Good morning.
Futures are GREEN into a tape where the 10Y is printing fresh multi-year highs, Bunds have broken, Brent is through $101, and Trump just priced a $1.3T election-year giveaway at $5,000 a head — so the whole setup in one line: the bond market is repricing fiscal discipline while the AI numerator refuses to blink. TSM AUG REVENUE +53.3% Y/Y, FOURTH STRAIGHT RECORD and first month ever above NT$500B — with 1.4nm mass production possibly pulled into end-2027, the supply side is now saying exactly what the demand side has been saying. That's why equities can rally through a rates break: this isn't a beta tape, it's an earnings tape with a discount-rate problem stapled to it.
The headline print is ORCL tonight, and the bar is higher than the beat — RPO composition (OpenAI share and recognition timing), FY27 capex guide, and any actually verifiable framework for covering debt service with AI returns. Beat-and-raise probably doesn't re-rate at 4.7% on the 10Y; vague "trust us" gets sold. This is the first listed-company stress test of the borrow-to-build-AI model, and it precedes the SB Energy IPO — $430B of contracted datacenter, ZERO operating sites — which is the credit market's own temperature check two weeks out.
Asia's the tell again. PSMC (力积电) back above NT$70B for the first time in four years, +76% Y/Y, company blaming DRAM supply/demand and rising memory prices — a mature-node foundry dragged out of the ditch by a memory shortage is the cleanest evidence yet that this cycle is bleeding into second-order names. Murata cutting MLCC lines to free capacity for AI, Huawei lifting Ascend 950DT quotes +20-50% on HBM scarcity — the pricing-power chain is now closed from Samsung/Hynix to Apple to AT&T to Huawei, and four unrelated buyers eating full pass-through is not a cycle, it's a tax.
The bear case lives in the same window: DeepSeek V4.1 Flash cuts KV cache HBM needs to 1/4 and SSD to 1/8, and Kioxia's CEO is publicly calling the memory rally overdone. Architecture-as-HBM-substitute is the one variable that could bend the super-cycle, and it just got a real datapoint — worth pricing, not yet worth trading against the evidence chain.
Semicap advanced packaging is the quiet winner: KLA's adv-packaging inspection growing 70% Y/Y at roughly 2x market, with FORM and AEHR in the same lane. Our feed flags ASML bearish, so the dispersion inside semicap is the trade, not the sector. Also worth a flag: DOJ opening a probe into the NVDA-Groq structure — short-term noise, but it sets the tone for every future deal the ecosystem needs.
Apple's China Telecom preorder slots filled in 2 MINUTES and Far EasTone bookings doubled Y/Y — but telecom reservations are a zero-cost queue, not conversion, and we've seen "sold out then soft" too many times. The $100 price hike not killing demand is the first layer of proof; 9/18 first sale and late-Oct Duo weekend are the real reads.
We'll hit up ORCL, TSM and the memory complex first, then get to semicap (KLAC/FORM/AEHR), the neocloud/power names, and the software prints.
Verdict: the numbers beat, the guide beat, and the stock still got taken out back — down ~20% to the mid-$20s on a 3% revenue beat and a 26% op income beat. That's a positioning flush plus a dead narrative, not a broken business. But make no mistake: the "accelerating organic growth" story that got this thing 51% in six months is now on life support, and it needs Forge at the end of the month to revive it. No analyst downgraded. Nobody on the street is below $30.
No rating changes, all PTs up or held. Named targets sit $30-50, median ~$35 — the $32-38 band is where the cluster lives, Piper is the low at $30, Needham is the outlier at $50 and has been for a while.
Raises were modest and mechanical: Raymond James $27→$33, Stephens $31→$34, Piper $27→$30. Cantor and Goldman sit at $38; Oppenheimer $36; Stifel and BTIG $35; Mizuho $32. Zero downgrades on a -20% tape move. That tells you the sell-side thinks the flush is technical, not thesis-breaking — and also that they were all late marking to a stock that ran 51% in six months.
Bull: The beat-and-raise was real, and it was operational — op income beat by 26%, gross margin expanded 66.5% while services revenue scaled up, which kills the "services is a low-quality mix problem" argument. DBNRR inflected +200bps Y/Y on better renewal cohorts, which is the leading edge of a reacceleration, and cRPO at 24% still beats every comp. Paid AI attach jumped 9pts, Operator engagement is early but real, and the legacy marketing cloud displacement/consolidation motion (CRM, ADBE) is intact. At ~2.5x CY28 revenue vs a ~4x group average, RJ calls it attractive r/r for a premium grower.
Bear: The acceleration narrative is dead — Oppenheimer's words — and organic growth decelerated 250bps sequentially for the first time in several quarters. cRPO missed, backlog decelerated, and the near-term revenue and profit guide came down after lapping OfferFit and as services capacity builds for Decisioning Studio. The FY28 raise was mostly services assumptions, not subscription. And there's a live macro risk: AI spend crowding out 2H software budgets. Add an open CFO seat and the fact that the implied Q/Q growth is roughly flat to up 1%, and you have a stock that was priced for acceleration and just got told to wait.
New: Paid AI attach +9pts. DBNRR 110%/+200bps Y/Y. Services mix explicitly climbing, with most of the FY28 outlook raise tied to services assumptions (that's the tell the bears grabbed). Forge conference confirmed for later this month. CFO search still open. Intl decelerated to 25% from 29%.
Known: Legacy displacement and vendor consolidation. The <3-year path to 20% operating margins (Goldman). Low-to-mid-20s growth. The AI-native front-office positioning.
The market treated the services mix shift as a quality-of-revenue problem. Oppenheimer steelmanned the other side:
"The AI ramp does not appear to be structurally margin-dilutive for the company." — Oppenheimer
"Organic growth reached 24.4% in the quarter, though it did not accelerate for the first time in several quarters." — BTIG, Matt VanVliet
"Fundamental growth drivers remain intact, including legacy displacements and consolidation activity." — Stifel
"Shares offer attractive risk/reward at approximately 2.5 times its calendar year 2028 revenue estimate, compared to the group average of approximately 4 times." — Raymond James
Forge, late September. Both Oppenheimer and BTIG point at it as the potential reacceleration catalyst — Oppenheimer explicitly says growth could reaccelerate after the customer conference. That's the single most important date on the BRZE calendar.
Services mix in FQ3, and cRPO growth. If cRPO reaccelerates back above 25%, the deceleration scare was a one-quarter comp issue. If it doesn't, the bear case has legs.
CFO hire. Open seat at a company about to guide through an AI investment cycle is an overhang.
The big question this print raises is whether front-office AI is a budget expander or cannibalizer for martech. Oppenheimer's crowding-out concern is the same worry that's been pressing on CRM, ADBE (Experience Cloud) and HUBS — AI dollars coming out of the same CX software line items. BRZE's answer is the +9pt paid AI attach, which argues attach is additive rather than cannibalistic. Watch KVYO and TWLO next. And note the services build — implementation capacity for Decisioning Studio is a real tell on enterprise deal complexity, which reads across to any CX vendor selling AI workflow into large orgs.
The trade: mid-$20s, ~$3.43B cap, ~2.5x CY28 revenue vs 4x group. Street's already there, nobody's capitulating, and the catalyst is three weeks out. This is a show-me on Forge, not a fundamentals short.
Bulls and bears read the same event and land on opposite signs. Apple priced the new lineup for units, not for margin — and at 36x forward with memory costs running, that trade only pays if the Duo actually ships in volume. Right now nobody can build enough of them to find out.
A barbell, and an unusually wide one for a $4.6T mega cap. Bulls cluster at $365-370 (Evercore ISI $365, HSBC $366, Melius $370, BofA cut to $370 from $380). Bears sit at $250-263.66 (Lynx $250, KeyBanc $250, Jefferies $263.66). Oppenheimer slots in at Perform — no PT, no conviction, which for a name this well-covered is its own statement. Stock at $315.34 lands below the entire bull cluster and way above the bear floor.
The consensus that matters isn't the PT — it's the revision direction. 13 ANALYSTS HAVE CUT FORWARD EPS ESTIMATES. BofA took CY27 to $9.98 from $10.32 (down 3.3%) but held the 37x multiple flat, so the PT cut was pure earnings, not de-rating. That's the honest way to trim. The bears are the ones who think the multiple is the problem.
Bull (Melius, Evercore, HSBC, BofA): The Duo is a genuine new form factor, not a spec bump. Melius modeled 15M DUO UNITS IN FY27 with mid-teens production builds, and backs it into iPhone revenue growth of 20% over two years versus 9% consensus for FY27. Price increases are real even where they came in light — Pro/Pro Max +$100 TO $1,199/$1,299, higher storage tiers +$300-500. Evercore points to an upgrade pool of devices three-plus years old, plus the A20 Pro (40% better sustained performance) and a 48MP variable-aperture camera as the spec-forced replacement driver. Ternus deliberately set a low near-term bar.
"Apple could sell 15 million Duo units in fiscal year 2027 if supply allows, with mid-teens production builds appearing possible." — Melius Research
Bear (Jefferies, KeyBanc, Lynx, Oppenheimer-lean): Volume over margin is a bad trade at this multiple. The 256GB and 512GB iPhone 18 Pro/Pro Max came in 6-8% below Jefferies' estimates — the price increases are 7-9%, not the double-digit uplift the model needed. Add inflating memory and flash costs against a 48.65% gross margin and you get exactly the BofA move: more units, less EPS. Layered on top: Siri AI is still English-only beta and unavailable in the EU (Oppenheimer), and App Store revenue printed $6.3B, +0.1% Y/Y in fiscal Q4 — the Services anchor of the multiple is effectively flat.
"Apple appears to be prioritizing volume over margin." — Jefferies
"Lower-than-expected pricing on new iPhones... could drive higher unit sales but pressure gross margins due to higher memory and component costs." — BofA Securities
Two things. First, the foldable is real and priced at the top of the category — $1,999 base with a 256GB variant added. Second, and more important: SUPPLY IS THE SWING FACTOR AND THE TWO SIDES ARE ALREADY DISAGREEING ON IT. Oppenheimer pegs foldable supply at 8-10M UNITS THIS YEAR on capacity and yield constraints. Melius says 15M in FY27 if supply allows. That 5-7M unit gap is the entire bull case, and it's a manufacturing question, not a demand question.
Mac share gains, AirPods 5, Watch Series 12/Ultra 4, the Health app overhaul (60x heart rate data frequency), on-device AI with Private Cloud Compute. Real product, no numbers attached, don't underwrite it. Folding these into a thesis is how you end up long a story instead of a print.
Oppenheimer flags CEVA, Corning (GLW), and Universal Display (OLED) as direct beneficiaries — Ceramic Shield content on the Duo plus larger display area. GLW already ripped +127% OVER THE PAST YEAR to $168.46, which is a lot of foldable optionality priced in before a single unit shipped. Counterweight: the cancelled all-glass iPhone (originally 2027) took a Jefferies downgrade on Apple and is a cover-glass negative for the same complex. And the memory/flash tightness Lynx calls out — the thing that "might take a year for the new CEO to address" — read through to the entire handset BOM, not just Apple.
Net: this is a margin-debate stock into FY27, not a demand-debate stock. Both sides concede the units. Trade the gross margin line, and respect that 36x leaves no room for the bears to be right about supply.
Verdict: the quarter was better than the stock's reaction. Beat on revenue, beat on GBV, beat on margin, raised FY guide — and it still traded down 15% after hours. That's a positioning flush, not a fundamental miss.
Revenue $233M, +35% Y/Y, vs ~$220M consensus. Non-GAAP EPS $0.05 vs $0.04. Operating margin 7.4% vs 6.3% consensus — that's the line that matters, because the bear case has always been "great top line, no operating leverage." Gross booking value $3.0B, +45% Y/Y, vs $2.85B consensus (Rosenblatt had it 7.5% ahead of their own number). Platform usage revenue ~90% of total, +35%. New GBV signings $4B TTM, +60% Y/Y. 50 S&P 500 logos, up from 45. RFP activity +200% Y/Y in H1. Gross margin still 72-73%.
Q3 revenue $253-255M, ~+30% Y/Y, above the ~$248M bogey. Non-GAAP op income $35.5-36.5M. FY revenue $927-933M (+32% midpoint) vs $911M consensus, and op income $82-86M vs $78.6M consensus — raised from a prior $76-80M. Includes ~$2M from BoomPop.
So: guided up on both lines, above Street on both lines, and it sells off 15%. Stock was +59% YTD vs Russell 3000 +11% going in. That's your answer.
Consolidation: Rosenblatt to $29 (from $27), Buy. Needham reiterates $30, Buy. Citizens reiterates $38, Market Outperform. Note the dispersion — two of three cluster around $29-30, Citizens is a $38 outlier. The $29-30 cluster is doing the honest work here: Rosenblatt's build is 32.8% revenue growth in FY27, 25% in FY28, with the PT anchored on FY28-29 estimates. That's a "multiple stays compressed until growth decelerates gracefully" PT, not a re-rating call.
Needham flags the nuance that actually explains the tape — market expectations ran hot after last quarter's World Cup bump and a strong early public-company print. They also call out rising S&M spend (with a commission amortization quirk making the S&M line lumpy) as the reason profitability optics wobble. Their key structural call:
Navan can sustain growth above 30% even excluding recent strength in its Payments and Expense segments.
Read that again. If travel-only growth holds >30%, the Payments/Expense attach is upside optionality, not the crutch.
Bull: Durable 30%+ grower with 73% gross margins, real enterprise penetration (50 S&P 500s, RFP activity tripling), and operating leverage starting to show up in the guide — 7.4% op margin now, $82-86M FY op income. BoomPop adds Meetings & Events TAM on the cheap. If the growth is travel-core and not Payments-driven, the multiple has room.
Bear: Everything is priced for perfection, and the selloff says PMs agree. World Cup comps flatter the next couple of prints, S&M is stepping up and will eat margin, and the commission amortization treatment makes the profitability trend harder to trust quarter to quarter. Rosenblatt's own framework — 25% growth by FY28 — means deceleration is in the model, and a 30% grower that decelerates gets de-rated regardless of the beat-and-raise cadence.
Clean the after-hours move out and this is a beat-and-raise with an FY op income guide going up 8-10% at the midpoint. The 15% drawdown is a de-grossing event on a crowded momentum long. Not sure we can read too much into it either way until we see where it opens cash — but if the fundamentals hold, this is where the r/r improves, not deteriorates. Watch whether the $29-30 cluster migrates up or Citizens walks back toward the pack.
JPMorgan just flipped Neutral → Overweight with a PT to $820 from $640, and the reason matters more than the action: they're underwriting frontier models and non-ad products, not another ad-cycle beat. That's the first real sell-side capitulation on the "Meta is an AI loser" narrative. Evercore ($860), KeyBanc ($780), and Bernstein ($800) were already there. Street PT range $580–$1,000, consensus Strong Buy. The stock is UP 20% OFF RECENT LOWS but still DOWN 1% YTD against the S&P's +12% — that gap is the whole trade. PMs haven't paid for this story yet.
Muse, the consumer AI agent, hit #3 IN THE U.S. APP STORE ON DAY TWO with early usage running 10X TRAINING COHORTS. That's a real demand signal, not a press release. It ships alongside Muse Spark 1.3 (Zuckerberg flagged step-function gains in coding and agentic work), now live via Muse Code and the Meta Model API, with open-weights versions coming.
The TAM math is what Evercore hung its call on — 3.6B DAU, ~15M SMBs already transacting commercially across FB/IG/WhatsApp, and a >50% probability of successful consumer + SMB rollout. Note the sequencing: JPM says monetization is explicitly not the near-term priority for Muse. The near-term monetization case is still ad share.
On ads, Bernstein's line is the one to remember: Meta is on track to surpass Google Search in advertising revenue by end of 2026, and captured nearly half of every incremental digital ad dollar in Q2 2026. Gross margins 81.75%, revenue growth 27.65% — the core engine is not the debate.
Bull: You get frontier-model credibility (JPM says the summer-2025 goal of frontier models within a year is "essentially achieved"), a distribution channel no other AI lab can replicate — 3.6B daily users is the moat, not the model — and a core ad business compounding mid-20s while funding the whole thing internally.
Bear: The capex line is genuinely frightening. JPM models 2027 CAPEX OF $243B (+70%) and 2028 at $284B (+17%), BOTH WELL ABOVE CONSENSUS, with FCF of NEGATIVE $65–70B IN EACH YEAR against $41B LTM levered FCF. That's a company voluntarily torching its free cash flow profile for a product with no disclosed monetization. And a key AI researcher (Andrew Tulloch) walking right after the launch isn't nothing — though not sure we can read too much into one departure.
The honest bull rebuttal: Meta has said it will max compute in 2026–27 and retain flexibility from 2028, so the $284B number is a policy choice, not a contract. If the model layer commoditizes or Meta can't convert agents into take-rate/subscription revenue, the multiple compresses on a much worse FCF story.
JPMorgan said monetization is not the near-term priority for Muse but noted opportunities across a take-rate model and subscriptions.
That's the tell and the risk in one sentence — nobody is modeling agent revenue yet, so the upgrade is a re-rate on narrative + optionality, not on numbers. Watch two things: App Store rank durability past week two, and any commentary on capex discipline at the next print. If capex guidance walks up again with no monetization framework attached, the FCF bears get the tape.
Every analyst on the tape has a Buy and a PT 35-120% above spot — and the stock is still down 50% YTD. That gap is the trade. Either the AI story inflects revenue in 2H, or the multiple keeps grinding lower while the sell-side defends.
KVYO at $16.27, DOWN ~18% IN A WEEK and nearly HALF YTD. The K:BOS customer conference in Boston was the week's catalyst, and the analyst read-through was uniformly positive — Needham, Stifel, Citi all reiterated Buy, TD Cowen trimmed a buck but stayed long. The problem: PTs sit in a $22-36 cluster and the stock keeps making lows. When the street is this uniformly bullish into a 50% drawdown, the market is telling you it doesn't believe the guide.
Q2 was fine on the headline — REVENUE $370.6M, +26% Y/Y, beat the $362.1M bogey, ADJ EPS $0.19 in line. But TD Cowen flagged sequential growth as light vs. the prior quarter, and the 2H margin outlook is the real culprit. Gross margin 73.8% and net cash on the balance sheet (current ratio 3.13) — the foundation isn't the issue. The issue is what happens to opex as they push upmarket.
The headline product is the Klaviyo Data Platform (KDP) — positioning KVYO as the context layer and system of record for B2C retail data, not just an email/SMS sender. Stifel calls the context layer, data capabilities, and activation channels defensible. Agentic roadmap got expanded: updates to Composer and Customer Agent, 260+ MCP tools accessible from AI interfaces, plus Klaviyo SQL and a Headless strategy aimed squarely at larger enterprise accounts. Composer is at roughly 138K users.
Needham's line on why this matters against the model providers:
"The AI agent has a strong use case compared to native AI tools like Claude and GPT."
That's the whole bull case in one sentence — proprietary first-party commerce data beats a general-purpose LLM when the job is "run my campaign and don't burn my list." Needham also flagged new pricing aimed at reducing adoption friction, which is the right move if the goal is seat/usage expansion over ARPU near term.
Bull: KDP makes KVYO the data layer under B2C marketing, which is a stickier and more defensible position than point-solution email — 260+ MCP tools, agentic roadmap, and enterprise/Headless motion open a TAM well above the SMB base. Net cash, 73.8% gross margin, +26% growth, and a stock at half its YTD starting point is a r/r skew most software names don't offer.
Bear: Growth is decelerating into a softer 2H with margin pressure, and the AI features that are supposed to reaccelerate the model are exactly the features the LLM platforms are shipping natively for free. A $22 low-end PT vs. a $36 high-end doesn't reflect disagreement about execution — it reflects zero consensus on terminal multiple, and that's what a de-rating looks like.
Needham $30, TD Cowen $26 (from $27), Stifel $22, Citi $36 — the dispersion is the story, not the average. Conference vibes are great; conference vibes don't print. Watch 3Q for Composer attach and any enterprise logo disclosure, and watch opex. If margins stabilize and net-new enterprise lands, this re-rates. If not, the Buy ratings are just a slower way of being wrong.
Clean ARR beat, cosmetic revenue miss, and two brokers splitting $22 vs $19 — the entire debate sits on how much AI monetization you're willing to underwrite. SAIL is a 25% grower trading at a "prove it" multiple. The quarter did nothing to resolve that; it just moved the goalposts from "can they migrate to SaaS" to "can they monetize agents."
ARR $1.231B, +25% Y/Y, beat the midpoint by $11M. SaaS ARR $847M, +36% — that's 97% OF NET NEW ARR, well above the guide range. Total revenue $309M, +17%, and that's the headline miss: below the midpoint because the SaaS/term mix skewed harder than modeled. Term revenue gets recognized upfront, SaaS gets ratably, so the mix shift was a $5M revenue headwind. That's a timing artifact, not a demand problem — but it's why the top line disappointed while the ARR line beat.
The tell: NRR MAINTAINED AT 113%. Not up. Gross retention high-90s. So AI is 30%+ of net new ARR and >$70M of total ARR, yet the installed base isn't expanding any faster than it was. Read that two ways — AI is landing as new logos and add-on land, or it's partially displacing existing SKU spend. Not sure we can read too much into one quarter, but the flat NRR against a hot AI narrative is the thing to watch.
Adj op margin 20.3%, FCF $37M, GP margin 66%. That 66% is the legacy/on-prem hangover — Mizuho's point that the SaaS transition unlocks better economics is the second-order story here. GM should grind up as the mix completes.
$70M in AI-driven ARR (Agentic Fabric, Agentic Suites, agentic add-ons for human and non-human identities) against a $100M FY27 target. Tracking ahead. Pipeline doubled in three months, with demand for product that isn't GA until FQ3.
The AI-driven pipeline has more than doubled over the past three months, with early demand for Agentic Fabric and Agentic Suites ahead of third fiscal quarter availability.
That's the whole bull case in two lines. 125M human identities managed is the distribution surface for non-human/agentic identity governance — the attach math is what gets you a $22 target instead of $19.
Bull: ARR beat, SaaS conversion is basically done at 97% of net new, AI ARR is a year ahead of plan, and identity governance is a land-and-expand suite with rising ARPU as agentic modules attach. Evercore's $22 implies 25% upside and treats the revenue miss as noise.
Bear: Revenue guide didn't move despite the ARR raise, NRR is stuck at 113%, and the AI number is still small enough that the incremental monetization is unproven.
The firm said it remains uncertain about the degree of cross-selling success over time and the magnitude of incremental AI monetization.
That's Mizuho staying Neutral — $19 from $18, ~8% upside. They like the platform, they just can't size the attach. Fair.
Bottom line: If you own the AI-in-identity thesis, this print supports it. If you need the P&L to show it, come back in two quarters. The guide is the sword of Damocles — ARR up, revenue flat = management either sandbagging or the term-to-SaaS drag runs longer than the Street models.
Piper Sandler starts AVGO at OVERWEIGHT, $460 PT — and the tell is the multiple, not the rating. They're modeling ~51% EPS CAGR through 2030 and still only paying 14x FY28. That's a "we believe the numbers, we don't believe the terminal multiple" initiation, and it's genuinely conservative next to the post-print street: BMO to $575, Truist trimmed to $520, DA Davidson cut to $350. A $225 spread on the same tape isn't a fundamentals fight — it's a duration-of-AI-capex fight.
The differentiated work here is per-gigawatt content, not the EPS bridge.
Broadcom's networking attach rate is running at 30%, enabling $20 billion to $30 billion in total content per gigawatt. The company has visibility for 12 gigawatts of demand in fiscal year 2027, with estimates of 23 to 30 gigawatts in fiscal years 2028-2029 and 38 gigawatts in fiscal year 2030.
Run that curve: 12 GW → 38 GW FY27–FY30 is a 3x on volume, and the ASIC + networking bundle is what makes it a 3x on content too. Piper has them at 75% ASIC SHARE for AI inference, with Google TPU the current volume shipper (Anthropic riding it), Meta MTIA and OpenAI's "Jalapeño" ramping, plus two unannounced customers. Custom silicon AND the fabric that hangs off it — that's the toll road.
Bull: 75% share in custom inference silicon with hyperscaler co-design lock-in, attach rate compounding, and a GW backlog that's already visible three years out. Even a 14x exit multiple gets you to $460.
Bear: every gigawatt in that FY27 number is someone else's roadmap. ASIC programs slip, OpenAI's in-house silicon is the least proven thing on the board, and the market is now paying for GW conversions that haven't been taped out. Watch the software line too — Truist flagged it as the soft spot in the print.
Piper Sandler opens at NEUTRAL with a $190 PT, and the tell isn't the rating — it's the framing. Piper's whole bull case is already priced: "new design wins are already reflected in the current share price," so the stock needs something incremental to work from here. Datacenter goes from a rounding error to ~26% of FY29 revenue, handsets fade to ~50% of EBIT, and he still only gets to a 16% 4-year EPS CAGR and ~$15 in FY29. That $190 tag implies roughly 12.7x his own FY29 number — below the 14x five-year average he uses as the anchor. Even the neutral case is discounted. Stock's been marked near $176 (20.5x on $8.59 TTM), so call it ~8% upside in a base case where handsets keep eroding.
The bar is now HBC. Piper's explicit tripwire:
The firm said it would need to see greater traction or industry acceptance of Qualcomm's HBC technology to adopt a more bullish stance on the stock.
That's the whole debate in one line. Late to datacenter, winning on custom silicon/accelerators/CPU servers through FY29, Amazon as a named customer doing customized inference silicon and optical connectivity — but every one of those wins is a 2027+ revenue event competing against NVDA's roadmap and hyperscaler in-housing. No surprise the street's all over the map: PTs run $165 to $270. StoneX Buy $270 (Amazon partnership), RBC $180 Sector Perform, Barclays $180 Underweight, Bernstein $165 Market Perform, Cantor cut to $165 from $200 on accelerator/CPU competition. That $105 spread tells you nobody has conviction on terminal datacenter share.
The trade: QCOM is cheap (20x with buybacks running) and the AI narrative is real, but it's a show-me story four years out with the multiple already paying for it. Long here is a bet on HBC getting industry acceptance — Piper just told you he won't pay for that yet. Not sure we can read too much into a fresh initiation's PT, but the "wins already in the price" line is worth respecting given the 31.67% six-month run. Wait for the print that shows datacenter revenue actually materializing, or for HBC to get a second hyperscaler nod.
Piper Sandler starts at Neutral, PT $110 — likes the story, won't pay for it. Stock's at $106.24, +328% over a year, +127% in six months. That's the whole problem in one line. Piper's doing the math on what's already in the tape and coming up with "fully valued."
Agentic AI is the tell here — it's driving server CPU demand where supply is genuinely tight, and Piper doesn't see supply/demand normalizing until 2029-30. That's a multi-year runway for DCG to print high-teens revenue CAGR to 2030, more than offsetting the PC memory noise that's been spooking the group. Foundry is the swing factor: 14A customer evals are coming in better than expected, and management gets credit for the culture/product/customer-trust rebuild under Lip-Bu Tan.
But here's the punchline — Piper thinks the stock already discounts ~15% foundry share gains, which is a full fab module (Fab-62) and 45% of the share price. You're paying for the win before it's booked.
"The firm initiated coverage at Neutral on valuation with a price target of $110."
Not standing still on the buy side of the ledger — Lip-Bu Tan and a family member plan to buy $12M in shares as part of a $20B common stock offering. Northland flipped to Outperform on turnaround progress and the server CPU shortage. Mizuho sits at $92 — below spot — on Agentic-AI demand and server refresh tailwinds. Citi's bullish semi call lifted the whole SOX, INTC included.
The read: sell-side is clustering around "story's real, price is ahead." Neutral-at-$110 with the stock at $106 is a polite way of saying the easy money's been made. The bull case needs 14A to convert evals into actual share — until then you're long a fab-module option that's already marked at 45% of the equity.
Piper Sandler came off the sidelines with an Overweight and a $270 PT, and the tell is in the math: 18x CY2030 EPS of $19, which assumes ~45% EPS CAGR off today's print. Verdict — the bull case here is no longer "AI datacenter attach," it's "Marvell is the second source on hyperscaler custom silicon and the tape hasn't priced the TPU attach ramp." The bear case is equally clean: you're paying 79x trailing for revenue that doesn't start landing until FY2029. Oct 6 analyst day is the catalyst that forces the market to pick a side.
The Piper thesis rests on MRVL's ~10% share of datacenter components, leading DSP and custom attach connectivity franchises, and the hyperscaler relationships (Celestial AI tuck-in included) that position them to take custom share into the next optical wave. The reported $120B Google agreement is the validation piece, with TPU attach programs kicking in fiscal 2029.
Piper Sandler framed the $120 billion Google agreement as "validation of Marvell's strategy and transformative for the company," driven by TPU attach programs starting in fiscal year 2029.
The PT cluster tells you where the Street's anchoring: KeyBanc $400, Oppenheimer $325, Needham $300, Benchmark $275, Piper $270. Cantor is the lone Neutral and the only one saying the quiet part out loud — valuation versus NVDA. Note the Q2 setup was solid, not spectacular: REV $2.739B, EPS $0.94, modest beat, strong DC-led Q3 guide, and 26 analysts have revised up for the coming period. That's the honest read — no air pocket, but no upside surprise either. The whole trade now hinges on whether PMs will underwrite a 2030 model in September. If the analyst day delivers a credible long-term framework, the Cantor valuation objection gets steamrolled. If it doesn't, MRVL stays a range-bound funding short for NVDA bulls.
Piper Sandler opens ARM at Overweight, $320 — and the bull case is quietly mutating under everyone's feet. It's no longer the smartphone royalty annuity story; it's ACCELERATOR IP. Firm's math: just 10% ASIC share could DOUBLE earnings. That's the whole trade, and it's why nobody can agree on a number.
Stock sits at $264, UP 119% IN SIX MONTHS, trading at 270x P/E. You're not paying for the base case (50% CPU IP share, ~20% CAGR to ~$12B by FY31) — you're paying for the silicon pivot. AGI CPU chip ramps at Meta and OpenAI from CY27, $15B revenue target by FY31. The multiple is a call option on that working.
Piper Sandler: "approximately 10% ASIC share could double the company's current earnings."
Ten-plus markets of price targets and they span $250 to $641 (that RJ number is a genuine outlier, not a typo — Raymond James sees server royalties eclipsing a THIRD of total revenue by FY28). Rosenblatt and Wells Fargo both trimmed on margins and multiple compression, $270→$250 and $350→$280. Guggenheim steady at $255, AI strength offsetting smartphone softness. Q1 was clean — license +23%, royalties +22%, data center royalties doubled, both line items beat by 200bps.
Net: Piper's initiation is a valuation story dressed as a growth story. At 270x, r/r hinges entirely on whether the accel IP/ASIC optionality is real by FY27. If it is, the bears' multiple compression argument is noise. If it slips, $250 is a gift. 25 analysts revising up though — the tape likes it.
Stifel reiterates Buy, $60 PT, and the point isn't the quarter — it's the register. GFS closed its $375M quantum CHIPS research award, issuing 9,907,399 shares to Commerce at $37.85 (1.75% of diluted shares). That's worse than the ~1% the Street modeled, and the fixed award converted at a ~15% discount to the reference close. Stifel shrugs: the EPS hit is rounding error against $375M of funding plus the geopolitical seat at the table.
Stifel — EPS dilution is minimal next to the funding and geopolitical positioning gained.
Structure matters more than the headline. Fixed dollar award, variable share count — the government's stake is inversely proportional to the stock, so a lower GFS means more shares outstanding. At $45.75, Commerce is already up on a stake it took at a discount. Layer the pending $300M silicon photonics LOI on top and cumulative government ownership hits ~3.14% at $45. GFS is now co-owned by Abu Dhabi and Washington, two sovereigns underwriting the same AI compute buildout.
Bull case is mature-node scarcity. TSMC is repricing wafers, capacity is tight, and Cantor sits at $90 Overweight on exactly that. Q2 beat ($0.46 vs $0.43, $1.79B rev), comms infra and data center guides went up, and GM prints ~30% for FY26 a year early.
Bear case is the volume base. UBS cut to $55 from $61, Neutral, on mobile now tracking a low-double-digit decline for the year. PT spread sits at $55 to $90 with the stock at $45.75 — PMs are arguing about terminal margin here, not the near-term setup.
JPMorgan starts SK Hynix at Overweight with a $245 ADR PT (June 2027, 20% ADR premium to local, 7x FY26-27 EPS), implying ~24% upside from $198.63. Stock sits right at the 52-week high ($199.87) with a 7.09 P/E — that's the whole pitch. Market is pricing a cycle, JPM says it's pricing a multi-year platform: 34% EPS CAGR over two years, memory ASPs up every quarter from Q1'24 through Q4'28 and beyond, and total shareholder return yield of nearly 42% across '26-'28.
The idiosyncratic kicker is capital return. 50%+ of capacity locked into long-term agreements, and management committed to distributing 50%+ of FCF — Barclays frames it as ~15% of market cap coming back. That converts HBM leadership from a pure pricing beta story into something with a visible cash-return floor. £40T buyback already board-approved (Needham at $220 post-announcement).
Street cluster: JPM $245, Needham $220, Barclays $300, GS Buy on the dip. Range reflects uncertainty on how long pricing holds, not on SK Hynix's position.
JPMorgan projects memory average selling prices will trend upward from the first quarter of 2024 through the fourth quarter of 2028 and beyond. The firm said memory content and value share will continue to rise in next-generation AI architecture.
Bull case in one line: whoever owns HBM capacity owns the AI capex cycle, and SKHY owns the majority of it plus returning half of FCF. Bear case: PTs have run $220→$300 while the stock is at all-time highs — the multiple is 7x because the market doesn't believe '28 ASPs. GS flags exactly that: recent weakness is memory pricing fear, not a demand break. Not sure we can read too much into a single initiation on a name this well-covered, but JPM anchoring the long tail of the cycle to 2028+ is the note that matters.
The sell-side is chasing price, not leading it. Stifel nudges its PT to $225 from $200 — and keeps a HOLD — while the stock sits at $256, i.e. the new target is 12% BELOW spot. That's a catch-up raise, not a conviction call. Shares have ripped 33% off the June lows into the print, and the Street is repricing a narrative flip (freemium = user growth optionality) faster than it's repricing the actual ARR math.
Merge the tape: Mizuho to $260 (Neutral), Barclays to $295, Stifel to $225, TD Cowen parked at $245 (Hold). Read the cluster — targets are climbing but ratings stay NEUTRAL/HOLD. Nobody wants to underwrite this at the current multiple. The operative story is user growth over monetization: Adobe running a more aggressive freemium push means MAU is the headline, ARR is the casualty. Q2 organic net new ARR already printed -3%, Adobe guided H2 ARR down $500M, and TD Cowen models net new ARR growth at -25% in the back half. Barclays models $400M net new ARR for FQ3 (seq decline on richer freemium mix) but sees $420M+ if web traffic and app downloads hold. Implied Q4 ARR sits at 10.2%, FY27 ~8% organic — essentially unchanged, which is the tell: no one's marking up the forward numbers off this move.
"Moving parts around the executive transition, acceleration of the freemium push, and evolution of the competitive landscape through artificial intelligence."
That's Stifel's own hedge — CEO handoff (Chakravarthy in), no new CFO after Durn's exit, Creative chief Wadhwani departing, all while AI rewrites the competitive map. Clean 89.4% gross margins and 11.5% LTM revenue growth keep the quality bid alive.
Bull: freemium converts to paid, MAU inflects, the AI narrative flips from threat to TAM expander, and FQ3 clears the Barclays $400-420M bar. Bear: this is a 33%-off-the-lows stock with a guide-down already in the tape, decelerating net new ARR, a C-suite in flux, and a sell-side that keeps raising targets while refusing to upgrade. You're paying up for a pivot that hasn't shown up in the numbers yet. Not sure we can read too much into one PT bump — the tells here are the Hold ratings sitting under spot.
Nothing new in the model here. Evercore used a webinar with Ken Duda (plus PANW's Klarich and CISOs from Arista, Anthropic, PANW) to frame ANET as a security name rather than a switching name — EOS's single code base as the resiliency pitch. That's a smart reframe because the actual stock driver is already in the tape: FIRST $3B QUARTER, $3.04B VS $2.83B CONSENSUS, EPS $1.02 VS $0.88, +33% REVENUE GROWTH LTM, SHARES +47% YTD. PT cluster sits at $240-259 (BofA $240, TD Cowen and Evercore $250, UBS $259), so Evercore's $250 is consensus, not differentiated. 26% UPSIDE from $243B cap.
The bull steelman: AI-accelerated vuln discovery is real and getting worse — Anthropic's Glasswing logged 23,000 open-source vulns in a month, 37,000 combined with PANW's findings. If attack tooling commoditizes (Evercore's own point: open-source models close the gap to frontier in ~3 months), enterprise security budgets shift toward infrastructure that can patch and mitigate at machine speed. EOS as a common baseline is a defensible wedge. The bear steelman: this is a webinar, not a purchase order. No dollar figure attached to the security thesis, no attach rate disclosed, and ANET still sells boxes into a hyperscaler capex cycle where the marginal buyer is already long. The security narrative adds multiple; it doesn't add revenue until someone shows a line item.
"Arista's EOS operating system as a resiliency leader, built on a single code base that provides a common baseline for discovery and mitigation of vulnerabilities."
Not sure we can read too much into a reiteration with no PT change on a name up 47% YTD. The setup is fine, the catalyst is thin.
JPMorgan throwing in the towel is the signal — upgraded to NEUTRAL from UNDERWEIGHT, PT to EUR24.30 from EUR17.80. This is a capitulation upgrade, not a fresh idea. The stock is +68% over the past year and sits near its 52-week high, and the catalyst was Wednesday's 22-year Google PPA for a data center fed off the Loviisa nuclear plant.
The number that matters: JPM figures the PPA cleared around EUR80-90/MWH against Nordic forwards in the 50s. That's a data-center buyer paying a fat premium to the curve, which tells you the market is pricing tighter Nordic supply/demand into 2030 than anyone modeled — including JPM, which now admits as much after Nordic wholesale power ran >10% since July on higher gas and DC demand.
JPMorgan said it underestimated the upside risks from data centers and tighter Nordic power markets.
Worth flagging the messy bits underneath: Q2 comparable EPS slipped to EUR0.08 from EUR0.09, nuclear output guidance got cut to 23-23.5 TWh, and management is already hedging on future ancillary-services income. Cash flow was the bright spot (EUR324M vs EUR203M), and the conditional tender for Norway's Elmera adds ~3.2-3.3M retail customers. But the equity story now is entirely the nuclear-to-hyperscaler narrative. At "balanced" r/r with the premium already in the tape, this is a name to own on weakness, not chase here.
Piper Sandler initiates OW with a $300 PT, and the argument is not whether AI spend happens — it's whether NVDA keeps the lion's share of it. 80% SHARE of AI compute, 50% UNIT SHARE, supply constrained for another TWO TO THREE YEARS, and agentic workloads bending GPU rental pricing upward. O'Connor models a 47% REVENUE CAGR through FY2030 to ~$30 EPS, which puts the stock at 14x FY28 — PEG of 0.3. That is the whole bull case in one line: you are not paying up for growth, you are paying up for the scarcity rent.
80% market share in AI compute, 50% unit share, supply expected to remain constrained for the next two to three years. — Piper Sandler initiation thesis
The rest of the Street sits in a tight cluster above him — StoneX $335, BMO $340 (with a 70% FY28 growth projection), Needham $300, Cantor reiterating OW while flagging the durability-of-the-AI-cycle debate as the live argument. Nobody's fighting about the number, they're fighting about the multiple.
Worth flagging: the HUGGING FACE acquisition at $12.93B ($11.9B to holders, up to $1B retention equity). That's NVDA buying the model hub — vertical integration into the software layer where the agentic workloads actually get built. Not sure we can read too much into one wire, but if compute is the rent, this is the landlord buying the building's directory. Semi tape helped: SOX +1.6%, INTC +9.8%, AMD +6.6%, ARM +4.7% on Citi's bullish call.
Verdict: the capacity number is the tell, and it just got bigger. MPWR blew past its old $4B annual revenue capacity bogey and reset to $6B+, with an $8B long-term marker. That's management signaling the AI data center smart power stage ramp is a multi-year build, not a one-quarter pop. Stifel stays Buy on it. The math backs the ambition — $3.27B LTM revenue, 28.7% growth, and the Street modeling ~48% for the current year.
The Singapore GFS long-term agreement is the piece worth chewing on. It keeps the Asian-dominated cost structure intact while cutting China exposure — supply chain de-risking without eating the margin. Sequencing matters: MPWR expands its existing 60nm-class node processes on the 300mm Singapore fab first, then moves to 40nm-class for next-gen. All of it aimed at low-voltage VRM smart power stages feeding AI DCs, auto, and robotics.
Partner base went from 5 to 13 since 2021. That's actual multi-sourcing depth, not a press release. GFS sits on top as incremental supply, not a substitute.
"The company has reached its prior annual revenue capacity target of $4 billion and set a new capacity target exceeding $6 billion, with a long-term goal of $8 billion." — Stifel
Q2 2026 print was clean: adj EPS $6.50, revenue $980.6M, +48% y/y, beat, and they raised the full-year enterprise data outlook. The Street just repriced hard — Truist to $1,889, Needham to $2,000 on DC revenue, KeyBanc to $2,100 on the Enterprise Data segment's growth, StoneX initiating Buy on the lean expense profile and margin structure. That's a tight cluster, and the dispersion is small — nobody's fighting the direction.
Bull: capacity targets lead revenue, not lag it. Tripling the ceiling is a forward demand statement. Add diversification (13 partners, Singapore ex-China), best-in-class margins, and a TAM that's now three-legged — AI DC, auto, robotics.
Bear: capacity isn't demand. If AI DC orders normalize, you're carrying a bigger fixed cost and depreciation base into a softer tape. The $6B is management's number, not signed backlog. And at these multiples, a single guide-down wipes out the re-rating.
Not sure we can read too much into one Stifel note — but the capacity reset is a hard number, and PMs should treat it as a genuine change in the company's own view of the runway.
UBS nudges its PT to $101 from $99, keeps Buy — this is a "the compounding is underappreciated" call, not a new thesis. The lever is GlobeOp (~30% of revenue) riding AUM growth at the mega multi-strats: Bloomberg counts SIX funds north of $75B now, the most ever. Scale begets outsourcing, and SSNC is the #1 global administrator — bigger funds need third-party admin, reporting, and middle-office, and they don't insource that back.
The print already shows the inflection. Q2 organic ACCELERATED TO 7.6% FROM 5.0% in Q1, GlobeOp 9.0% FROM 6.7%, EPS $1.76 vs the $1.68 bogey on revenue of $1.7B vs $1.66B. Needham flagged records on revenue, adj EBITDA, and EPS. PT cluster now sits in the high-$90s to low-$100s — DA Davidson at $96, UBS at $101.
UBS expects momentum to continue to build at the company.
Steelman both sides: the bull case is operating leverage — accelerating organic growth plus float/AUM tailwind means mid-single-digit revenue growth drops through at a much higher rate, and the hedge fund complex keeps concentrating toward the exact clients GlobeOp serves. The bear case is that a $2 PT bump is a nothing-burger, the multi-strat AUM stat cuts both ways (fund blowups and redemptions hit admin revenue too), and after the Q2 beat the re-rate may already be priced. Not much edge in a $2 nudge — watch GlobeOp organic next print, that's the number that matters.
Stifel reiterates Buy with a $225 target — UBS is sitting at $275 on the same name. That's a 22% gap between two bulls, and it maps exactly to the debate: does the AI narrative re-rate the multiple, or does the decelerating outsourcing book drag FY27 guide below consensus? Stock has ripped +37% IN THREE MONTHS vs S&P EQUAL-WEIGHT +2%. So the easy money's made and now it's a print-and-guide event.
Stifel's modeling an in-line FQ4 (Aug-end), implied midpoint of +1% ORGANIC CC with modest margin expansion. Bookings are the wildcard — consulting bookings trend "generally positive," outsourcing bookings decelerating, and management's own forecast is muddied by industry headwinds. FY27: Stifel at +1% CC ORGANIC VS CONSENSUS +2%, expects a 0-4% guide with consensus at the midpoint. Skepticism persists unless FQ4 bookings or FQ1 guidance supports acceleration.
"The stock is likely to trade based on the fiscal 2027 guidance and fourth quarter bookings."
Bull: Trades at 11x FCF INCLUDING SBC — cheap for a franchise levered to enterprise AI spend. Google Cloud partnership targeting midmarket AI, McCoy acquisition building the Dutch SAP/Edge business, investor day OCTOBER 14 as a potential re-rate catalyst. The 3-month move says positioning already leaning long.
Bear: 1% organic is not a growth stock's number. Outsourcing decelerating is the tell that discretionary IT budgets stay frozen. A 0-4% FY27 guide is a wide range that lets management sandbag — and the market's been rewarding that less and less. At 11x FCF the stock is cheap for a reason if AI revenue doesn't show up in bookings.
Net: Two bulls, two very different targets. $225 vs $275 is really a vote on whether the guide clears 2%. Watch bookings, not the EPS line. Investor day Oct 14 is the second catalyst — this is a two-event setup, not a one-print trade.
Stifel takes its PT to $130 from $110 and still won't put a Buy on it. That's the tell — the debate here was never the business, it's the multiple. Pre-quarter-end IR calls gave Stifel enough to underwrite 5.5% FY27 revenue growth (stable employment and pricing, Paycor ramp, PEO trends, ancillary attach) and, more importantly, 60BP of EBIT margin expansion versus a 25-50BP guide. Paycor cost synergies kick in mid-FY26, so the margin story front-loads into the back half. LTM gross margin sits at 74%, which is the foundation under that math.
The FY26 sign-off was already clean — FQ4 adj EPS $1.32 on $1.61B revenue, +12% YoY, full-year revenue +17% to $6.5B. The bar is set.
This is the line worth flagging to PMs:
"Paychex and the payroll sector appear to be moving out of the AI-loser category, with the stock re-rating to 18 times free cash flow."
That's a positioning trade, not a model. HCM and payroll have sat on the AI-disruption short list for two years — that's unwinding. 18x FCF for high-single-digit EPS growth plus a 4.1% yield (39 straight years of dividends) is Stifel's definition of fair, not cheap.
Rest of the street clusters well below: Wells $111 on float income, Cantor $107 on resilient payroll, UBS $98 on post-Paycor growth. Stifel now owns the top of the range with a Hold rating — a PT hike that says "I have to mark it up, I don't have to like it." Employment is the swing factor. If hiring cracks, the revenue build and the float income both go with it.
The print wasn't bad. The print was fine, and that's the problem. ServiceTitan beat revenue by 2.7% — versus a 3.7%-5.7% historical range as a public company and Stifel's ~5% bogey — then guided ahead by less than the beat, and the tape took the stock to $55.99, near its 52-week low of $54.17 (article also cites a prior close of $87.92 and a ~20% drawdown; can't make those numbers square, but either way this is a broken-chart tape). Not one of the six firms that moved cut the rating. Every single one cut the PT. That's the tell: nobody wants to be the downgrade, nobody wants to underwrite the number either.
Two separate problems got compressed into one selloff. First, real demand softness: usage revenue levered to gross transaction volume, and GTV came in light on weaker job growth and lead volume — particularly May and June, before a healthier July. Second, an accounting transition. Max is taking a bigger share of new deals, and the rev rec around it (ramping structure, consumption-driven models, management absorbing onboarding fees) creates $4-5M of top-line headwind across subscription and services as Max mix grows. Stifel's channel check says H2 subscription is largely within plan despite a smoother Q1-to-Q2 ramp. So: one of these is cyclical, one is self-inflicted and temporary. The market isn't paying up to distinguish.
"The subscription revenue beat was thinner than usual and usage revenue faced headwinds from slower than expected gross transaction volume growth due to lighter job growth and lead volume, particularly in May and June before a healthier July."
Stifel $100 (from $125), TD Cowen $100 (from $125), Truist $100 (from $110), BMO $90 (from $103), Piper $110 (from $115), KeyBanc $110 (from $120). That's a $90-110 cluster from a prior $103-125 band — and every one of them stays Buy/OW. Collective thesis: the beat cadence is compressing, the guide is conservative-to-cautious, and Max is a margin/revenue-mix story that resolves in '27, not next quarter. BMO flags "moderate upside against high expectations" — which is the honest version of "we still like it, just less."
Bull case: Max attach ramping, July inflecting, H2 sub within plan, and these PTs are mechanical resets off a lowered near-term model, not thesis changes. The onboarding-fee absorption is a deliberate land-grab — trading revenue now for a stickier consumption-based base later.
Bear case: a 2.7% beat is the wrong direction of travel for a name that used to clear 5%+, and the guide-ahead-by-less-than-the-beat pattern says management sees the same deceleration. Usage revenue is a macro derivative they don't control, and Max's rev rec change is a very convenient place for a real slowdown to hide.
r/r here is whether you believe July's "healthier" trend is the turn or the noise. I'd want one more print before paying up — but at a 52-week low with six Buys intact, the setup is at least interesting for a starter.
Verdict: Truist's $355 initiation (roughly 48% above the $240.35 print) isn't a multiple call, it's an earnings-model call — they think consensus 2027 ARR is simply too low, not that NBIS deserves a richer multiple on the same numbers. NBIS is +187% YTD, +157% over twelve months, so direction stopped being the debate a while ago. The debate is whether the buildout is supply-constrained enough to keep repricing, and Truist's answer is yes.
LTM revenue growth of 507% and Q2-26 revenue of $582.3M (+454% y/y, beat the $569.9M bogey) give them cover. ARR went $1.9B → $3.0B. That's the base they're extrapolating from.
Recent pricing ranges from $20 million to $25 million per megawatt for one-to-three-year deals and $40 million to $50 million per megawatt for shorter-duration contracts.
Read that second range carefully. Short-duration contracts clear at roughly 2x the per-MW rate of multi-year deals — that's a scarcity premium, not a discount for flexibility. Customers are paying up to avoid locking in. That's what Truist means by "contract economics," and it's why they model 2027 ARR well above consensus against a $2.1T cloud-layer revenue pool by 2030.
Compass Point went to $300 from $260 (Buy) on the AI cloud contract book. DA Davidson went to $250 from $175 — still Neutral — after Vineland data center approval removed a chunk of permitting risk. That's the classic "we're behind, but we're not believers yet" PT raise, and it tells you more about DA Davidson's positioning than NBIS's fundamentals.
The $4.5B convertible (two series, private to QIBs) funds the build with dilution and leverage, and 20-30% adjusted EBIT margins are a medium-term aspiration while capex stays front-loaded. Q2 adjusted EPS printed -$0.68 — the "-$51.46 expected" comp in the tape looks like a data artifact, don't read anything into it. If pricing rolls over, the whole 2027 ARR bull case deflates at once.
Not sure we can read much into a single initiation, but the per-MW print is the first hard number anyone's put on NBIS contract economics. That's worth tracking into the next print.
Guidewire beat the quarter and the stock got taken to the woodshed anyway — DOWN MORE THAN 20% on the print and -22.3% on the week. Revenue +3% vs. D.A. Davidson's bar, non-GAAP op income +25%, net new ARR of $95M (a $5M beat on guidance), ending fully ramped ARR up 21% y/y and total ARR at $1.24B (+19%). Nothing in here is bad. The problem is FY27 initial guidance came in IN-LINE, and a name trading at this multiple doesn't get paid for "in-line" — it gets paid for a raise. So you get a 20%+ drawdown on a quarter that would've been a non-event at a lower starting valuation.
The sell-side reaction tells you exactly where the debate sits. D.A. Davidson stays Buy, $222, arguing results were solid and management is making the right long-term moves. Citizens keeps Market Outperform, $220, pointing at upside hidden in the FY27 guide. Then you have Wells Fargo raising its PT to $195 while KEEPING an Underweight — that's a firm telling you the numbers are fine but the price isn't. PT cluster sits roughly $195-222, so even the bulls aren't screaming upside from here; this is a valuation argument, not a fundamental one.
"Guidewire ... believes management is making the right moves to allow Guidewire to remain the long-term leader in the InsureTech space." — D.A. Davidson, Buy, $222
Bull: Recurring revenue compounding 20%+ with $95M net new ARR and 25% op income upside on the quarter — operating leverage is real, InsureTech leader in a sticky, mission-critical vertical, and the FY27 guide is likely conservative given the pattern of beats. The selloff is a multiple reset on a name that needed perfection, not an execution break.
Bear: You don't de-rate 20% on an in-line guide unless the bar was already stretched. If FY27 is "in-line" at the top line, the ARR-to-revenue conversion isn't accelerating, and Well Fargo won't even upgrade at $195 — the r/r on further multiple compression from an elevated base is still unfavorable until estimates prove the guide was sandbagged.
Bottom line: Numbers good, guidance just OK, stock paid for great. Watch whether the FY27 range gets revised up on the next print — that's the tell on whether this was a flush or the start of a grind lower.
Oppenheimer stays at Perform and there's no PT to argue about — because this isn't a fundamentals story, it's a refi story with a melting revenue line attached.
The quarter was mixed in the least exciting way: revenue $98.2M vs Oppenheimer's $99.0M (essentially in line, not the disaster the tape wants to make it), adjusted EPS $1.17 vs $1.03. Management cut the FY27 revenue outlook by $12M at the midpoint on accelerating consumer softness plus the lingering hole from the H1 government bookings shortfall — but held adjusted EBITDA at $108-116M and FCF at $14-22M. Read that again: revenue guide down, profit and cash guide unchanged. Cost control is doing all the work. Enterprise held in line at -1% y/y, gross margin sits at a genuinely impressive ~76%, and the GTM redesign plus workflow cuts drove roughly 7 points of EBITDA margin outperformance.
Oppenheimer: "Skillsoft maintained its adjusted EBITDA outlook of $108 million to $116 million and free cash flow guidance of $14 million to $22 million through cost management."
One flag on the tape: the article cites a $125.4M revenue bogey as the miss that sparked the selloff. That number looks stale or wrong — Oppenheimer's own estimate was $99.0M and the print came in at $98.2M. Don't anchor to $125.4M. The real miss was the forward guide, not the quarter.
TSM (bullish/high) — August revenue hit NT$514.81B, the first >NT$500B month, +53.3% Y/Y and +10.8% M/M. Four straight record months. Q3 guide sits at $44.6B–$45.8B and FY USD growth runs slightly above 40%. Supply is the constraint, not demand. The 1.4nm pull-in talk is the leverage point: Taichung P1 trial April 2027, mass production 2H 2027, and four fabs needing 9,000–10,000 workers. The short side is a crowded-long/geopolitical hedge, not a fundamental call today.
MU (bullish/high) — Huawei raised Ascend 950DT indicative price to >250,000 yuan, +20–50% in two months, on HBM cost inflation. That closes the pass-through loop: Apple retail +$100, AT&T tariff pass-through, Huawei accepting full memory tax. One desk note says DRAM prices will not start declining double digits in 2H CY27. The bear case is real: DeepSeek V4.1 Flash cuts inference HBM needs to 1/4 and SSD to 1/8 via KV cache compression. Kioxia’s CEO also publicly called time on the rally. Memory bulls now have an architecture headwind and a credible bear voice to price.
ORCL (neutral/high) — Reports tonight. Street broadly expects the OpenAI contract to drive a cloud beat. Options chase an upside gap, but pricing is for beat, not re-rating. Asymmetry sits to the downside if guidance goes vague. Key items: RPO composition, OpenAI-linked recognition cadence, FY27 capex guide, and any verifiable framework for returns covering debt cost. This is the first public-company test of borrow-to-build-AI in a 4.5%+ yield world. Symbolically, ORCL matters more than its single-stock value; SB Energy IPO pricing two weeks later is the second credit test.
AMD (bullish/medium) — A whale bought AMD calls as part of a $315M+ premium spend across AI assets. Strikes are deep OTM and Jan-2027 dated. That is a big-money expression of an AI melt-up into 2027. No fundamental update, just a flow signal.
LITE (bullish/high) — Lumentum management at Citi framed scale-in optics as 1,000x bandwidth vs scale-across 1, scale-out 10, scale-up 100. Die-to-die and die-to-memory optical links under one metre are the next S-curve. Management said Lumentum’s best days are yet to come. LITE remains the favorite for high-power external lasers in CPO/NPO, but the moat is physics-based, not permanent. Watch InP/MOPA challengers.
COHR (bullish/medium) — Coherent-lite optics is the next major incremental opportunity for COHR, MRVL, and NOK. The market underestimates scale-across TAM. COHR is one of the few US high-power laser competitors. Optical interconnect scarcity is broadening beyond pluggables.
NOK (bullish/medium) — Nokia showcased an integrated optical offering at the optical communications expo. The market significantly underestimates scale-across TAM. Coherent-lite is a major incremental opportunity for NOK. This is a cheap optionality read on optical infrastructure, not a standalone fundamental breakout yet.
SMTC (bullish/medium) — Semtech’s call was incredibly bullish and the stock lags fundamentals. It remains a consolidation candidate on optical interconnect exposure. All-time highs do not define a stock. The tape treats it as a lagging optical recovery name.
FORM (bullish/high) — FormFactor’s TRITON is moving from concentrated customer development to multi-fab and fabless production adoption. CPO test is shifting from R&D infrastructure to manufacturing infrastructure. FormFactor is architecture-agnostic across CPO, NPO, and pluggables; the bottleneck is capacity, not demand. NVIDIA is paying premiums to lock probe-card and test-socket capacity in Taiwan, and FORM is named alongside AEHR, Winway, and MPI. Testing is becoming a structural AI hardware bottleneck.
AEHR (bullish/high) — AEHR spiked 24% as the market priced burn-in and wafer-level reliability as a structural bottleneck. It was named in NVIDIA’s probe-card/test-socket capacity lock. The test-capex recognition is finally happening. This is a high-beta expression of the same theme as FORM and TER.
KLAC (bullish/high) — KLA revenue grew 12% in 2024, 17% in 2025, and is on track for low-20% growth in 2026. Operating margins expanded ~400bps over that timeframe. Advanced packaging inspection share rose from ~2% in 2023 to 7–8% in 2026, growing 70% Y/Y at ~$1.1B, 2x market growth. A disclosed book held KLAC long alongside SWKS, AAPL, DELL, LLY, DHR, RGEN, and INTC, and shorted ASML. That pair expresses preference for process control and packaging inspection intensity over leading-edge litho units.
ASML (neutral/medium) — EUV transient thermal deformation causes up to 2nm image shift with dose, deteriorating contrast and overlay. That raises metrology and inspection intensity at the leading edge. It also highlights a physical limit that benefits process control suppliers. The disclosed book shorted ASML while long KLAC. Treat that as relative positioning, not a standalone fundamental call.
TER (bullish/high) — Teradyne’s CEO said AI increases test required per chip across advanced nodes, chiplets, HBM, and CPO. CPO test-equipment TAM could go from $100M in 2026 to $700M in 2028. Exposure is broadening across AI ASICs, networking, HBM/DRAM, SSDs, silicon photonics, and system-level burn-in. Test intensity is a structural tax on AI hardware complexity.
STX (bullish/high) — High-capacity HDD demand is outstripping supply and the gap is widening. Seagate targets at least 25% exabyte CAGR over the next 2–3 years via HAMR rather than unit growth. This is a margin-accretive scarcity posture. HDD is no longer just a value trade; it is a capacity-tightness trade.
SNDK (bullish/medium) — A whale bought SNDK calls alongside SK hynix, Roundhill Memory ETF, INTC, AMD, BE, and CRWV. The expiry is Jan-2027 and deep OTM. That expresses a NAND/memory scarcity view through sophisticated call flow. NAND remains the cheaper way to express memory tightness versus HBM.
TSEM (bearish/low) — Tower Semiconductor was short in a disclosed book alongside TSM, VRT, ASML, and AVGO. The position is a contrarian bet against crowded foundry strength. It signals concern over Taiwan premium and foundry cycle positioning. No company-specific catalyst in the feed.
SWKS (bullish/low) — Skyworks was long in a disclosed book alongside KLAC, AAPL, DELL, LLY, DHR, RGEN, and INTC. The position is an RF-content play in the Apple ecosystem. It is a portfolio signal, not a new fundamental datapoint.
MRAAY (bullish/medium) — Murata is EOLing a large number of commodity MLCC part numbers across multiple applications. Capacity redirects to high-end miniaturized high-capacitance MLCCs for AI servers, data centers, and autos. Taiwan and China suppliers could pick up replacement orders. This is a mix-shift and pricing-power signal for Murata.
ASX (bullish/high) — ASE’s packaging/test capacity is sold out and monthly revenue broke above NT$800B for the first time. The read-through is advanced packaging and test capacity remain the bottleneck for AI chips. This supports the entire OSAT/packaging tightness thesis. It also validates the FORM/AEHR/TER test-capex chain.
KYOCY (bullish/low) — Kyocera is named in the AI server PCB/substrate supplier list alongside Ibiden, Unimicron, Shinko, Toppan, SEMCO, Daeduck, Korea Circuit, LG Innotek, ZDT, Nan Ya PCB, Kinsus, Shennan, and Fastprint. The list is a direct ODM shipment beneficiary set. The signal is thematic and low specific.
FSLY (neutral/medium) — Agent-driven traffic from Instinct, Grok Bot, and Muse is a live disruptor for travel/shopping/dining sites. Edge players with agent-friendly delivery exposure are the read-through. Fastly is named in the agent traffic debate. The signal is thematic, not company-specific.
NET (bullish/low) — Cloudflare is cited as an edge player with agent-friendly delivery exposure. The meme “Always Buy Cloudflare” appears in the agent traffic debate. The read-through is agent-driven traffic disrupts travel/shopping/dining sites. Low-conviction, but directionally supportive for edge delivery.
AKAM (neutral/low) — Akamai is named in the agent-driven traffic debate alongside FSLY, NET, META, and SHOP. Agent traffic could disrupt travel/shopping/dining sites and change edge delivery demand. The signal is thematic, not company-specific. No numbers to underwrite.
SHOP (bearish/medium) — Shopify is mentioned in the context of being stuffed back into the AI loser basket. The risk is not extinction but missing the agent layer, which is the entire growth market. Incumbents still lack an answer for agent commerce. This is a narrative short, not a print-driven call.
CRWV (bullish/medium) — CoreWeave calls were part of a $315M+ whale premium spend across AI assets. The strike is deep OTM and Jan-2027 dated. This expresses a leveraged AI compute melt-up view. CoreWeave also sits among compute race names alongside MSFT, AMZN, NBIS, IREN, HUT, WULF, CIFR, and AMD. Financial backstops from NVIDIA, Google, and Broadcom fund the buildout. Vendor financing concentration is the key credit question.
IREN (bearish/medium) — IREN’s bull case is challenged by the argument that contracted power is not delivered capacity. Transformers, switchgear, and cooling are borked and 2x more expensive. Power and shells are not the bottleneck; the rest of the supply chain is. That is a direct challenge to the “own the power” thesis.
HUT (neutral/low) — Hut 8 is listed among the compute race names alongside MSFT, AMZN, CRWV, NBIS, IREN, WULF, CIFR, and AMD. The buildout is backed by NVIDIA, Google, and Broadcom financing. The key credit question is vendor financing concentration. No company-specific catalyst.
WULF (neutral/low) — TeraWulf is listed among compute race names alongside MSFT, AMZN, CRWV, NBIS, IREN, HUT, CIFR, and AMD. The buildout is being funded with vendor backstops. The market is still debating credit concentration. No fresh numbers.
CIFR (neutral/low) — Cipher is listed among compute race names alongside MSFT, AMZN, CRWV, NBIS, IREN, HUT, WULF, and AMD. The buildout is backstopped by NVIDIA, Google, and Broadcom. Vendor financing is the key credit question. No company-specific update.
MSFT (neutral/low) — Microsoft is listed among the compute race names alongside AMZN, CRWV, NBIS, IREN, HUT, WULF, CIFR, and AMD. The AI buildout is being funded with NVIDIA, Google, and Broadcom backstops. The key debate is vendor financing concentration. No new fundamental signal.
AMZN (neutral/low) — Amazon is listed among compute race names alongside MSFT, CRWV, NBIS, IREN, HUT, WULF, CIFR, and AMD. The buildout is being backstopped by NVIDIA, Google, and Broadcom. Vendor financing concentration is the key credit question. No fresh catalyst.
GOOGL (bullish/medium) — Google signed a 25-year PPA for NextEra’s Duane Arnold nuclear restart, with DOE loan up to $1.9B for 615 MW, target Q1 2029. Google also has a 22-year deal for up to 50% of Fortum’s Loviisa nuclear plant. Google pays production costs, not Iowa customers. Watch undisclosed overrun and price terms. Google also acts as one of the financial backstops helping fund the OpenAI/Anthropic compute race. That deepens its role as both competitor and financier; circular credit risk is the debate.
NEE (bullish/high) — NextEra closed a DOE loan up to $1.9B to restart the 615 MW Duane Arnold nuclear plant, shut since 2020. Google’s 25-year PPA provides contracted revenue to recover investment; CIPCO buys remaining output on same terms. Target Q1 2029. Electricity prices, financing terms, and overrun responsibility were not disclosed. This is a concrete AI-power infrastructure win.
BE (bullish/medium) — Bloom Energy calls were part of a $315M+ whale premium spend across AI assets. The position is deep OTM and Jan-2027 dated. This expresses an AI power and fuel-cell melt-up view. Pure flow signal.
VRT (bearish/medium) — Vertiv was short in a disclosed book alongside TSM, TSEM, ASML, and AVGO. The short is contrarian versus the datacenter construction/equipment orders narrative. It signals concern that cooling/equipment names are over-owned. No company-specific catalyst.
PLTR (bullish/medium) — NVIDIA is deploying Palantir Foundry/AIP with customized Nemotron models across its own supply chain. Enterprises can customize with proprietary data and deploy on-prem or cloud. Rackspace’s sovereign pod also combines Blackwell, Palantir Foundry/AIP, and managed ops. This deepens Palantir’s sovereign AI and industrial stack. It is a distribution and credibility win.
RXT (bullish/medium) — Rackspace joined NVIDIA Cloud Partner Program and launched an Institutional Sovereign Pod combining Blackwell compute, Palantir Foundry/AIP, and managed ops. It targets regulated and government customers. This is a specific sovereign AI product signal. Small stock, but it puts RXT in the sovereign AI conversation.
ABNB (bullish/medium) — Chesky’s GS fireside chat was among the most bullish AI commentary, signaling a vibe shift from a few months ago. He argued consumer AI token consumption is nowhere near its upper limit. This is a token-consumption bull datapoint beyond single-player chatbots. It supports the agent commerce and consumer AI theme.
PANW (bearish/low) — A cryptic post argues Palo Alto’s margin is Anthropic’s opportunity, with a short PANW disclosure. The read-through is AI agents could pressure security software margins. No elaboration in the source. Treat as sentiment, not a model change.
TCEHY (neutral/medium) — Tencent WorkBuddy enterprise SaaS is priced at Rmb 198/month/seat including 2,000 credits. The smallest credit pack is 2,000 credits for Rmb 100. Included credits doubled July–September. CSO James Mitchell said paying-user gross margins match Tencent Cloud’s overall gross margin, though Tencent Cloud’s margin has never been disclosed. Watch the credit-based SaaS pricing model.
UBER (neutral/low) — Uber priced a €4.5B debut euro bond. Commentary linked it to hyperscaler supply indigestion and broader credit demand. The read-through is large AI-driven issuance is competing for credit market capacity. No equity fundamental signal.
NDAQ (bullish/medium) — Nasdaq is investing $100M in Payward, Kraken’s parent, at a $21B valuation. Kraken will distribute tokenized Nasdaq-listed stocks; token holders get same voting rights. Nasdaq plans its own token debut Q2 2027. This is traditional finance building tokenized equity infrastructure. It is a quiet but strategic crypto/TradFi convergence signal.
AMD (bullish/medium) — Already covered above. Whale call flow is the only new signal.
FCX (bullish/high) — LME copper rose to $14,858.50/t, surpassing prior peak of $14,779. Comex copper hit a fresh record of $6.894/lb. The red metal has gained 19% YTD on data center, renewable, and grid demand. This validates the grid/buildout trade in price. FCX is the liquid large-cap expression.
TECK (bullish/high) — Copper hit fresh records on LME and Comex with data-center demand cited as a structural driver. Teck is a direct copper producer levered to the grid and buildout trade. The same supply-demand imbalance that lifted copper 19% YTD supports Teck. No company-specific news, pure commodity leverage.
SCCO (bullish/high) — Copper records across LME and Comex, with data-center demand cited as structural. Southern Copper is a pure-play copper producer. The 19% YTD gain and record pricing directly support earnings leverage. Watch Peru/Mexico political risk as the offset.
BHP (bullish/high) — Copper records on LME and Comex, with data-center demand and grid buildout cited. BHP is a diversified miner with major copper exposure. The commodity squeeze is broadening into metals and agriculture. Copper is doing the talking.
VLO (bullish/medium) — Refiners are the big active debate on ULSD backwardation, with diesel crack at record $5.94/gallon and Brent above $100. Valero is directly levered to distillate cracks. The macro squeeze is broadening into energy. This feeds higher-for-longer inflation.
MPC (bullish/medium) — Refiners are active on ULSD backwardation, with diesel at record $5.94/gallon and Brent above $100. Marathon Petroleum is levered to distillate cracks. The energy inflation is feeding higher-for-longer macro. No company-specific update.
PSX (bullish/medium) — Phillips 66 is in the refiner debate on ULSD backwardation and record diesel cracks. Brent above $100 and product scarcity support refining margins. The macro read-through is energy inflation is not easing. No standalone catalyst.
DK (bullish/medium) — Delek is named in the refiner debate on ULSD backwardation and record diesel cracks. Brent above $100 and product tightness support refining margins. The signal is macro-driven, not company-specific.
TSLA (bullish/low) — Tesla’s virtual powerplant grid stabilization system is cited as an under-priced optionality line alongside the auto business. It ties Tesla into the AI power and grid management theme. The signal is early and small versus the core auto business. Low conviction.
AVAV (bullish/medium) — AeroVironment beat, with defense demand strong. The print is part of a broader strong defense tape. The read-through is defense spending remains a safe haven amid geopolitical risk. No specific numbers in the feed.
HAWK (neutral/low) — Hawk is mentioned as post-lockup with strong defense/consumer earnings context. No specific numbers were provided. The signal is flow-related rather than fundamental. Keep on the lockup-expiry watchlist.
SIG (bullish/medium) — Signet was part of strong consumer earnings. The consumer tape was healthy across SIG, AEO, and CHWY. The read-through is the consumer is not cracking yet despite rate pressure. Watch discretionary trade-down risk.
AEO (bullish/medium) — American Eagle was part of strong consumer earnings. The consumer tape was solid across SIG, AEO, and CHWY. This supports the soft-landing consumer narrative. No company-specific number in the feed.
CHWY (bullish/medium) — Chewy was part of strong consumer earnings. The consumer tape held up across SIG, AEO, and CHWY. Pet spending remains resilient despite macro pressure. Good read on non-discretionary pet wallet.
LOVE (neutral/medium) — Lovesac Q2 revenue $161.2M missed $165M and grew +0.4% Y/Y. EPS was $0.51 vs estimate -$0.36, but FY27 revenue guide $690–710M missed $715M and Q3 guide was below. FY27 EBITDA guide $31.5–35.5M missed $38.8M. Mixed print with cost control but weak top line. Cost cuts are carrying the print.
FLWS (bearish/high) — 1-800-FLOWERS Q4 revenue $293.1M missed $294M and fell -12.9% Y/Y. Adj EPS -$0.80 missed -$0.69; adj EBITDA -$31.0M missed -$30.3M. FY27 guide EBITDA $10–15M badly missed $27.7M with mid-single-digit revenue decline. Company is evaluating asset sales and capital raising. This is a broken consumer discretionary tape.
NVS (bearish/medium) — Artisan Partners is calling for a board shake-up at Novartis after drug trial setbacks. David Samra said successive chairmen failed on acquisitions and the deal team is uninspiring at best. This is an activist governance signal. Watch for board changes and portfolio review.
DHR (bullish/low) — Danaher was long in a disclosed portfolio alongside KLAC, SWKS, AAPL, DELL, LLY, RGEN, and INTC. The position is a quality defensive long within an AI-heavy book. It is a portfolio signal, not a new fundamental datapoint.
LLY (bullish/low) — Eli Lilly was long in a disclosed portfolio alongside KLAC, SWKS, AAPL, DELL, DHR, RGEN, and INTC. The position is a healthcare defensive long. It is a portfolio signal, not a new fundamental datapoint.
RGEN (bullish/low) — Repligen was long in a disclosed portfolio alongside KLAC, SWKS, AAPL, DELL, LLY, DHR, and INTC. The position is a bioprocessing long. It is a portfolio signal, not a new fundamental datapoint.
DELL (bullish/low) — Dell was long in a disclosed portfolio alongside KLAC, SWKS, AAPL, LLY, DHR, RGEN, and INTC. The position is an AI server and enterprise hardware long. It is a portfolio signal, not a new fundamental datapoint.
ODD (bullish/low) — ODD moved +35% in the trading picture. No fundamental detail was provided in the source. Treat as a momentum move without a confirmed catalyst. Do not chase without a print.
ADI — No fresh RSS/Twitter signal. Nothing actionable. Keep on watchlist for next print/flow. AEP — No fresh RSS/Twitter signal. Nothing actionable. Watch power/AI infrastructure read-through. IONQ — No fresh RSS/Twitter signal. Nothing actionable. Quantum remains a narrative trade. KDK — No fresh RSS/Twitter signal. Nothing actionable. MCHP — No fresh RSS/Twitter signal. Nothing actionable. MP — No fresh RSS/Twitter signal. Nothing actionable. Watch rare earth/defense policy flow. MXL — No fresh RSS/Twitter signal. Nothing actionable. NOW — No fresh RSS/Twitter signal. Nothing actionable. ON — No fresh RSS/Twitter signal. Nothing actionable. PSQL — No fresh RSS/Twitter signal. Nothing actionable. TTD — No fresh RSS/Twitter signal. Nothing actionable. Agent commerce read-through only. TXN — No fresh RSS/Twitter signal. Nothing actionable. VRNS — No fresh RSS/Twitter signal. Nothing actionable. ZS — No fresh RSS/Twitter signal. Nothing actionable. Security margin debate remains PANW-specific.