versus the $10.73 share price (86% UPSIDE), arguing record backlog and next-gen broadband demand from Charter, Comcast and smaller operators position the company to beat in Q3 and beyond. The bull case has operating support: LTM revenue grew 11% TO $444M with a 50% GROSS MARGIN, Q2 comfortably beat consensus, management raised FY guidance for the second time, and Q3 guidance already sat above the Street. At 12.2x Rosenblatt’s 2027 EPS estimate, the r/r hinges on execution convincing PMs to pay a low-20s multiple for the pure-play broadband transition.
ETSY finally delivered a clean turnaround print: Q3 revenue of $692M beat by ~4%, adjusted EBITDA beat by ~9%, and GMS rose 8.8% TO $2.65B, while active buyers grew y/y for the first time since Q2’24. Street PTs now sit at $100-$115 after Guggenheim lifted its target to $105 from $95 (Buy); the bull case rests on double-digit app GMS growth, improving marketing efficiency and internal execution—not some broad e-commerce recovery. At $92.35, near the 52-week high and +67% YTD, the easy re-rate looks done; sustained repeat/habitual buyer growth needs to carry the next leg.
RBLX — AI fixes the content funnel, not the P&L yet. Roblox now has MORE THAN 120 MILLION DAUs, AAA ports in HALF A DAY versus TWO MONTHS, and two fully AI-made games cracking the top 20. But the print missed the Street by 2% against a 7% beat bogey, Q4 disappointed, and above-Street 2027 bookings still leak into below-Street EBITDA as fixed costs grow 16%. Buy the distribution moat if you want; don’t call this a clean earnings inflection.
AMZN — AWS killed the deceleration debate; cash flow replaced it. AWS grew 39.3% to $46.0B with $17.8B of operating income, and management now expects ABOVE 40% growth in both 2027 and 2028. That guts the near-term reset bear case, but roughly $77.3B of Q3 property/equipment spend and NEGATIVE $42.4B of reported trailing FCF move the argument to returns and funding (definitions vary; direction doesn’t). External Trainium sales in late 2027 add strategic option value—and more capital intensity.
META — The AI-ad leaderboard could change next quarter. Meta expects its ad business to overtake Google Search for the first time while growing at MORE THAN 2X the rate. Treat that as a forward estimate, not disclosed segment math. Amazon/Muse chatter adds agentic-commerce option value, but no confirmed deal or economics means zero model credit for now.
NVDA — Bottlenecks protect the near term; ASICs and financing risk complicate the long term. The ASIC-to-GPU ratio could move from roughly 1:1 in 2026 to nearly 2:1 in 2028, while Rubin’s final HBM4/NVLink spec cuts temper peak performance without breaking NVL72 rack economics. Nscale credit-guarantee chatter matters more than the reported $1.1B KYEC test-fab prepayment: customer financing would pull tenant credit, lease economics, and residual-value risk toward NVDA’s balance sheet. KYEC helps US test capacity but adds capital intensity.
TSM — Advanced packaging keeps becoming a bigger battlefield. COUPE bandwidth should double to 6.4Tbps by 2027, with 51.2Tbps+ targeted later; FAU coupling equipment looks like the picks-and-shovels angle. The CPU message—manufacture at scale regardless of instruction set, no Intel CPU required—adds foundry optionality. No customer win or volume, so no revenue model change.
UMC — Useful Intel Foundry validation, no numbers yet. Intel tapped UMC to help build a more customer-friendly 12nm PDK, giving Intel some ecosystem credibility and UMC a real process-transfer role. Nice strategic fit for mature nodes. Zero disclosed volume or revenue means this stays optionality, not an earnings catalyst.
LRCX — WFE setup stays good; timing and memory inventories are the swing factors. 2027 wafer-fab spending above $200B versus an illustrative $155B 2026 base supports the cycle, while roughly 12-month lead times help visibility. Facility delays can still push tool acceptance and revenue right. LRCX offers memory-capex exposure without underwriting peak DRAM ASPs, but genuine capacity discipline would eventually hit equipment demand.
RMBS — Positioning signal, not a company update. Tight DRAM/HBM supply and rising interface complexity fit the multiyear AI-data cycle. No orders, revenue, or margin data means no standalone earnings call. Keep this as relative exposure to memory-content growth.
CBRS — The inference-cost benchmark gets attention. A Cerebras/GB300 setup delivered about 13 MILLION TOKENS FOR $1 and up to 1,000 TOKENS PER SECOND per user with four GPUs. That supports the cost pitch, not scale economics across every workload. Portfolio inclusion says PMs want differentiated inference exposure; backlog and revenue conversion remain the missing pieces.
SPCX — Real AI profits, real concentration. SpaceX generated $14.575B of Q3 revenue, with AI contributing $7.56B and its first profit. Anthropic supplies roughly two-thirds of AI revenue and 34.5% of the total, while the maximum $120.5B arrangement carries 90-DAY termination rights after the initial periods. Bulls buy profitable compute leasing; bears buy customer, financing, GPU-residual, and power risk.
POET — The theme helps; earnings proof hasn’t arrived. Shortages across chips, memory, optics, and power force redesigns that favor differentiated optical platforms. The source still trimmed POET. Needs design-win and revenue proof, not another AI-optics narrative.
SOITEC — Same redesign tailwind, same evidence problem. Supply-chain redesign supports differentiated photonics and compound-semiconductor solutions. The source trimmed SOITEC because company-specific earnings evidence remains thin. Treat it as optionality until orders change the story.
PLTR — This reads like discipline, not deterioration. The source took profits while adding other AI exposure, pointing to valuation and crowding concerns. No fresh operating data. Not sure we can read much more into the rotation than that.
ADT — Deal stock now. Thrive Holdings agreed to acquire ADT for roughly $15.3B in cash, and shares jumped about 33–34%. R/r now centers on closing and spread rather than standalone AI upside; the buyer’s agent ambitions add strategic rationale, but no penetration, ARPU, or savings support extra value.
RDRT — Traffic and monetization are rolling over together. Weak guidance, declining US active users, and consumer usage deteriorating from NEGATIVE 2.5% y/y in Q3 to NEGATIVE 4.3% in September make this more than a soft print. Google crawls Reddit aggressively but sends less traffic, while AI Mode drives no referrals to Reddit or known peers—direct pressure on the low-cost acquisition funnel. Login tests could improve targeting, subscriptions, and licensing, but risk sacrificing anonymous reach first.
CAMT — KLAC creates multiple risk before estimate risk. KLAC’s move into higher-end advanced-packaging inspection brings a better-funded competitor that can reuse front-end technology. CAMT retains roughly 150nm inspection capability plus differentiated throughput and metrology. No current revenue loss, so watch customer evaluations and share rather than near-term numbers.
ONTO — Same KLAC threat, same sequencing. High-value advanced-packaging applications now face more competition, but nobody quantified immediate revenue loss. ONTO keeps roughly 150nm sensitivity and differentiated measurement functions. The risk hits the multiple first unless high-end evaluations start going KLAC’s way.
ASML — Installed Base Management delivered. Q3 IBM revenue reached €3.3B, above guidance, as customers funded productivity upgrades. That validates monetization beyond leading-edge system shipments and adds a resilient service layer beneath the equipment cycle. Clean positive.
ASMI — One of the cleaner multiyear semicap signals. Management kept 2027 revenue above €5.7B, reported a solid double-digit SAM increase at 1.4nm versus 2nm, and expects higher share. That gives the leading-edge spending runway more than one node of support.
ALGM — 800V delays push the ramp right. Large data-center programs slipped one or two quarters after hyperscalers postponed 800V programs, weakening sequential growth. Don’t call demand destroyed yet, but timing bogeys need a reset. Bigger risk: the power-architecture transition slips across the supplier base.
AVGO — Another custom-silicon proof point. FuriosaAI is developing its next chip with Broadcom and other partners, targeting 2NM and HBM4/4E. Good for AVGO’s custom-accelerator franchise; also more validation for the ASIC mix risk facing merchant GPUs. The partners disclosed no economics, so keep the read-through strategic.
LITE — Strategic scarcity is real; the rumored premium isn’t. Marvell reportedly considered an acquisition above $80B, but no formal approach or agreement emerged. Without board support, financing, or a regulatory path, this stays merger-arb noise. Don’t mark the premium as realizable value.
DELL — Huge TAM, not booked orders. Dell forecast 200GW of AI data-center growth through 2030, including 100GW from enterprises and neoclouds, and framed a $1T INFRASTRUCTURE MARKET over four years. That expands the AI-server runway narrative. Keep valuation tied to orders, margins, and conversion—not headline gigawatts.
WMT — Retail media and automation pull in different capital directions. The reported $1.4B Vibe deal adds agentic streaming-TV targeting on top of Walmart shopper data, an asset-light commerce-media option. Roughly $6B of expanded Symbotic spending plus $125M for next-generation racking and picking raises the productivity ceiling—and the execution and capital-return bar.
SYM — Big backlog validation, concentrated buyer. Walmart’s expanded agreement could add as much as $6.92B to future backlog, alongside the $125M next-generation technology commitment. Strong anchor-customer validation. Don’t underwrite it like diversified backlog when so much depends on one buyer.
PWR — Demand is converting into cash. Q3 revenue hit $12.01B, adjusted EPS reached $6.06, backlog climbed to $63.7B, and quarterly FCF reached $1.2B. That’s execution across power and infrastructure, not just awards headlines. One of the cleaner non-semi ways to stay long the electrification buildout.
CROX — Beat, raise, and broad-based execution. Adjusted EPS of $3.69 beat the $3.30 bogey, operating margin reached 22.3% (+150 BASIS POINTS), HEYDUDE grew 6%, wholesale grew 14%, and international DTC rose 24%. Management lifted FY26 revenue growth to roughly 3% and adjusted EPS to $14.45–$14.65, with HEYDUDE Q4 revenue guided up 10–12%. Modest top line, real earnings momentum.
HOOD — Limited valuation cushion, weak operating validation. The stock trades near 2.2X ARR after bouncing roughly 25% from recent lows. The source cites 21 total customers and seven additions without specifying units, so that data proves little. This signals crowding and growth expectations, not fresh fundamentals.
AMAT — NOVEMBER 12 is the catalyst. No fresh forecast, order number, or margin target here. AMAT needs to provide the company-specific read on the above-$200B 2027 wafer-equipment outlook. Until then, it’s sector beta with a date.