Good morning.
Futures mixed into a pre-Warsh squat — nothing wants to commit before 10pm ET, and volume will be garbage. The movers tell the positioning story: PYPL -15% (M&A premium dead, Advent/Stripe walked), MRVL -8% (beat, guided, and still sold). THE MRVL PRINT IS THE MORNING'S REAL SIGNAL. Revenue +37%, optics driving F27/F28 up — and the stock drops 8%. That's the bar problem, officially spreading from NVDA to custom silicon: the street had Google's order magnitude baked into FY28, and the guide didn't confirm it. Either it's expectations, or share shifting to AVGO — TAIWAN Q4 SHIPMENT DATA WILL ADJUDICATE. Demand is not the question — ASIC visibility runs to 2028, Anthropic's sniffing around a training chip, NVDA's guiding ~16 months of capex visibility. Memory's the tightest tape but the rent's getting contested: CXMT H1 revenue up ~10x, Micron Taiwan unions ~80% ready to strike over profit-sharing, Samsung leading the custom NVHBM race. EVERYONE WANTS A CUT — labor, governments, CXMT, LTA-hunting customers. Warsh is the swing factor tonight: Bund 10Y at 15-year highs, Brent above $90, BNP calling an ECB hike "all but nailed on" — the discount rate backdrop is stagflation-shaped. Goldman's already packaging TRS on hyperscaler credit because CDS is widening (Meta 75–92, NVDA 54–82). If Warsh can't deliver a clean inflation framework, the AI credit leg tightens into next week, and multiple compression follows. Watch three things: the MRVL distribution argument, Micron's strike vote, and whether Anthropic drops the S-1 before Monday — the Pentagon cloud just got cleared by a federal judge.
We'll hit up MRVL, MU, and SNDK first, then get to the software security group — CRWD, OKTA, WDAY.
The demand inflection narrative is no longer theoretical. CRWD printed NNARR of $333M, +51% Y/Y — a massive acceleration from FQ1's +32% — and the ARR beat was the largest in years. The street is scrambling to get PTs into the $250 zone, and only Bernstein is still standing at $119 with a Market Perform. That's not a bearish statement on the quarter; it's a valuation objection at 30x CY27 revenue. The question now is durability, not direction.
PT range: $119 (Bernstein, Market Perform) to $260 (RBC, Outperform). The rest of the street clusters $240–250. Consensus is Overweight/Buy. The bull/bear axis has shifted from "can CRWD reaccelerate?" to "how much of the AI security budget can one platform consolidate, and what multiple does that deserve?"
The new information: demand is inflecting. Management called it out explicitly — an inflection driven by agentic AI threats, and it's broad-based across the Falcon platform. Piper calls it a "step-function change." DA Davidson points to record pipeline, accelerating Flex, early AIDR contribution, and customer urgency post-Mythos.
We already knew CRWD was the cybersecurity consolidator. We knew the platform was sticky and the go-to-market was working. What we didn't know: whether growth could reaccelerate at this scale. Now we have FOUR STRAIGHT QUARTERS OF ACCELERATING ARR GROWTH. That's the "show me" moment for PMs who were waiting for proof before paying up.
Bull case: The demand environment changed, not just the comps. Agentic AI created a new threat surface, security is the first budget line, and CRWD is the vendor with the platform to consolidate spend. The results prove it. RBC raised estimates and says 2H FY27 numbers are still biased upward. The acceleration is a step-function, not a one-off.
"CRWD delivered blowout FQ2 results w/ NNARR of $332.8M, well ahead of investor bogeys & accelerating to +51% Y/Y vs. +32% Y/Y NNARR growth in FQ1." - DA Davidson
Bear case: It's priced like a step-function that hasn't fully proven out. Bernstein's customer conversations suggest demand has NOT fully inflected — the acceleration may be concentrated among platform vendors right now, not broad sector demand. At 30x CY27E revenue, there is zero room for a 2H miss. Piper said "room for misexecution is virtually nonexistent" — that cuts both ways. One wobble and the multiple compresses faster than the growth rate.
The next set of security prints is the tell. Watch PANW, ZS, S (or whoever prints next) for language around agentic AI threats, pipeline acceleration, and platform consolidation. If the inflection is real across the sector, the whole security complex re-rates and CRWD's premium holds. If it's CRWD-specific, then the AI security leadership position is worth more — but the peers are set up to disappoint against their AI narratives. Either way, this quarter confirms security is the first place CIOs deploy AI-related budget. That's a read-through for the entire TMT software complex.
THE DEBATE JUST FLIPPED. No longer "is AI capex real" — the question now is whether Jensen can physically ship enough silicon. Management dropped a first-ever full-year guide: FY28 revenue growth ~70%, implying AT LEAST $690B vs. the Street bogey of $573.5B. Demand signals say 100% growth. The guide says 70%. Supply is the ceiling, not demand.
Nine firms out today, PTs ranging $270-$515. The cluster sits at $300-350: Benchmark $335, Truist $346 (raised from $307), Cantor $350, RBC $330 (from $300), Freedom Broker $305 (from $275), UBS $300. TD Cowen holds $275 with Top Pick designation. Argus sits at $270. Raymond James is the outlier at $515 — Strong Buy, citing FY27/FY28 outlook. Consensus is emphatically Buy, and the through-line is uniform: demand visibility has never been this strong, and at 32x with a 0.29 PEG, the valuation debate is basically over. Truist's math: CY27 EPS of $15.71 (raised from $12.29), 22x multiple, a 12-turn DISCOUNT to datacenter peers. UBS reads the guide as CY27 EPS north of $16.
Three things, and none of them are the Q2 beat (which hit every line of the P&L).
First-ever full-year guide — and management framed it as a supply ceiling, not a demand estimate. That's a watershed. They're telling you what they can ship, not what customers want. The gap between the two is the entire ballgame.
The demand-supply spread. Truist flags demand supports 100% growth, supply caps it at 70%. That 30-point gap is forward estimate upside. Benchmark's Acree nails it: "unusual execution room" from a quarterly revenue base approaching $100B.
Gross margin is now a known headwind. FY28 guide at 72-73% vs. 74% TTM. Memory cost inflation is the culprit. RBC says it matched their recent model — not a shock — but it's the one blemish on an otherwise clean print.
The under-discussed part: $88B in FY29 purchase commitments, Rubin ramping hard, CPU revenue doubling next year, ACIE (non-hyperscaler customers) outpacing the big guys, and neocloud revenue-sharing creating a recurring revenue stream. This isn't one quarter of visibility — it's a multi-year demand map.
Bull: The demand function is no longer theoretical. Customer forecasts support roughly doubling revenue. This is the highest-conviction growth story in the market, and the guide is the conservative version of reality. Every incremental wafer is incremental revenue, and the supply ceiling gets raised every quarter. The 0.29 PEG makes the r/r asymmetric — even a modest de-rate leaves you with compounding growth most of the S&P can't touch. The supply-constrained guide is a setup, not a setback.
Bear: The balance sheet is doing heavy lifting nobody's pricing. Freedom Broker flags "quality-of-earnings and balance sheet problems" — supply commitments, equity stakes, customer guarantees, extended payment terms, weakening cash conversion. Memory cost inflation squeezes GM. And the 100% demand number is customer self-reported — neoclouds and strategic investors have every incentive to overstate. When the music stops, some of those commitments get renegotiated. A "supply ceiling" guide is also a tell: they can't convert demand to revenue on the timeline the market wants.
"We believe the difference between the demand signal and the supply-backed guide creates unusual execution room for further estimate upside even from a quarterly revenue base approaching $100B." — Benchmark
"The disclosure shifts the debate from whether the cycle persists to whether Nvidia can physically serve it." — Freedom Broker
This is a sector signal, not just an NVDA print. Truist explicitly says the broader AI trade — particularly high-AI-concentration suppliers — starts to recover. Semis traded up in sympathy. The logic: if NVDA is supply-constrained at 70%, the suppliers (memory, HBM, advanced packaging, power, networking) face the same demand curve with easier comps. AVGO, MRVL, TSM, and the memory complex get a bid.
The GM compression on memory costs is the tell — HBM pricing has real power right now. That's a direct positive read for MU and the memory names. They're the binding constraint inside NVDA's supply ceiling.
One more thing for PMs to watch: NVDA just set a precedent. First full-year guide, framed as supply ceiling. Expect the rest of the complex to adopt the same framing — guide to what you can serve, let demand sit above the guide. That structurally compresses risk for the whole trade. The AI cycle just got a multi-year visibility stamp from the one company that matters most.
THE ACCELERATION IS REAL — AND THE STREET IS PAYING UP FOR IT. Post-Q2 beat, this is no longer a show-me story. cRPO re-accelerated to ~14% YoY (from 12.2% last quarter, vs. 10.8% consensus), bookings were the strongest ever for a non-Q4 period, and management raised guidance by 2x the beat. The stock is up 85% in six months — and PTs are chasing it.
The street view is now firmly constructive: PT range is $143-$200, with Truist, Needham, and Cantor at the $200 top end. That's a massive re-rating from the prior $105-$180 cluster. Piper's $160 is the outlier — and notably, it's below the current price. That's your tell on how far this move has run.
The headline numbers matter, but the rate of change matters more:
"Management noted dozens of Okta for AI Agents deals, several of which are greater than $1 million and the average AI agent deal size exceeds traditional deals."
That's the setup for the next leg. Identity is becoming the control plane for agentic AI — and Okta is the early leader. But the guide says the dollars don't show up in the P&L yet.
BULL: The inflection is finally here. cRPO acceleration is the leading indicator that points to durable revenue re-acceleration — the "finally showed what we've been long waiting for" dynamic, per Bernstein's Peter Weed. Large enterprise execution is improving, new products are 30% of bookings, and AI agents provide a visible up-sell vector that most software names can't point to. At 77% gross margins and accelerating forward metrics, the operating leverage story is intact. Truist's raise to $200 from $165 is aggressive — but it's justified by a 330bps beat versus their cRPO estimate.
BEAR: The stock is already pricing in the good news. Piper's Neutral with a $160 PT below the current $165 price is the honest take — at ~8x CY27 revenue, you're paying for the AI narrative to be real, not the core business to be solid. The cRPO acceleration is off a depressed base; the 14% growth is still well below the 25-30%+ this stock traded at during its peak multiple days. AI monetization is a FY28+ story, and the guidance raise explicitly assumes minimal AI contribution in 2H. Net: the de-rating risk is real if AI demo days don't convert to committed spend within the next two quarters.
"The risk-reward profile is balanced at current levels." — Piper Sandler
That's the bearish rebuttal in one sentence. The stock ran 85% in six months. The good news is already in the price. You're now paying for acceleration to persist — and for AI agents to start contributing to bookings.
Okta is the canary for the identity security complex — and for enterprise software more broadly. What this print says:
1. Enterprise budgets are rotating to security controls for AI, not away from them. Identity is the choke point for agentic access. If Okta's AI agent deal pipeline is real (and KeyBanc's commentary suggests it is), watch CrowdStrike and Zscaler for the same dynamic. 2. Large enterprise spend is back. Record non-Q4 bookings with success in 7-figure deals. This is corroborated across the security space — the digestion phase of 2025-26 is over. 3. The "AI is only for mega-cap infra" narrative is dead. Okta is a pure software name getting AI-driven upside. That's a signal for the whole application layer.
The risk? This is a forward-looking story. The FY27 guide is conservative by design. No one is paying for FY27 numbers — they're paying for the FY28 AI agent ramp. If that slips, a 50% drawdown is on the table. But in a market that's starving for growth with visibility, Okta just gave PMs the cleanest re-acceleration narrative in software this quarter.
Verdict: AI narrative finally has receipts. NNAOV at a 4-year high, Agentforce ARR near $1.5B, and an Anthropic partnership that changes the model — the street rotated from "show me" to "how high" in one print. (Stock trades low-to-mid $240s, ~$200B market cap.)
PT range got dragged up: $240 (UBS, Neutral) at the low end, $300 (Truist, Buy) at the top. The bull cluster sits $266–$280 — Monness $266, Guggenheim $270, Oppenheimer $275, TD Cowen $280. Cantor holds Overweight at $250. KeyBanc stays Sector Weight, no PT. Net: Buy consensus, wide target band. The debate isn't whether Salesforce is good anymore — it's how fast the reacceleration compounds.
Bulls anchor on three data points: 1) NNAOV GROWTH AT A 4-YEAR HIGH, 2) AGENTFORCE ARR ~$1.5B WITH PRODUCTION CUSTOMERS UP 70% QoQ — production, not pilots, 3) a $100M organic raise to FY27 guidance. TD Cowen admitted the ~6% cc subscription growth missed their ~7% bogey but called the cRPO print and AI ARR beat (second consecutive quarter) the real signals.
Truist had the most quotable take:
"NNAOV Best in 4 Years; Expanding Set of Growth Accelerants; Maybe SaaSrenAIssance Not SaaSpocalypse; Reit. Buy; We feel that 2Q was an important proof-point qtr for Salesforce w/ the pieces needed to drive sustainable organic growth beginning to come together." > — Truist Securities
Bulls: Salesforce just became the AI platform trade. Claudeforce (GA in September) positions CRM as the data, context, and action layer for third-party agents — it monetizes AI without needing to win the model war. Consumption KPIs are inflecting, NNAOV is accelerating, and the multiple still isn't stretched: P/E ~23.9, PEG 0.6. Dreamforce September 15–17 gives management a stage to pound the Agentic Enterprise narrative. Truist's framing is the right one — SaaSrenAIssance, not SaaSpocalypse.
Bears/neutrals: Core subscription growth still lagged — ~6% cc versus a ~7% bogey. KeyBanc stays measured for a reason: Claudeforce is "potentially powerful" but early, and they want harder organic growth evidence before underwriting the reacceleration. UBS's $240 Neutral is the valuation argument — the stock already scores a recovery, so what's the next leg? Monness flagged the macro environment as "treacherous" and competition dynamic. AI momentum doesn't override a cautious CIO.
The Anthropic deal is the genuinely incremental piece. Not just another model partnership — it signals Salesforce is building an AI aggregation strategy rather than hitching its wagon to a single LLM. KeyBanc also flags Claude Cowork capabilities getting replicated inside Salesforce. That's a product expansion vector the market hadn't priced. (Truist calls it an "expanding set of growth accelerants.")
Already known: Q2 beat, cRPO strength, AI ARR momentum, H2 reacceleration guide. All in the script. The PT raises are the street catching up to the print.
This is the loudest "software is back" data point in the tape. Salesforce was the SaaSpocalypse poster child, and it just printed a 4-year NNAOV high with AI ARR compounding. If that doesn't crack the application-layer short thesis, nothing will.
Claudeforce also validates a bigger theme: MODELS ARE COMMODITIZING, DISTRIBUTION WINS. Good for the app layer across SaaS, bad for anyone selling a model as a moat. And if Agentforce is at $1.5B ARR and compounding this fast, PMs should be hunting the next enterprise agent trade. This isn't a CRM story — it's the template.
VERDICT: VEEV JUST PROVED THE VERTICAL AI MOAT IS REAL. Seven analysts, seven PT raises to $283–$325 after a quarter where revenue, EPS, billings AND guidance all beat — and management raised FY27 revenue by MORE than the quarterly beat. A top-20 pharma pushed Agentic Call Report into production across its entire U.S. field force. That's the single most concrete enterprise agentic AI print of this earnings season. The only question left is what you pay for it.
Bull — Vault CRM is becoming the default, not the challenger: 12/20 top-20 pharma, 180+ live, and every win is a multi-year migration annuity that locks in the installed base through the next upgrade cycle. The AI moat is real — a top-20 paying for agentic AI on regulated workflows, at the exact spot where a generic LLM can't go without the compliance layer on top. Billings +19%, guide up more than the beat, Rule of 50.
"Veeva Systems remains a high-quality software company with relative insulation from AI disruption and is positioned to sustain Rule of 50 status." — Piper Sandler
Bear — You're paying 43.8x P/E for sub-20% growth after a 34.5% six-month run. The stock sits 5% below the $310.50 high — revisions are chasing a tape that got there first. Aspen CRM at half price says the core vertical TAM isn't enough; that price war with Salesforce funds out of the 44.8% op margin that justifies this multiple. And "insulation from AI disruption" is what legacy vendors say the year before agents chew their workflow layer — one top-20 decision goes sideways and the whole narrative cracks.
All seven firms raised, all at Buy/Outperform/Overweight. Range: $283 (TD Cowen, from $235) to $325 (RBC, from $275). Cluster: $295 (Jefferies), $300 (Stifel), $305 (Truist, from $262), $310 (Needham), $320 (Piper, from $295). Median ≈ $305 — roughly 4% above the pre-print spot. TD Cowen's $283 already sits below the tape. That's what a momentum move past the models looks like.
The collective message isn't cheap — nobody says that. The message is quality, visibility, and rate of change:
"Normalized billings growth of 19% indicates strong underlying demand and improved visibility." — TD Cowen
The visibility claim has teeth: the guide raise of more than the quarterly beat says backlog, not hope. As positioning, the strongest r/r is owning VEEV into the last two top-20 pharma decisions, not shorting the strength. This is a compounder — but the multiple is the risk, not the story.
VEEV is the first real proof that regulated vertical workflows survive the LLM disruption trade. And it's a direct negative for Salesforce in life sciences — 12 of the top-20 pharma logos now on Vault CRM, two left. Every one of those migrations is a logo SFDC used to own. Aspen CRM at half price extends the attack into horizontal CRM, so the peer read is: Salesforce's vertical wedge just got weaker.
The agentic AI angle matters more than the CRM war. Agentic Call Report at scale across an entire field force validates the "AI agents > co-pilots" narrative — and tells you the next big software winners are the ones shipping production agents in compliance-heavy workflows. That's the theme PMs should be screening for today.
Printing $241 after a beat-and-raise that still fell after hours. That tells you everything about how high the bar got — the Google ASIC headline had everyone positioned for a blowout, not a "slightly ahead." The real story isn't the quarter. It's the FY28 guide jump to $18B. That number is the entire bull case in one line.
Record revenue $2.74B (+13% QoQ, +37% YoY), non-GAAP EPS $0.94. Both just ahead of consensus. Stock STILL went red after hours. That's not a fundamental message — that's a positioning message.
The guidance is where the juice is:
The Google TPU business doesn't hit until FY29. Management says custom business more than doubles in FY28 and exceeds $10B in FY29. So the FY28 guide is mostly the existing book, with Google upside still ahead of the numbers. That's a nice problem to have.
Wolfe is the loudest bull — PT to $280 from $250, Outperform, modeling $6.57 FY28 EPS on $18B revenue at 57.5% GM.
"The guidance improvement was driven by broad-based strength across connectivity franchises and robust demand in core businesses."
The key nuance: upside comes from BOTH better demand (optical + custom) AND improved supply. MRVL is actively securing incremental supply in an industry-wide constraint environment. You don't hear that every day in semis — that's a demand-pull signal.
Morgan Stanley and Goldman both parked in the lot. MS to $246 Equalweight (from $224), Goldman stuck at $195 Neutral. The stock trades ABOVE both — Goldman's sitting in the "overvalued" bucket.
The critical pushback from both: Google was ALREADY in prior guidance. MS says expectations ran too hot on the announcement. Goldman thinks the stock stays range-bound until the custom compute ramp in 2H shows up in actuals. They're watching the shape of the Microsoft ASIC ramp, not the press releases.
They have a point. You can't beat-and-raise and still fall after hours without acknowledging the tape was ahead of the story.
Visible PTs range from Goldman's $195 to Wolfe's $280 — with Rosenblatt at $300 on the pre-earnings bull path. That's an enormous spread. The Street hasn't converged on this story.
The bull case is real: MRVL is the only scaled custom ASIC play with two hyperscaler logos. The bear case is also real: expectations have run so far that a beat-and-raise now produces a red candle.
October Analyst Day is the catalyst. Everything between now and then is noise. The one thing to watch in the interim — the shape of the Microsoft Maia ramp in 2H. If that shows up in the next guide, Wolfe's $280 starts looking conservative.
Both PTs jump — TD Cowen to $95 from $70, BofA to $108 from $90 — and both still sit at Hold/Neutral. That tension IS the story. Great quarter, real execution improvement, AI starting to contribute. But nobody's ready to call the multiple re-rating. Fair.
FQ1 beat was clean: $0.70 EPS vs $0.58 est, revenue $478M vs $470M est. CLOUD REVENUE +$18M QoQ vs $13M Street est — that's the headline. Sales-led sub beat guidance by $5.5M. Full-year raised to 17% CC growth, a 100bps acceleration. RECORD 80 NET NEW $100K+ ACV CUSTOMERS. Gross margin 75.6%. Quality asset, no two ways about it.
TD Cowen was candid about the math:
"The results were better than expected. The implied second-half acceleration leaves little room for upside."
That's the crux. You're buying the re-rating because the guide implies the beat repeats. Fine, but that's a priced-in scenario.
Bull: Execution is finally matching the product story. BofA points to improved sales execution and prior commitments converting — that's a 3-4 quarter narrative, not a one-off. AI contributions are increasing. At 3.4x EV/CY27 revenue vs 5.8x infra peers, the stock carries a valuation catch-up option if the AI story keeps building. BofA says it outright: "improving execution leaves it more constructive."
Bear: Billings growth of 6% missed expectations. cRPO flat at 20% growth — stable, not accelerating. NRR down 100bps to 111%. And the AI contribution, while increasing, is still "below the mid-20s level typically associated with established AI beneficiaries." That's a nice way of saying Elastic is an AI story in progress, not an AI story in the numbers. Both firms had to raise targets 20-35% and still wouldn't step to Buy. That's the tell.
Watch billings and cRPO next quarter. If billings reaccelerate with the execution narrative, the Hold crowd flips and this is a $100+ stock. If it's another 6% print, then NRR decel plus soft billings means the re-rating was just a sympathy bounce with the sector. The 75.6% GM and $8.7B market cap say this deserves a premium — but 16% LTM revenue growth doesn't scream buy-the-dip at these levels.
Verdict: Clean quarter, ugly forward tell. WDAY beat FQ2 bogeys — non-GAAP EPS $2.75 vs $2.61 consensus (+5.4%), subscription rev $2.471B vs $2.456B. But nobody pays you for the quarter you just printed. They pay you for the slope. The slope is decelerating. SUBSCRIPTION GROWTH 13.9% YoY, DOWN FROM 14.3% LAST QUARTER. Backlog growth down to 14.2% from 15.4%. And the big one: TOTAL SUB BACKLOG MISSED CONSENSUS BY ~$1.2B — $27.4B vs $28.58B. That's the number that matters.
Nobody's moving on the print. Citizens reiterated Market Perform. William Blair reiterated Outperform. No PT hikes, no dramatic cuts — the street is in "show me more" mode. (When a stock's already up 71% off the April low, the bar for fresh enthusiasm resets.)
The company attributed the lower total backlog to a mix-shift toward growth in the customer base, with renewals having "a slightly shorter duration than a net new offering."
Bear: Seat-based software is structurally challenged. Employment growth is soft. And the backlog data says customers are committing to less. Analysts reiterating without PT changes isn't conviction — it's watchfulness. Blair flags financials traction with enterprise customers as a key risk, and that cross-sell is the whole second act of the growth story. If financials doesn't land, 12.4x FCF is a value trap, not an opportunity.
Beat-and-raise. AI ARR tripled. Stock still down 4% after hours — that's the whole ballgame.
SentinelOne printed a clean Q2 FY27: revenue $292M (+21% YoY) just above consensus, ARR $1.218B, and NON-GAAP OP MARGIN OF 10.5% vs 8.4% consensus — the big beat. RPO growth accelerated to 45% YoY. Net-new ARR topped estimates by ~$5M. Large-customer NRR improved for the third straight quarter. On paper, this is the platform inflection story finally showing up in the numbers.
The problem? The stock ran 73% in six months. Shares sit at $22.71, near the $23.95 52-week high. Both houses land at $25 — Citizens raised from $23, Baird just reiterated. From this close, that's ~10% upside. The aftermarket fade tells you the tape was already there.
The quality metrics all moved in the right direction:
"Net-new ARR exceeded estimates by approximately $5 million, RPO growth accelerated to 45%, and operating margins came in around 10% — the platform momentum is broadening, not just endpoint share gains."
That's the Baird/Citizens consensus in one breath: the beat is real, the quality is real, and the story is now platform-led.
Bull: RPO acceleration plus AI ARR tripling is the proof point bears demanded. Margins at 10.5% with 21% growth means the model works while they invest. Large-customer NRR inflecting + raised guidance = the setup for H2 reacceleration. At $25 PT, limited downside if they execute.
Bear: The stock's up 51% YTD versus R3K's 12.5% — the easy money's made. Both PTs at $25 cap the upside. A 4% aftermarket fade on a beat-and-raise is the market telling you expectations ran hot. InvestingPro's fair value screen has the stock at ~0.95x, i.e. slightly overvalued here. NRR is improving, but from a base that was never screaming. 21% growth is fine, not exceptional, in security land.
Great quarter, no argument. But this was a "sell the news" reaction — the position was long, the print was good, and the stock still got sold. That's the tell. Fresh money at $22.71 with a $25 target is a low-velocity, low-dollar trade. If you're already long from the teens, the trim decision writes itself. If you're not, wait for the post-earnings drift to settle and see if $21.50 holds.
FIVE RAISES, ONE STORY: NTNX IS NOW A SHARE-TAKE COMPOUNDER, NOT A RELAXATION TRADE. Stock's up 71% in six months to ~$70 and the sell-side still feels behind — PTs jumped to a $65-90 range (RBC led at $90 from $84; Piper's $60→$75 was the steepest hike, a 25% PT bump in one morning).
Beat and raise, plain and simple. Q4 FY26: non-GAAP EPS $0.60 vs $0.49 consensus, revenue $757.1M (+16% YoY) vs $738.31M est. But the headline metric is ARR: $2.549B, +16% WITH ACCELERATING GROWTH, sitting on an 87% gross margin (yes, 87%). FY27 guide of 11-13% revenue growth says the engine keeps turning.
"A solid finish to its fiscal year with outperformance across the board and annual recurring revenue acceleration." — RBC Capital (Outperform, PT $84→$90)
The growth drivers are stacking, not single-threaded: VMware replacement cycles, external storage partnership contributions, AI/Kubernetes adoption, and future-start-date orders from big government contract wins. Piper's framing is the most useful — top-line acceleration with positive lead metrics, plus potential FCF upside through the year as supply chain headwinds ease.
Bull: This is a migration supercycle with multi-year visibility. VMware displacements are sticky, long-duration wins that convert into durable ARR. External storage partnerships open a TAM legacy HCI never touched. Government contracts in the bag de-risk FY27. At 87% gross margins with accelerating ARR, the operating leverage story is just starting to compound.
Bear: Hardware supply constraints are capping near-term FCF guidance — that's a real issue for a standalone cash-flow narrative. And the available-to-renew base is "larger but slowing as anticipated" — the first whisper of the decel that follows every pull-forward spike. At $70, the beat is priced in. The next leg needs FCF proof, not just ARR headlines.
Quality name, quality multiple, and the debate is now timing, not direction. We lean constructive but want to see the FCF re-acceleration before chasing at these levels. Supply chain commentary next quarter is the swing factor — that's the one number that turns this from a hold into a buy or a trim.
The beat is real; the raise is mostly M&A. ADSK printed Q2 FY27 at $3.30 EPS vs $3.12 cons, $2.05B revenue vs $2.01B — revenue +16% reported, +14% CC. Billings ~2% above. FCF beat. And the stock still caught a "mixed" tape. That's the tell: the quality of the beat is the debate.
UBS to $325 from $290 (Buy), Guggenheim to $283 from $277 (Buy). $42 of daylight between two Buys — that's the whole conversation in one number. UBS values it at 18x CY27 EV/FCF and calls it durable low-teens organic growth. Guggenheim sees the same quarter and says "show me the ACV recovery first."
Adjusted New ACV -40% YoY after +15% growth in Q1. Tougher comp, yes — but that's a trajectory that usually precedes a guide-down, not a raise. ADSK guided FY27 organic CC billings growth to 9-10% from 8-9% (ex-MaintainX, ex-model change). The $135M revenue guide raise at the midpoint breaks down as ~$60M MaintainX, ~$36M Q2 over-performance, and only ~$40M of actual demand improvement. The $70M billings raise? Guggenheim says 100% MaintainX.
"Absent dilution from MaintainX and related transaction and financing costs, Autodesk likely would have raised its fiscal 2027 EBIT margin outlook by approximately 25 basis points and increased rather than reduced its fiscal 2027 free cash flow guidance."
That's the bull case in one blockquote. Underlying ops are better than the headline guide. Margins inflect in FY28, 41% FY29 target intact. UBS is willing to look through the M&A optics. Guggenheim, at $283, is not yet.
Bull: Model transition is the distortion, not the business. Adjust for it, revenue and billings both grew ~12% CC. 92% gross margin, perfect Piotroski score, low-teens grower with FCF inflection ahead at 18x EV/FCF. Buy the mixed tape.
Bear: New ACV -40% is a demand signal, not a transition artifact. The "improved demand" embedded in guidance is $40M — noise. Billings raise is acquired, not earned. If the model change was truly neutral, why is the ACV cliff this steep?
Verdict: Both shops are Buy-rated, but the PT spread says it all. At $270.58 with a $57.1B cap, you're paying for the bull case. UBS's $325 works if you trust the transition math. Guggenheim's $283 is the price of proof. Position size accordingly.
VERDICT: THE MARKET IS PRICING SNPS AS AN AI VICTIM. THE NUMBERS SAY IT'S AN AI BENEFICIARY. Benchmark out with Buy and $570 PT after fiscal 2H beat-and-raise — and the stock still sold off after hours. That gap between the print and the tape is the whole story.
Non-GAAP EPS $3.91 on revenue $2.48B vs consensus $3.67/$2.44B — REAL BEAT, AND THEY RAISED FULL-YEAR. LTM REVENUE GROWTH 39.5%, GROSS MARGIN 82.6%. This is not a melting ice cube. Benchmark raised near-term estimates by a "material amount."
Here's the kicker: STOCK ENTERED THE PRINT AT 24x FORWARD EARNINGS — LOWEST MULTIPLE IN YEARS — AND STILL GOT SOLD. The AI-disruption narrative has been the overhang on all EDA names, and the tape is telling you the multiple compression isn't about the quarter.
"AI is serving as a tailwind to EDA and simulation software rather than a headwind." — Benchmark
That line is the entire bull/bear debate in one sentence. Benchmark thinks AI is pulling design complexity forward. The market has been acting like AI will replace the tools altogether. The September 30 analyst day is the clearing event — show the AI pipeline, and this thing re-rates fast. Waffle, and sub-20x is in play.
Baird upgraded to Outperform with a $560 PT (only +$2 from prior, but the rating change is the signal), calling for double-digit organic revenue growth by FY27. Same thesis: AI disruption is a tailwind, not a headwind.
Bottom line: Two analysts, one narrative — the worst case is already in the multiple. The r/r is asymmetric into the analyst day. If they map AI contribution to revenue, 24x forward looks silly in hindsight.
Citizens says the Q4 mess doesn't matter — the transition math does. Reiterates Market Outperform and $80 PT (stock at $40.53, down 47% from highs — that's nearly a double from here).
Headline print was horrid: revenue $137.2M vs $157.1M consensus, plus a $684M net loss from non-cash impairments tied to the bitcoin exit. But look through it. AI Cloud revenue hit $70.5M, more than double the prior quarter's $33.6M, now 51.4% of total. BTC mining shrank to $66.7M as capacity flips to GPUs.
The forward book does the talking: ARR ~$500M today, rising to $1B once Microsoft accepts Horizon 1, and >$4B ALREADY CONTRACTED for the December quarter. They've raised ~$19B over the past year via customer prepayments and converts — customers are pre-paying to lock capacity, and the market is rewarding the land grab (revenue +105% TTM to $757M, stock +76% past year).
"Near-term reported results may remain uneven as additional capacity is commissioned."
That's the polite way of saying tread water through the build-out. $80 PT implies ~5.5x FY28 EV/EBITDA — reasonable if the contracted ARR converts to cash revenue. Bears will hammer the dilution and the headline miss. But this is a funded, contracted AI-infra story, not a hopium narrative. The risk is AI capex wobble, not execution. Stay long the transition.
Evercore ISI raises PT to $355 from $315.16, keeps Outperform. The target matters less than the survey proof point: 57% OF ALEXA AI USERS BOUGHT A PRODUCT THEY WEREN'T PREVIOUSLY AWARE OF — first hard evidence Agentic AI is additive for Amazon Retail, not just a slicker search box.
"57% of Alexa AI users bought a product they were not previously aware of, marking the first survey evidence that Agentic AI is additive for Amazon Retail."
Penetration dipped to 92% from last year's record 95% — still 34pts ahead of Walmart. Same-day delivery usage rebounded to 49%, and Prime same-day users spend 3.1x non-Prime, the widest gap Evercore has recorded. Perishable Checkout works: 46% of respondents who've seen it added fresh groceries to their carts.
Bull case is clean. AMZN trades at 20.6x P/E with a 0.23 PEG — market's pricing it like a utility while the survey says AI discovery drives incremental spend. The $355 target sits at 30x 2028 EPS. Rosenblatt initiates Buy at $335; Citizens keeps Mkt Outperform on logistics strength + AI model growth. 22 analysts have revised EPS up; consensus implies 27% upside. Bear argument is thinner: penetration rolling over from a record high, and AI-driven discovery could be novelty spend that fades. But the 3.1x same-day spend gap argues spending intensity is compounding, not decaying. Side noise: AWS picks up DuckLabs (DuckDB) for faster/cheaper analytics; four Swedish wind farms add ~200MW. Incremental, not thesis-changing.
Wolfe reiterates Outperform and raises PT to $120 from $95. Not a massive gap over the current $107 print (stock's up 106% in six months, basically parked at the $107.91 high) — but this is Wolfe arguing the tape is wrong, not chasing it.
The fight: the post-print dip on the "Cloud NNARR expectations miss" doesn't hold water. Wolfe says the market is watching the wrong metrics — SUBSCRIPTION ARR AND NNARR BOTH ACCELERATED, and those are the numbers that matter now that the Cloud migration is largely done. Hardware and memory price spikes? No impact on H1, and Wolfe sees no reason that changes.
Cloud NNARR expectations miss does not warrant the stock's decline.
The math backs the bull case: LTM revenue growth 46%, gross margins 80.6%, and growth ex-migrations still ~20% against a 5pt tougher comp. EPS $0.20 vs $0.04 consensus — tenth straight beat. Wolfe models Subscription ARR growth of 31%+ this year, settling into durable mid-to-high 20s. Identity and RAC are the next upside levers.
Call it what it is: RBRK is an AI + cyber risk compounder, and Wolfe says it's one of the best security names on that trade.
Settlement overhang gone — the hard part was never the money, it was the uncertainty. KeyBanc keeps Overweight/$780 (stock at $578.77) with the math on their side: TEENS ARE LESS THAN 1% OF REVENUE and gross margins run 81.75%. This is a narrative reset, not a P&L event.
The Street broadly agrees — UBS Buy/$715, Rosenblatt up to $886 ("significant win"), Truist trims to $763 but stays long. Collective read: the $12.7B guaranteed over 10 years is the price of admission to focus purely on the AI product cycle — business messaging, consumer AI, compute optionality.
The bear case gets its due from Needham's Laura Martin, who stays Hold:
"Meta's costly strategy diffusion" — ventures into scattered tech verticals while the settlement locks in a decade of regulatory oversight.
She's not wrong about the sprawl. But the market's telling you the overhang removal outweighs the strategy drift. 35% to KeyBanc's PT from here, and for once the path is clean.
MS flips top pick to Legrand — from Schneider. PT EUR160 from EUR155, Overweight maintained. The bull case in one line: the de-rating is a narrative problem, not a fundamentals problem.
LR +8% YTD vs ABB +35%, Schneider +27%. Down 11% in six months. Multiple compresses 22% vs ABB and 18% vs Schneider. But 2027 consensus op profit estimates are UP 9% YTD — matching Schneider, trailing ABB's +13% by a hair. The de-rating happened without earnings revisions. That's the setup.
Two bears hanging on the tape: data center growth questions as 800-volt DC solutions enter later this decade, and H1 EBIT margin slippage while comps expanded. MS says those concerns are priced.
"The underperformance has not been tied to earnings revisions."
Girtz acquisition ($80M rev, modular power integration) gives the data center story a plug. RBC to Outperform, Kepler to Buy with EUR170 PT — sentiment is already flipping. The question is whether 800V DC fears are just 2028 noise hitting a 2026 multiple.
TRUIST PUT A REAL NUMBER ON THE IRM BULL CASE THIS MORNING — RAISED PT TO $155 FROM $140 AND FLAGGED A PATH TO 13% REVENUE GROWTH IN 2027 VS 9% CONSENSUS. That's not an incremental tweak, that's a step-change in the growth narrative. Stock's already up 48.6% YTD, so this is the firm saying the easy money is behind us, but the hard money is ahead.
Truist sees Asset Lifecycle Management staying hot and Records Storage pricing firming — that combo gets you to 13% in 2027, well above the street's 9%. Even their conservative internal number is 10%. Q2 backed it up: revenue $2.03B, +19% y/y, $60M beat, record EBITDA and FCF, full-year guide raised. At 18x 2028 EV/EBITDA, you're inside the 17-19x peer range, so the multiple isn't broken — the growth assumption just needs to hold.
"The firm's analysis indicates a potential path to 13% year-over-year revenue growth in 2027, compared to the 9% consensus estimate."
48.6% YTD means a lot is already in the price. If ALM decelerates or Records Storage pricing doesn't firm up, you're looking at 10% growth on an 18x multiple — that's a de-rate, not a crash, but it's not the 20% upside either. No M&A fuel in the tank right now. Best idea for 2026, but this is a conviction hold, not a fresh entry.
William Blair opens WULF at Outperform with a $31 base-case fair value — roughly 92% upside from $16.49. The firm's stepping into the dip (stock still +75% over the last year, +44% YTD) and calling it favorable r/r. This isn't a bitcoin mining call anymore. It's a power play: WULF transformed into a leveraged provider of critical IT load for Anthropic, Google, and other hyperscalers.
Blair sees 54% FY26 revenue growth as the backlog converts into contracted revenue. That's the multiple expansion path.
The bears have ammo though. Q2 was horrid on the bottom line: adjusted EPS -$1.94 vs -$0.25 consensus. Revenue $44.77M grew 32% QoQ but still missed the $46.52M forecast. S&P affirmed BB- but cut the outlook to stable from positive — flagging financing and construction risk on that 20-year, $19B Anthropic Kentucky campus. Rosenblatt remains Buy ($30 PT) but acknowledges AI bubble anxiety and the debt load are what's capping the tape.
Binary-ish setup: execute the Kentucky build, backlog converts into a rerating; stumble, and the leverage punishes you. Watch the financing news flow and backlog disclosures — that's the near-term catalyst loop.
BofA keeps the boot on DOCU — reiterated Underperform, PT to $58 from $52 (stock at $63.80). That PT hike is pure mechanics: PEER MULTIPLE EXPANSION (9x CY27 EV/FCF, up from 8x). No change to the fundamental view. They still don't think the Q2 print gives you line of sight on IAM-driven growth. Core e-sign is healthy, retention improving — that's not the debate. The debate is whether IAM gets ARR growth sustainably above 10%. The guide sits at 8.5%.
"The most important signal that the new product cycle is working would be an increase of fiscal 2027 ARR growth guidance, currently at 8.5%."
That's the whole ballgame. BofA's bar for a positive reaction isn't the quarter — it's a guide raise. IAM adoption is early, the category is unproven, and AI budgets are a knife fight. Add an evolving GTM and pricing/packaging shakeout and you get a high execution bar. Meanwhile shares are +35% off Q1 and +42% over six months. You're paying for a reacceleration that isn't in the numbers yet. Citizens is the bull — Market Outperform, $86 PT (modest Q2 improvement). Needham sits on the sideline at Hold. The $58-to-$86 spread is the market's honest disagreement on whether IAM is narrative or product cycle.
BIG Q4 BEAT, IN-LINE FY27 GUIDE. Citizens holds Market Outperform / $25 PT after Aviat blew past every bogey — revenue $120.9M vs $110.1M consensus, EBITDA $11.9M vs $10.6M, EPS $0.64 vs $0.55, revenue +4.8% YoY. Clean sweep in a stock that was only +3% YTD against a +12% Russell 3000. The setup was right.
The catch: FY27 guidance validates the base case, not a re-rate. Revenue $455-470M (midpoint ~$462.5M vs $462.7M consensus), EBITDA $50-55M (midpoint $52.5M vs $52.6M consensus). No guide-up.
Aviat Networks issued fiscal 2027 revenue guidance of $455.0 million to $470.0 million, compared with consensus estimates of $462.7 million.
Bull case: EBITDA margin expands from ~10% to 11.4% at the midpoint — real operating leverage, and net income grows this year. Bear case: 32.5x P/E on 5% revenue growth with a base-case guide. r/r comes down to whether Q1 shows this beat was a new run-rate, not a one-off. Watch the margin line, not the topline.
Freedom Broker lifts Ooma to $24 from $20, keeps the Buy. Stock sits at $23.04 — so the raise reads as a rubber stamp, not a new idea. Q2 FY27 beat across the board: EPS 35c vs 33c est, revenue $83.2M vs $81.72M. RECORD ADJ EBITDA OF $27.4M on a richer mix and tight opex.
Two-engine tape. Ooma Business is accelerating on AirDial's copper POTS replacement run, with FluentStream and Phone.com (acquired late 2025) stacking acquisition fuel. The bigger structural shift: MyPhone reversed the long-standing residential subscriber decline, and management embedded residential growth in FY27 guidance for the first time. That's not a tweak — that's a model change.
"The results were driven by accelerating growth in the business segment and the successful product launches across both business and residential divisions."
At 75% YTD and 66% over six months, the market already got the memo. The $24 target leaves ~4% upside — this is a momentum setup now, not a value one. Bull case: POTS replacement compounds for years and Ooma AI monetization is unmodeled upside. Bear case: AirDial growth decelerates, MyPhone was a one-quarter pop, and a $619M market cap means the multiple snaps back hard on any miss. Next print decides the narrative.
Q4 beat means nothing. FY27 guide is the drainer. DOWN 47% YTD TO $345.60 and management just told you next year is a transition — price cuts in TurboTax DIY, Mailchimp stuck in neutral, new SBC accounting method muddying the optics. That's a recipe for continued multiple compression, not a dip-buy.
The Street is resetting, not defending. Freedom Broker cuts to $370 from $430 (Hold), Truist and Stifel both land at $300, Piper at $290. BMO still Outperform but even they concede moderating growth. Wolfe downgraded outright — tax revenue guidance and the shift toward lower-end customers is a red flag for the premium multiple.
Revenue did come in hot — $4.354B, +14%, 2.1% AHEAD. CREDIT KARMA +16.5%. But with 6 analysts marking FY27 estimates down, the forward curve is all that matters.
Freedom Broker noted the transitional nature of fiscal 2027 and temporary pressure on client monetization growth.
CLSA's upgrade is a cost story, not a growth story. Outperform from Hold, PT to $3.80 from $3.20 — and the kicker is a 20% INCREASE to FY26-27 adjusted net profit estimates on AI-driven cost savings and better BEC profitability. Stock sits at 3.2x forward P/E and 0.33x book. Market's pricing in structural decline; CLSA's saying the margin work changes that calculus.
Q2 was "in line," but the mix is the message. Product sales -9.4% — deliberate pruning of home/furnishing, beauty, appliances. Services +10.2%, led by luxury and outdoor. Gap China +22% revenue but still a 33M yuan loss. They're trading top-line for quality.
"CLSA raised its fiscal 2026-27 adjusted net profit estimates by 20% to reflect better BEC profitability and AI-driven cost savings."
The reported EPS "miss" ($0.42 vs $1.68) has to be a consensus quirk — a literal read implies a 0.4x P/E, which contradicts the upgrade entirely. Non-GAAP operating income +25% is the cleaner signal. They also raised the 2028 profit target, which is the real strategic tell.
Bull case: AI savings compound, services mix keeps improving, and you're paying distressed-asset multiples for a profitable company. Bear case: Gap China bleeds, product sales shrink, and the China discretionary tape is horrid. CLSA's 4x FY27 multiple isn't demanding — it's a bet that the cost cuts are structural, not a one-off. At 0.33x book, the downside is arguably already in the price.
BMO stays Outperform at $30 after the investor day, but the real story is AI as a TAM doubler, not a feature. WAY's LLM-agnostic approach is smart — pick the best model per use case instead of hitching the wagon to one vendor.
THE KICKER: existing clients are hitting up Waystar first to talk AI strategy. That's not a vendor relationship, that's a strategic perch. Management reconfirmed LT revenue growth and EBITDA margin targets — steady as she goes. Q2 already showed the model works: REVENUE +18% TO $320M, ADJ EBITDA +21.5% TO $137M, and they raised FY guidance. 69% GPM at 19% growth — that's the software math PMs want to see.
"Existing clients are approaching Waystar first to discuss AI strategies."
The $27-$44 PT range tells you the street hasn't converged. BMO at $30 is the conservative bookend; Mizuho at $42 is the AI bull. Truist actually trimmed PT to $33 from $38 — so the dispersion is real. Net: this is a name you own for the AI cross-sell option, not for multiple expansion. 3p.
Jefferies trims PT to $31 from $45, keeps Hold — and the stock at $33.37 already trades above that. Their view: 2026 is the trough year, but the recovery is slower and shallower than the Street's back-half numbers imply. Not a compelling long, not a great short either. That's the whole trade in one sentence.
The firm matches 2026 consensus but sits BELOW Street on revenue and EBITDA for 2027-28. Core problem is still the U.S. residential market — softness plus slow margin recovery means guidance comes in light. C&I and Europe are improving, batteries are selling, but residential remains THE earnings driver until proven otherwise. Gross margin at 21% is the tell; that's a structurally weak number, not a blip.
"Jefferies characterizes the shares as an unappealing short position but not a compelling long investment."
The SST business gets a ~$5/share SOTP value at 2% market share — real option, but don't hold your breath for a major disclosure.
Crossing the tape: UBS went the other way, upgrading to Buy with a $42 PT on the FCC inverter import ban — reads as a structural gift to SEDG as a domestic manufacturer. Mizuho cut to $38 from $71 after the Q3 guide missed, echoing the U.S. softness concern. Net: three analysts, three different conclusions, same bearish residential backdrop. The bull case hinges entirely on import restrictions + Europe, the bear case on the margin recovery never arriving. At $33, the market's pricing in exactly that coin flip.
TD Cowen bumps PT to $45 from $43, keeps Buy, and sticks the "top services pick" label on EXLS after meeting with the CFO and IR. Light article, but the signal is loud: demand messaging stays consistent, AI-led momentum is broad, and the fundamental story is outpacing the IT services pack.
Q2 already did the heavy lifting — EPS $0.59 vs $0.55 est, revenue $594.8M vs $574.03M, FY guide RAISED. Organic growth hit 16%, THE STRONGEST IN YEARS. Baird came along for the ride, upgrading to Outperform with a matching $45 PT (from $35). Now TD Cowen's meeting notes confirm the narrative holds.
The honest caveat: TD Cowen admits the shares will trade with AI fund flows. But that's the entry point — the fundamental outperformance is becoming more visible as AI/IP scales.
"Shares may largely move with AI fund flows, but the company's fundamental outperformance versus peers and IT services is becoming more appreciated as it benefits from scaling AI and intellectual property."
At $37.80, that's ~19% upside to the $45 PT cluster. Not a screaming bargain, but a quality compounder with real AI revenue in a tape full of vaporware. For PMs looking for services exposure, this is the name two independent shops now land on at $45.
AAPL — September 9 is a 2nm capacity event wearing a product-event costume. First foldable, iPhone 18 Pro, and A20 Pro all land the same day. Read-through is N2 allocation: Apple grabs the initial 2nm wafers, everyone else eats tight supply. Watch DDI/panel constraints if foldable demand surprises.
AMD — ROCm 10 is real, but the ATOM inference engine saga is the bigger signal. Word is 20+ full-time engineers and a Fellow's dissent are burning on a side project with zero customers. AMD's AI inference future rides on vLLM/SGLang compatibility, not a separate engine. Bulls get software progress; bears get execution dysfunction.
ASML — EUV is the long-dated bottleneck on AI compute. If $1 of fab capex prints $100 of token revenue, ASML and Zeiss should be cranking mirror production harder. No print in this window. The signal is structural pricing power, not quarterlys.
ASMPT — Wire bonding is now shortage-bound. ASMPT and K&S are the nearly irreplaceable duopoly, and customers lock capacity with upfront cash. That broadens the semicap upcycle beyond advanced packaging. Traditional packaging finally has pricing power.
AVGO — Custom silicon remains an AVGO/MRVL duopoly. A MRVL guide miss could be AVGO share if Google splits orders. The total ASIC pie expands into 2028, so demand is not the issue. Taiwan Q4 shipment data is the arbiter.
BABA — Alibaba's model strategy: free now, monetize later. Cloud is the value-capture layer. Open weights commoditized the Chinese model layer. The AI revenue path runs through BABA Cloud, not model API pricing.
COHR — LITE publicly attacked COHR's 200G EML yield on 6-inch InP. Yield is the choke point at 200G. If LITE is right, COHR loses the next-gen optics socket. No COHR-specific numbers in the source.
CRWV — CoreWeave converts power delivery risk into refinancing risk. Source expects AI DC rents to stabilize 26Q4, then moderate 20–30% as supply lands. That's a headwind to the passthrough pricing underpinning CRWV's leverage.
CXMT — Blowout print: 1H revenue RMB150.31B, +873.6% Y/Y; net profit RMB77.61B swung from a loss and landed 36% ABOVE THE TOP OF ITS OWN GUIDANCE. Gross margin 84%. HBM2E in production, HBM3 in trials — China DRAM now self-funds future capacity. Bear read: the big three's commodity DRAM pullback hands CXMT DDR5 share, and shortage rents fund the future price warrior. Export controls didn't stop CXMT becoming China's most valuable company.
DDOG — In the cyber organic/inorganic revenue breakdown, flagged not a pure cyber play. No specific print. The AI observability/security tailwind applies. Signal is group comp, not idiosyncratic.
DELL — Ugly preview. F2Q revenue $42.2B vs $44.5B guide mid. EPS $4.70 vs $4.84 consensus. AI server guide raise to ~$75B likely priced in already. NVL72 shipments cut to 17k FY27 / 20k FY28. Traditional servers +75%, CSG +4%. Enterprise/PC shows early cooling after a strong 1H26.
FAST — Quality-management signal, not an AI data point. Florness compounded steadily. No earnings or valuation update in the source.
FTNT — Cyber group repriced after CRWD's record ARR. No FTNT-specific print. AI security tailwind real, not name-specific.
GEV — Gas OE pricing debate: if GEV takes more price than Siemens Energy, the peak-earnings bear case breaks. No hard data. GEV remains an AI power buildout derivative.
GOOGL — Google is the arbiter in the MRVL custom-silicon debate. TPU/Axicom Q4 data will show the share split with AVGO. Google and Nvidia add ~45GW of compute next year. Capex visibility enormous.
GTLB — No data on ultimate upgrades yet. Open question: does AI-assisted coding actually expand seats? Not actionable until GitLab prints upgrade or retention data.
INTC — Commerce is embedded at Intel. $23B raised. Tariff regime likely directs US volume to Intel Foundry. 14A called one of Intel's most successful nodes. Policy-engineered foundry demand, not just process execution.
KLIC — Wire bonding capacity secured with upfront cash. K&S and ASMPT nearly irreplaceable. Traditional packaging now shortage-bound. KLIC has durable pricing power.
LITE — OCS shipments doubled q/q; next quarter should break $100M. 200G EML has almost no competition except AVGO. NPO timing pulled forward to late 2027/early 2028. Net beneficiary of US action on Chinese transceivers — though China still holds 70% of the US market. And the public challenge to COHR's 200G EML yield is a share-shift signal in the highest-speed socket.
LRCX — Expect a lot of etcher sales over the next 12 months as memory and advanced-packaging capex inflect. WFE imports across Korea/Taiwan/China +14% YTD. No print. Directional semicap breadth.
MSFT — Grok is now in Microsoft Foundry. Broadens xAI distribution, adds token demand. Co-signed the global AI cyber-defense letter. No financial print. Signal is AI distribution share and security posture.
MU — Taiwan union strike support at 80%. Taiwan is ~60% of Micron global capacity and anchors advanced HBM output. Mediation runs before mid-September, strike vote if it fails. A strike would tighten supply — pricing call option, narrative bear put. Bull case: shortage call extends to 2030, MU at ~5x with HBM allocation precommitments. But CXMT's blowout shows China capturing shortage rents to fund future capacity.
NBIS — Signed at 3–4x its installed-base unit price within 18 months, runs on customer prepayments. Shows demand urgency but also high future unit costs. Prepayments are a moat in tight credit.
NOW — Added to the cyber breakdown, flagged not a pure play. No specific print. AI workflow and agentic security tailwind. Comp-driven signal.
ORCL — Named as a recipient of Vera CPU racks. Co-signed the AI cyber-defense letter. No Oracle-specific print. Still sits at the center of enterprise AI infrastructure.
PANW — CRWD's print drove cyber higher. PANW noted at >100x fwd P/E in one comparison. No specific print. Multiple leaves zero room for execution error.
PYPL — -15% PREMARKET after Advent and Stripe abandoned acquisition plans. Press reports built the premium, never a confirmed process. If acquirers don't publicly deny, $52.50 might be near a cleaned-out premium. Low confidence without confirmation.
QCOM — HBC could capture up to 20% of the data center accelerator market. Negative sentiment means positive surprise potential is large. Mixing HBC with edge/smartphone DRAM stacking muddies the datacenter message.
SKHY — CEO calls the memory shortage structural to end-2030; oversupply risk low because memory is no longer just commodities. $4B Indiana HBM packaging plant targets 2029 production. Bullish for the complex, but positioning is one-sided long. Bear wart: NAND node lag — 176-layer still >50% of production, revenue share slid 21.4%→16.4%, YMTC nearly caught up. HBM strength doesn't fix the NAND margin drag.
SMEGF — Comparison leg in the gas OE pricing debate. If it takes less price than GEV, the "gas OE is over-earning" bear case loses its own logic. No company-specific data in the source.
SNDK — Sees structural massive NAND demand through 2030, non-GAAP gross margin around 80%. Memory at ~5x while software trades at 150x. Allocation precommitments may alter commodity-like dynamics.
SNOW — After the 100% rally off lows, screens above 150x fwd earnings. R/R flipped unattractive — software went from dirt cheap to uninvestable. A better LLM handling complex workflows could re-trigger disruption fears.
SSNLF — Early lead in Nvidia's NVHBM. 8-high HBM4E at 17–18Gbps trades stack height for speed. Integrated memory plus logic base-die gives Samsung an edge in custom HBM. NVHBM could show up in Rubin Ultra at 576-GPU scale.
SIVE — Record $1.2B pipeline. But pipeline is not design win. Jabil beta Q4, first production orders 1H27, ramp 2H27. Down 75% from the pink-candle high — potential bullflag. High optionality, high execution risk.
TCEHY — Hunyuan Hy4 open-weight preview: 770B total / 49B active, 1M context, Apache 2.0. Internal eval claims are narrative noise. Real signal: China's model layer is free and open, value accrues to distribution. That's Tencent's edge.
TSM — Moody's confirmed Aa3, outlook positive. Foundry share 73%. Apple M6 is the first 2nm processor. Higher rating lowers financing cost for the 2027 N2 capacity war. Q2 employee profit share NT$36B, +50% Y/Y — labor grabbing AI rent.
TTWO — GTA 6 first gameplay footage dropped, chosen for Netflix release. Stock up on launch-window certainty, not fundamental sales data. And it competes with Apple's September 9 event for the same autumn consumer attention.
UMC — 22nm fully utilized for Samsung's Fold 8 DDI. Samsung asked for a super hot run, UMC said difficult. Mature-node pricing power at maximum utilization. DDI supply remains a chokepoint.
ZS — Cyber group repriced after CRWD's record ARR. No Zscaler-specific data. AI security tailwind real. Group beta.
Custom silicon