Good morning. Futures grinding higher into the AMAT/TSM tape — semis leading, and AMAT's record print sets the tone: BEST-EVER QoQ GROWTH, MANAGEMENT SIGNALING >30% YOY UPSIDE. TSM keeping the bid under the group: Arizona ramping ahead of plan, 2nm potentially FOUR QUARTERS EARLY, overseas fabs now profit-contributive — the geo-risk discount keeps deflating. Ugly duckling on the same tape: Ulvac (6728 JP), 7-YEAR WORST DROP on a guidance miss the company blamed on production constraints, not demand. Delivery, not demand, is the bottleneck. Asia: China July data missed all three bogeys — industrial, consumption, investment — but FX reserves printed the LARGEST QUARTERLY GAIN IN 12 YEARS and Qwen downloads topped the charts. Macro sinking, AI floating, RMB in the middle. And the policy fuse just got lit: Washington drafted an ultimatum to 35 countries — join China's AI bloc, you're out of the US framework. Three themes frame today. First, NVDA halved its OpenAI Ohio backstop to BELOW $120B but is negotiating a $3B EQUITY STAKE in SB Energy — credit risk and demand exposure are being split apart. If neocloud (CRWV, NBIS) opens weak on this headline, that's positioning artifact, not re-rating. Second, pricing power is broadening from memory into equipment and materials — JPM's semis report, AMAT's record, Phison saying 2027 TIGHTER THAN 2026, Samsung/Hynix capex +35% YOY TO KRW 43.2T. Market still prices this as memory single-cycle — that's the mispricing. Third, the AI credit layer is becoming a macro rates variable: $1.5T OF AI-RELATED DEBT ISSUANCE YTD (+36% YoY) crowding out Treasuries, holding yields up — the slow-burn killer for high-duration AI multiples. We'll hit up AMAT and TSM first, then get to the optics complex (LITE/COHR/AAOI) — InP substrate prices jumping >10% IN Q4, biggest hike on record — and keep one eye on neocloud's open post-backstop.
FOUR FIRMS LIFTED PTs TO THE $230-235 CLUSTER FROM THE $175-181 RANGE AND THE TRADE ISN'T DONE YET. This is the AI security consolidator play that's actually showing up in surveys, not just sell-side pitches. At $216.95, with a 52-week high of $227.50 overhead, this is a breakout-or-bust setup into earnings.
TD Cowen is the high-end print at $235 (Buy) — they're calling 24% subscription growth in fiscal Q2 on Falcon platform consolidation. (Falcon now has 33 modules vs 30 last year.) They also re-rated the TAM to $149B from $116B — that's a 28% expansion in a single year, which tells you the market's starting to price AI security as a separate line item, not a feature.
Wells Fargo goes to $230 (Overweight) and their partner survey is the meat. Here's the line that matters:
"Pipeline strengthened quarter-over-quarter to 34% net above normal seasonality from 19% last quarter."
That's the second-half re-acceleration evidence the market wants. The 31% net above plan was a decel from 47% last quarter — don't bury that — but the pipeline number is the leading indicator. WF also notes CRWD tied PANW as the most likely consolidation winner and ranked #2 in AI positioning.
Here's the nuance: Q1 NNARR upside was only ~2% vs the historical 8-12% beat pattern in FY26. That's why the stock isn't at $250 already. But — and this is the key — Q2 guidance embeds NNARR growth accelerating ~100bps to 29% organic y/y. That's a step-change in the growth algorithm, not a one-off.
Bull: Black Hat checks confirm enterprise budgets are shifting toward securing AI at the endpoint — runtime protection, governance, autonomous defense. CRWD is the category leader in the place where those spend decisions originate. TAM expansion plus consolidation share gains plus AI attach = compounding multiple expansion.
Bear: The survey decel (31% vs 47% net above plan) says the field isn't as strong as the stock's 103% one-year run. The XM Cyber deal is IP-only — 45 patents, no revenue, no customers. That's not incrementally helpful to FY27 numbers. At ~$217 you're paying for perfection.
The setup is simple: a re-rating trade on pipeline inflection, with a technical breakout at $227.50 as the confirmation. The $230-235 PT cluster gives you 6-9% upside from here, but the real juice is in the NNARR print. If Q2 shows that 29% organic acceleration — with AI security as the named driver — this thing trades through the targets. If not, you've got a $230 lid on a $217 stock. The Cerebras partnership is the narrative kicker, not the earnings driver — at least for this quarter.
DB throwing in the towel is the signal. Not on fundamentals — on price. WDAY rips ~26% since June 1 while CRM/NOW/ADBE/ADSK/INTU AVERAGE DOWN 4% over the same window. That's a valuation call, pure and simple. The business didn't change. The multiple did.
DB cuts to Hold from Buy, raises PT to $220 from $180. No fundamental deterioration — they still call it "one of the highest-quality franchises in enterprise software." This is a risk/reward call after the stock ran away. 6-MONTH RETURN ~39%, RSI screaming overbought, and now a Silver Lake take-private headline doing a lot of the heavy lifting.
"The rating change is not driven by a deterioration in fundamentals."
That's the quiet part. Business is fine. Stock got ahead of the story. DB explicitly says valuation now reflects the quality of the business plus a probability of a strategic outcome. (Translation: you're paying up for a deal that may not happen.)
Meanwhile BMO still at Outperform with a $182 PT — STOCK TRADES $201.68. Their target is 10% BELOW the market. That's a stale, non-covering PT if I've ever seen one. MCH upgraded to Buy with a $150 PT. CLSA sits at $92 Underperform. MS at $145 Underweight. UBS Neutral at $220. DB Hold at $220.
The target range is $92-to-$220. That's not analyst dispersion, that's narrative chaos. Nobody has conviction in the same direction.
Bull case: Mission-critical HCM/FIM, 76% GROSS MARGIN, switching costs for days, high recurring revenue. Bhusri back in the CEO seat and employee/customer morale sounds genuinely buoyed. AI products early innings but workday's data position in HR/payroll is a legitimate moat. BMO sees a reasonable print and a PEG of 0.78 — if growth holds, the stock isn't even expensive on a relative basis.
Bear case: Core growth decelerating, full stop. AI disruption overhang on seat-based enterprise software is real — why pay per head when agents do the work? And the biggest near-term risk: a chunk of this valuation is Silver Lake option value. If the deal talks go quiet, this stock gives back the whole Musk-to-Mar-a-Lago move in a week. CLSA at $92 isn't crazy — it's just early.
Ten days out. BMO says results are "unlikely to quell debate about the durability of the mid-term growth framework." That's analyst-speak for: even a good quarter won't settle this one. Watch for AI agent attach commentary, any Bhusri color on the take-private rumor mill, and billings trajectory. But the real tell is how the stock reacts TO the print. At this valuation, it's a multiples story, not an EPS story. Good numbers with a fade = the re-rating is done. Bad numbers outright = air pocket.
AMAT blew past the bogey — October guide at $10.25B vs ~$9.6B consensus. The whispers ran $10.5B, and that gap tells you everything: expectations are priced for perfection at a 51.5 P/E. Demand isn't the debate. Margins are. The bull case won this quarter; the r/r from here is a margin-story fight.
Q3 FY26: EPS $3.50 vs $3.40 est, revenue $9.12B vs $8.99B. Clean beat, then the guide does the heavy lifting. Semi systems tracking ~40% growth in CY2026, raised. AGS compounding at 20%+. Management called CY2027 strong, flagged multi-year visibility from lead customers. Stock at $508, market cap $405.8B, up 186% over the past year. That multiple is the bull/bear dividing line.
Cantor framed it best:
Applied Materials raised its growth outlook for semiconductor systems, now tracking to approximately 40% growth in calendar year 2026.
Bernstein up to $700 from $675; Stifel and Mizuho hold $650 Buy/Outperform. The cutters: BofA $720→$650, UBS $705→$675. Look at why they cut — gross margin, not demand. Nobody is short demand here. Bernstein's own math shows the multiple doing less work: $700 at 33x avg FY27-28 EPS of $20.98, DOWN from 35x. EPS carries the PT, not valuation. Cantor sits at $850 Overweight — the outlier, comfortable on the limb.
Bull: This is a WFE supercycle, not a two-quarter pop. 40% systems growth across leading-edge foundry/logic, DRAM, and advanced packaging — AGS gives the recurring revenue layer — and management explicitly guiding 2027 strong. AMAT keeps outgrowing WFE itself.
Bear: Margins lag peers, and two firms trimmed PTs on a beat-and-raise print. That's the crack. The whisper miss says buy-side expectations ran hotter than the guided number. At 51.5x trailing, the stock prices in perfection. Any gross margin wobble creates air pockets.
JD IS DOWN 13% IN A WEEK BECAUSE THE MARKET FINALLY DECIDED THE TOPLINE MATTERS MORE THAN THE MARGIN STORY. Q2 revenue fell 3% YoY to 346B yuan (decel from +5% in Q1) and that overwhelmed a clean earnings beat — 6.29 yuan per ADS, +12% vs consensus. Management says growth reaccelerates in H2. The stock says prove it.
The revenue miss is real but narrow: 1% below Susquehanna's bogey, 1% above consensus. The mess is inside the mix. Product revenue printed 2% above consensus (JD Retail held up), but services revenue came in 4% below the Street. Electronics/home appliances took the hit — higher component costs plus category base effects. Not a demand collapse. Not a reacceleration either.
The margin side is actually the good news:
The Street is constructive but shaving numbers. Bernstein reups Outperform/$40 and calls the dip unsurprising after the recent rally. Benchmark holds Buy/$42. Mizuho trims to $39, Macquarie to $37 — both Outperform. Susquehanna is the outlier: cut to Neutral/$30, basically saying this is dead money.
"The decline in JD.com shares following the quarterly results was unsurprising after a recent rally. Management reaffirmed expectations for growth reacceleration in the second half." — BERNSTEIN
That's the bull case in two sentences. Cheap valuation ex the optics, margins inflecting, losses peaking, aggressive buyback under the stock. The bears just say the revenue tape is the only thing that matters, and it's pointing down.
BULL: Margin recovery is compounding — retail margins up, food delivery losses halved, buybacks aggressive. At $29 you're getting the H2 reacceleration call nearly free. If management is right, the $37-42 PT cluster is the setup and this is a gift.
BEAR: The decel has momentum: +5% in Q1, -3% in Q2. That's a trend, not a base effect. Services revenue missing by 4% says the high-margin mix shift is stalling. And the food delivery loss path is "more gradual" than bulls wanted — so the earnings bridge has a hole. Susquehanna's $30 PT says the r/r is symmetric at best.
H2 reacceleration is the entire ballgame. The first real tell is the Q3 print — revenue back to growth with margins still expanding puts this 30-40% higher. Decel persists and the stock grinds sideways. Watch SAMR's merchant fee reduction plan too — modest headwind to take rates, though JD's marketplace mix is lower than Alibaba's. And watch buyback pace: management has been aggressive, and that's the floor under the stock.
Verdict: Downgrade on valuation, not technology. Deutsche Bank moves S to Hold (from Buy) but RAISES PT to $24 from $17 — that's the tell. They're more constructive long-term, but the r/r after a ~90% rally off April lows is shot. This is a "price already paid for the good news" call.
The market is pricing a demand inflection from Mythos. DB isn't there yet. Their proprietary checks — 51 cyber resellers plus Black Hat field work — came back mixed. And the AI-era share battle looks like it belongs to the platforms: PANW and CRWD now trade at CY27 FCF multiples roughly 4x growth while everyone else scrapes for single-digit revenue multiples. S is fighting the wrong war at the wrong valuation.
The raised PT is the long-term optimism. The downgrade is the near-term reality. New CFO Sonalee Parekh gets a watchful eye — the Netskope analog (CFO exit + ARR decel) shows how fast this group punishes execution wobbles. S needs to prove it can compound before it deserves sponsorship up here.
SentinelOne and Netskope are likely "show me" stories at current levels and will need to show consecutive quarters of improving execution with sustained margin leverage to warrant further investor sponsorship.
That's the whole debate in one line. Steelman the bull: Mythos is real, the platform transition is working, and a fresh CFO can tighten the ship. Steelman the bear: 90% rallies in crowded cybersecurity names with decelerating growth and mega-cap competition — you know how that movie ends when the print disappoints. Buyer beware.
GS starts it Buy, PT $400 — 19% upside from $331.71. This is a conviction call, not a catch-up trade: stock already ripped 210% in the last year. The thesis is structural — AI-driven complexity in 3D integration and advanced packaging breaks yields, and that makes process control intensity (read: Onto's inspection/metrology gear) a bigger slice of every fab's capex dollar.
Growing 3D integration and design complexity make yield preservation the industry's critical focus — a structural shift that raises process control intensity across chip manufacturing.
Fundamentals back the story. Q2 was a beat AND raise: EPS $1.93 vs $1.69 est on $343.1M revenue vs $325.25M est — A RECORD QUARTER, +35% YoY, with H2 guidance lifted on top. New Dragonfly G5 and Atlas G6 platforms give GS confidence on share gains, and they see Onto outgrowing the WFE market through 2028 with margin expansion.
Risk is the obvious one — $16.3B market cap with consensus PTs clustered $330-450, so at $331 you're sitting at the low end of the range. If the AI capex narrative wobbles, multiple compression comes fast. But with the product cycle and beat-and-raise momentum, r/r still skews positive into the $400 bogey.
RBC lifts PT to $1,600 from $1,300 — stock sits AT $1,585. They keep Sector Perform. That's a grudging mark-to-market, not conviction. RBC saw the analyst day, liked the long-term targets, still won't bless the multiple until NAND pricing survives a downturn.
Management's targets now EXCEED RBC's model. The kicker: financial guarantees from three U.S. hyperscalers underpin NBM pricing. Fixed-price NBMs could cover >50% OF VOLUME THROUGH FY2028. That's the bull story in one line — contracted durability, not spot-NAND roulette.
The bear rebuttal, in RBC's own words:
"The firm expects the stock to trade on traditional industry metrics until pricing durability through a cyclical downturn becomes more apparent."
RBC leaves estimates unchanged (proof, not guidance, please). NAND supply/demand reaches better balance in H2 2027 by their math — that's the clock they're running.
Rest of the Street is much hotter. Bernstein $3,000, Goldman $2,200, Mizuho $1,900, Argus upgraded to Buy. Even Jefferies — cutting PT to $1,750 on margin concerns — stays Buy. SNDK IS UP 3,174% OVER THE PAST YEAR and the bulls aren't blinking. HBF samples in 2027 are the next secular catalyst; not in anyone's model yet.
Key level: $1,600. Clears and holds it, RBC's Sector Perform at target looks stale fast.
H.C. Wainwright is the only one at the party. Reiterated Buy and $4 PT — 205% upside from $1.31. Nothing new in the note, just repackaging the AI narrative after Thursday's Q2 print and the company's first-ever earnings call.
The hook: management ballooned the total development pipeline to 6.3 GW from 4.3 GW, adding ~300 MW across Rosa, Hedy, Ellen and Fei — all dedicated to AI/HPC workloads. Kati 2 Phase 1 carries a July LOI for 100 MW of critical IT load. HCW thinks the first AI data center lease lands within months.
"Receiving little to no credit from the market for its recent expansion into AI data center development."
Correct — because there's no signed lease yet. This is a pipeline story, not a revenue story. The base business is healing: Q2 revenue $15.1M (+145% YoY) but adjusted EBITDA still -$1.6M. Stock up 152% in a year, yet still 75% below the $5.14 high. That's the whole r/r — full optionality on 6.3 GW at a $1.31 price. If a lease prints, this thing re-rates violently. Until then, HCW carries the flag alone.
Boone at Citizens keeps GOOGL at Market Outperform with a $515 PT — street high-end — and it's all about YouTube becoming the default TV distribution layer, not a standalone app. THE FLYWHEEL IS SIMPLE: subscale streamers come to YouTube for audience access, each new partner makes the platform stickier, and single-app competitors can't break the loop. Stock's up 70% in a year with revenue growing 20% and 18 estimate revisions to the upside — the narrative has fundamental backing.
"The dynamic creates a flywheel effect that makes it harder for single apps to compete for default status."
The key reframe: YouTube is transitioning from streaming app to broader TV/content platform. That's the moat — aggregation of content plus discovery engine. Every new distribution deal (and there are more coming) makes the default status harder to dislodge.
Elsewhere on the tape: $25B senior notes offering closed (floating-rate tranches + $2B of 4.625% due 2029), Gemini 3.7 Flash launched for coding/automation — Pro still no timeline. French media complaining about AI-generated summaries eating traffic; Meta precedent means that one lingers, but it's not a bogy at this price.
Berenberg says the market got this one wrong. They upgraded MBLY to Buy from Hold, PT to $11.00 from $10.80, arguing the stock OVERCORRECTED after Amnon Shashua's resignation — despite a Q2 that beat by a mile. At $8.92, the tape is pricing low-single-digit revenue growth for a business Berenberg sees turning profitable this year ($0.48 EPS) with a product cycle inflection ahead.
"The stock is mispricing the mid-term growth prospects of the business."
The bull case is simple: automakers prioritizing cost/efficiency/scale plays into Mobileye's L2-L2+ lane, SuperVision commercial launches land in the next 12-18 months, and the vertical robotaxi announcement is free optionality. The bear case, per UBS ($9 PT) and Mizuho ($8), is the leadership transition and business model shift create real overhang. Canaccord cut to $12 on China competition and limited traction beyond VW.
Net: Q2 revenue $508M vs $475M consensus, FY26 guide ~$2.0B, EyeQ units raised to ~39M. The fundamentals are fine — this is a positioning call on sentiment washing out. Piper already went Overweight at $12; Tigress is at $18. Berenberg's the third firm stepping in front of the narrative. That's a cluster worth respecting at $8.92.
Stifel is the low-end of the new target cluster — reiterating Buy at $188 while UBS ($225), Benchmark ($230), TD Cowen ($210), and Needham ($200) have all repriced higher. Stock at $140.35, UP 182% OVER THE PAST YEAR. That's the tell: the fundamental story is intact, but the PT dispersion says either Stifel is late or the street is early. Momentum says the street wins.
The thesis is straightforward. 800G and 1.6T optical shipments are accelerating — Infra was 34% of April quarter revenue and Stifel thinks July quarter prints ABOVE their $328M estimate. October guidance has similar upside. Beyond AI: fourth-gen LoRa drives Industrial sequential growth, HighEnd Consumer keeps gaining share, and TVS billings data points are positive. This isn't a one-trick AI play, which is what gives the story durability.
"October quarter revenue guidance may also exceed expectations, with upside potentially coming from Infrastructure based on an expanding AI infrastructure cycle."
The margin angle is the sleeper. Gross margin sits at 52%, heading toward management's >60% target, and the pending cellular module divestiture to Compal clears the path. Thirteen analysts just revised FY estimates upward. The only real bear case is math: 10.8x CY27 EV/S on a stock that's already up 182% — the easy money got made. But with AI capex still compounding and a beat-and-raise machine in motion, fighting the tape here is a losing r/r.
DT is playing offense. The $915M Arize acquisition buys a seat in agent observability before the market fully forms, and DA Davidson's Buy/$65 re-up says the tape hasn't caught up — stock at $50.09, roughly 30% below the PT.
THE DEAL ADDS 200BPS TO ARR GROWTH. Against a quarter that already printed 41% organic net new ARR, that's acceleration on top of momentum. Scotiabank and Canaccord both lifted PTs after the FQ1 beat — $61 and $60 — so the Street cluster now reads $60-65.
"Observability requirements are rapidly evolving in response to AI."
That's the whole thesis. Traditional APM doesn't catch AI agent failures. DT is buying its way into the next cycle before Datadog owns it.
Bear case: $915M for an early market. Integration risk is real, TAM math is fuzzy. But with a $275M buyback last quarter and 18% revenue growth, this is a company buying from strength — not desperation.
DB throws cold water on the Mythos trade, cutting NTSK to Hold from Buy — despite a PT bump to $17 from $16. The stock's ~90% run off April lows has it at $23.11, NEAR THE $23.95 52-WEEK HIGH, and Zelnick says the market is pricing a demand inflection that isn't coming soon. The PT raise is a nod to momentum; the downgrade is a call on r/r at these levels.
Valuation is the bogey. NTSK trades only about ONE TURN CHEAPER than Zscaler on CY27 EV/Rev. That's a thin discount for a vendor whose core market is saturated and competitive, and whose own channel checks from the 51-participant reseller survey and Black Hat field work came back mixed. You're paying near-peak multiples for a story that still needs to prove itself.
"Netskope will likely need to show consecutive quarters of improving execution with sustained margin leverage to warrant further investor support."
That's the bar. Two more things on the watchlist: no permanent CFO yet, and SentinelOne getting downgraded alongside tells you this is a sector-valuation call, not a Netskope-specific breakdown. Position feels crowded. Wait for the CFO hire and next two prints before adding.
Bullish setup confirmed. TD Cowen raises PT to $400 from $360 after Black Hat 2026 channel checks — and the signal is stronger than the number. Citizens ($415), Evercore ($415), Tigress ($430) all stacking higher. Four firms, one narrative: AI security is the new budget line, and PANW owns the runtime layer.
Prisma AIRS is the hook. Management calls it the fastest-growing product in company history with clear line of sight to $100M ARR over the next few quarters. That's not a slideware promise — that's a monetization proof point. Enterprise conversations have shifted from "protect the network" to "secure our AI adoption." Runtime protection, identity, governance, autonomous defense. Platform vendors are winning the wallet.
"Palo Alto Networks faces a favorable setup that should allow it to guide above the fiscal 2027 revenue consensus of $13.8 billion."
That's the bet. 21% YoY growth is the bogey, and TD Cowen thinks PANW guides OVER it. Stock sits at $384.27, a rounding error from the $398.88 52-week high. Market cap $313B. Shares up 117% over the past year — the momentum is real, and platform consolidation keeps feeding it: organic platformization deals +32% YoY, +6% QoQ in Q3.
Bull case: AI security is incremental spend, not re-labeled budgets. PANW's platformization drives attach rates, and Prisma AIRS gives them the fastest-growing product in company history to sell into every enterprise AI conversation.
Bear case: this thing is up 117% in a year. The setup is good but the bar keeps rising. Nobody's pressing the short — but the question PMs should ask is whether the multiple already embeds the beat. At $313B, you're paying for perfection on the guide. The street is uniformly long, which makes the marginal buyer less obvious.
Light coverage, but the tape says it all: the street keeps reaching for higher PTs, and the AI security narrative gives them a reason. Favorable setup, well-telegraphed. The guide will matter more than the print.
BWS FINANCIAL RAISED THE PT TO $50 FROM $30 — A 67% JUMP IN THE TARGET, WITH THE STOCK AT $29.56 OFFERING ~69% UPSIDE FROM HERE. On a name already up 94% Y/Y and 67% YTD. Bold.
The thesis is hybrid bonding, pure and simple. Samsung just shipped new flash memory using hybrid bonding WITHOUT an Adeia license. SanDisk held an investor day flagging the same tech. Sounds like a negative at first glance — but BWS reads it as the opposite: wider adoption means more per-unit license signings as the tech standardizes. Adeia gets paid on volume, not on being first.
"Adeia should benefit from wider adoption as the company signs per-unit licenses."
The 2028 DRAM license renewal cycle is the forward catalyst. BWS thinks investor attention pivots to that timeline and the stock re-rates quickly. Near-term prints are messier — Q2 2026 EPS beat ($0.34 vs $0.31 est.) but revenue missed ($96.1M vs $96.79M est.). Mixed tape, but that's noise if hybrid bonding is the bridge to the renewal wave.
At a 0.55 PEG, bulls say the growth-adjusted valuation is cheap. Bears say it's cheap for a reason — IP licensing is binary, and the equity already ran 94% on narrative. BWS is comfortable with that r/r; the PT implies they think the license wins compound from here.
TD Cowen bumps ZS to $200 from $180, Buy maintained — 17 days from the FQ4 print, they see in-line or better. Stock sits at $183.60, so this is a modest step-up, not a barnburner. The bull case: security demand is healthy off the back of breaches, global conflict, and CISOs prioritizing AI. 41 ANALYSTS REVISED ESTIMATES UP, and the fundamentals back it — REVENUE +25% LTM, 77% GROSS MARGIN.
Z-Flex is the new large-deal accelerator. Options flow agrees — 39,437 CONTRACTS, majority calls.
Wolfe is the bear-ish counter: cut PT to $150 from $183 on valuation, still Outperform. That tells you this is a multiple debate, not a model debate. Not sure we read too much into one PT trim when the rest of the Street is moving estimates higher.
TD Cowen cited healthy security demand driven by global conflict, high-profile breaches, and CISOs' prioritization of AI.
Mizuho hosts Datadog's IR and FP&A brass, walks away reaffirming $300 PT. Stock sits at $255.46 — UP 100%+ IN A YEAR — and the bull case is simple: AI customers are hitting scale and feeding straight into the observability flywheel. Same camp across the street: Needham at $300, Macquarie at $260, Stifel Buy. (PT cluster now $260–300, consensus Buy.)
The numbers do the talking. 79.5% GROSS MARGIN, and Needham flags Q2 revenue at +36% YOY — a $42.9M beat. Macquarie counts five straight quarters of acceleration. Stifel's kicker: OPENAI SPENT LOW $100M RANGE with DDOG in Q2 alone. That's real revenue, not a keynote slide.
The bear case isn't competition — it's one customer. Management explicitly flagged a POTENTIAL USAGE DECLINE FROM ITS LARGEST CUSTOMER STARTING IN Q3. One tenant rolling over dents the growth narrative, and that overhang is exactly why the stock's not already at $300.
"Datadog continues to solidify its position as a leader in observability with multiple product areas at scale and newer products growing rapidly."
Net: the AI-native concentration fears are real but priced-in at these levels. The fifth consecutive acceleration quarter says the modernizations story has legs beyond one big spender.
Mizuho's the last bull standing, and even they're cutting — PT to $430 from $500, Outperform kept. The thesis: the Aug 25 print is when management resets FY27, and that reset is a clearing event, not a death spiral. Stock at $345.66, DOWN 47% YTD, ~13x FY27 non-GAAP P/E — trough valuation for a durable double-digit grower with real FCF.
The rest of the Street has already capitulated. TD Cowen to Hold ($304, later $328), Truist to Hold ($350), Morgan Stanley to Equalweight ($335) — all citing TurboTax worries and AI disruption. Mizuho's not dismissing those, but they're framing the setup differently: below-consensus FY27 guidance is already expected after intra-quarter commentary. The real question is September Investor Day.
"A credible reset followed by consistent execution is typically how companies like Intuit begin to re-rate."
Swing factors into FY27: low-end tax monetization, lighter QuickBooks pricing uplift, Credit Karma normalization. Mid-market and assisted tax growing >35% — that's the quality underneath the mess. Not a buy signal yet, but at 13x with a reset coming, the r/r is starting to look interesting. Watch the guide, then watch execution.
Dead-money wind-down, not a turnaround. Citizens keeps Market Underperform, PT $2.25, after Domo filed the 14C laying out the Progress Software asset sale. The $400M headline looks big, but that's GROSS for substantially all assets. Net to equity after assumed liabilities is the actual bogey. Shares off 53% YTD — the market already smells the dilution.
Progress Software walks away with the AI/data platform and customer contracts; Domo shareholders wait for the leftover. Citizens isn't buying the "strategic premium" narrative. For a PM, this is a skip. There's no catalyst off the deal — just a messy liquidation path with a low PT as the anchor.
Stifel lifts PT to $275 from $230 (Buy). Pure beat-and-raise setup. Firm models Q2 subscription growth at 15-16% vs guidance of 14-15%, and sees H2 guidance (currently 13-14%) ticking up ~1pt with Copli and FX tailwinds. STOCK'S ALREADY UP 37% SINCE THE JUNE PRINT vs IGV +4% — the market's front-running this exact scenario.
Real work behind the call: Stifel spoke with three large pharma customers on the Veeva vs Salesforce CRM dynamics, AI strategies, and vendor consolidation. Conclusion: defensibility is improving, AI is the wedge. The Copli acquisition / Falcon MLR launch automates medical-legal-regulatory review — 70% of manual labor gone over five years, per Veeva. That's the narrative keeping multi-year software multiples sticky through the AI digestion phase.
Valuation is the fault line. 21x/31x FCF with/without SBC, ~44x P/E. Stifel calls it appropriate. Guggenheim ($232) and Oppenheimer ($225) are less aggressive but same direction. Stifel's got the momentum and the customer checks — that's the number to watch into the July print.
"The re-rating suggests the stock is likely pricing in a typical modest beat-and-raise quarter."
H.C. Wainwright opens coverage at Neutral, $0.25 PT — ~47% above the $0.17 print. That rating-vs-PT spread IS the thesis: intriguing asset, unproven scale, not enough capital to brute-force it. HCW says ALPHA-01's deployment is already in the price. This is a "show me" initiation, not a conviction call.
The bull side: ALP deployed ALPHA-01 in May off an initial offtake from an undisclosed AI Frontier lab and claims a $200M ARR PIPELINE from the AI infrastructure supply/demand imbalance. The bear side: FY26 NET LOSS OF $38.6M (more than half from legacy biotech/digital-asset one-timers), cash only $10.3M at year-end, and confidential compute is a niche hyperscalers can crush anytime. GPU procurement, financeability, and data center access are the real bogeys.
"Wainwright said it will wait for greater visibility into future offtake agreements, GPU procurement and financeability, and access to scalable data center infrastructure before becoming more constructive on the shares."
New President Enzo Villani (already Exec Chairman/CIO) takes on strategy and capital allocation — the right role for the raise-and-build grind ahead. Next catalysts: another offtake, any financing, proof the pipeline converts. At $0.17 with ~$10M cash, this is binary. Wait for a better entry or a real catalyst.
Cantor cuts PT to $235 from $320, stays Neutral after a CFO call post-FQ3. Stock sits at $211.20 — still up 140% over a year, but 35% off the $328 high. The move isn't about the quarter; it's about the growth algorithm past the current $2.4B backlog.
The quarter itself: revenue $312M, EPS $1.42 — both barely missed (~1.5% on revenue) despite 9% YoY growth. But orders hit a record and backlog hit new highs. The real signal is capacity. Management says the leased footprint already committed (50k sq ft Houston + Ohio) is ENOUGH to execute the existing backlog. That 275-300k sq ft facility they're finalizing? That's for growth beyond the backlog, NOT a near-term constraint.
"This represents a de-risking compared to its May call with the company, when capacity remained the primary gating factor."
Bull case: the de-risking removes the "can't build it" discount. Backlog is executable, cash flow is strong, and record orders mean the next leg isn't dead. Bear case: Neutral rating + PT to $235 says the market already paid for the 140% run — and the next facility isn't online yet, so growth beyond backlog is a 2027+ story. r/r is okay, not great. You're paying for execution that the market now mostly believes.
DA DAVIDSON BUMPS PT TO $42 FROM $34 BUT STAYS NEUTRAL — AND THAT'S THE WHOLE STORY. Stock's already up 62% in six months trading at $36.15. The AI narrative is getting real, but the firm's not ready to chase it.
Analyst Lucky Schreiner met with CFO Serge Tanjga on Wednesday and came away "incrementally constructive" on positioning and the moat. AI opportunity is becoming more tangible — Q2 reinforced that: EPS $0.13 vs $0.02 expected, revenue +19% to $203.3M, cloud sub revenue +23% YoY, FASTEST GROWTH IN 2+ YEARS. 85% of new customers bringing AI capabilities. Raised full-year guidance. That's a strong tape.
But Neutral is Neutral. The firm wants evidence that these emerging drivers turn into sustained growth acceleration, not just a beat-and-raise quarter. Fair — the stock's already re-rated hard.
"It remains neutral as it waits for additional evidence that emerging growth drivers will translate into sustained growth acceleration."
Bull case: execution clearly improving, AI attach is real, public-sector pipeline is robust. Bear case: the re-rating is done — at $36 you're paying for the acceleration to persist, and DA Davidson isn't willing to sign that check yet. The $42 PT just formalizes where the stock already is. Risk/reward is balanced at best.
UBS re-upped Buy at $210 PT with the stock at $140 — 38% below the $225.64 high. Conviction from the bottom, not the top. Hodulik's framing: SpaceX is now a frontier-model/cloud provider, and the next Starlink leg is femtocells embedded in V3 terminals to cover indoor spaces. THE GROWTH INFLECTION IS NEXT YEAR, when V3 fixed-broadband satellites hit critical mass.
The fundamental scoreboard backs the bull: $23B LTM revenue, 138% growth forecast this year, Q2 AT $7.8B (+92% YOY), AI cloud services up 247% to $1.6B. Anysphere ($60B all-stock) folds the AI layer in. MS at $300 Overweight, Mizuho Outperform on Grok 4.6.
This is the real swing factor. The V2 mobile LEO constellation can't reach dense locations and office interiors — the femtocells in next-gen Starlink terminals solve that. But two keys: low-band spectrum below 1GHz (T-Mobile's 800MHz sold to Grain, EchoStar's 700 E Block, NextNav/Anterix 900MHz) and deployment density.
"The femtocells would radiate spaces that the V2 mobile LEO constellation cannot reach, including dense locations and office building interiors."
The catch: ~30% of Americans live in multi-dwelling units — precisely the dense indoor market this targets. UBS models US subs going 3M → 6M by end-2027 → 20M by end-2031.
The $75-$300 PT spread is a 4x gap — the market pays for optionality, not visibility. Phillip Securities is the Sell at $75, calling revenue durability into question. With AI revenue booked through Cloud Services Agreements and satellites still scaling, the bear case is multiple compression on lumpy prints. Fair debate. Not a coin flip — more like 60/40 the bulls hold unless the low-band spectrum layer trips.
Cantor keeps Overweight and lifts the PT to $550 from $510 — two days before the August 19 print. The call: October (fiscal Q4) comes in at $4.25B revenue / $3.75 EPS vs $4.07B / $3.54 consensus — a beat-and-raise that would mark the 10th straight quarter of above-seasonal growth. This isn't a semi-cycle recovery story anymore; it's an AI story.
AI-related revenue ran ~19% of mix last quarter, LTM growth near 30%. And the comp bar is low: analog peers guided Q3 to 6-8% sequential growth vs ADI's typical low-single-digit seasonality. Cantor sees GM at 73%+ but concedes upside is capped at best-in-class levels. DA Davidson echoes the bull case — Buy, $498 PT, Q3 FY26 revenue above $3.90B (up 7.6% QoQ) on ATE and AI data center demand.
"The firm believes there is upside from both a multiple and earnings per share perspective as investors shift focus to calendar year 2028, where earnings per share upside is closer to $20."
Options market prices a 5.8% move on the print. ADI has beaten the implied move in 6 of the last 8 earnings reports. Bias is clear.
The take: Top line wins, bottom line loses, nobody cares. SE printed Q2 REVENUE +48% YOY TO $7.8B against a $7.09B consensus bogey. ADJ EPS $0.70 MISSED $0.83 — the blemish on the print. The market has decided the growth spend is worth it (the EPS miss is a feature of the investment cycle, not a bug).
Three shops, one narrative: growth justifies the investment. Presidents Capital reiterated Strong Buy / $170 PT. Benchmark raised to $175 from $140. Morgan Stanley to $153 from $130. At $121.94, that PT cluster implies 25-43% upside. The whole thesis boils down to: SHOPEE +49% AND MONEE +59% ARE THE ENGINE — and S&M +52% / credit provisions +65% is the cost of running it.
Morgan Stanley: "strategic focus on growth investment while maintaining financial discipline as a positive factor."
That's the bull case in one line. Spend hard on winners, keep the balance sheet clean. r/r skews positive here.
Garena bookings +16% YoY — decelerating. Not the story today. H1 REVENUE 52% OF FY ESTIMATES, PATMI 48%. Setup intact.
KeyBanc still screaming UNDERWEIGHT at $306 — $250 PT implies ~18% downside, and they're not budging. The entire call is multiple: 34x P/E VS 28x THREE-YEAR AVERAGE, and they think the Street is paying up for price-led growth that won't compound. The firm's own card data shows spending +9% m/m (vs 2% avg) — they read it as neutral, so don't chase that as a demand signal.
The bear logic is clean: higher unit prices slow user growth, Services growth follows, and the multiple compresses as investors realize the mix shift. Not sure we can wave that away — a $4.48T MARKET CAP leaves zero room for multiple expansion, only earnings delivery.
"Investors will pay a lower multiple for price-led growth compared to volume-led growth."
Two sidebars: Siri AI content licensing talks with publishers (per WSJ) give the bull camp a 2027 hook, but it's early innings. And Jefferies just downgraded on the all-glass iPhone death — poor yields killed a 2027 launch. China share ticked up in June on promo-driven +62% sales, but 1H shipments STILL -7%. Lots of noise, little margin of safety.
Seaport calls this the REZI setup all over again — and they're buyers. Initiated ADIG at BUY, PT $35 (stock at $25.82, so ~36% upside). Thesis: ADIG printed a record quarter but guided toward near-term margin weakness on non-core items. Same playbook REZI just ran. Market sold both. Seaport viewed the REZI drop as an overreaction, and ADIG gets the same framework.
The Q2 print: REVENUE $1.29B, +1% YoY, ALL-TIME RECORD. Commercial demand stayed robust. Prior market share losses largely recaptured. Average daily sales +2%. The warts: gross margin 22.3% (thin), weaker cash flow, cost headwinds near-term. Both names disclosed flaws while waving at proof points of the core strategy.
"Both companies disclosed imperfections in their business while pointing to proof points of their core strategies."
If you buy the pattern, the margin guide-down is noise and the strategic proof points carry the multiple. $35 says the market over-indexed on the miss.
Piper Sandler bumps 3Q ad spend estimates +230bps on July strength. Market-wide July ad growth ran 130bps ahead of plan, and Reddit led the positive revision pack alongside ROKU and NFLX. Bull case writes itself: RDDT is taking share in a tape that's broadly healthy.
The headline that'll spook PMs first — Reddit's ads manager overhaul cuts audience size estimates ~20% in early September. Ignore it. That's a model forecasting fix, not a user demand problem. The underlying trends argue the opposite: end-July users ran +5% vs the start of the month, then grew another 7% in the first two weeks of August.
Only blemish in the print: average July users dipped 2% MoM. Feed added 6M while Conversation bled 26M. That's engagement mix rotating toward the monetizable surface, not churn.
"The changes are a function of model forecasting improvements, not underlying user trends."
Translation: better measurement, same users. If the September recalibration knocks the stock, that's the buy window.
Roll-up story works. Until it doesn't. Bernstein bumps BSP to $47 (from $40), Benchmark sits at $50, Mizuho's way out at $72 — three buys, same thesis: the M&A machine prints, and the market keeps paying up for the next deal.
Q2 was a bomb: ADJ EPS $0.46 vs $0.27 EXPECTED, REVENUE $704M, +126% YoY. Trailing revenue $1.65B at 67% GROSS MARGIN. That's not a software company anymore. That's a holding company with an app-studio brain — AOL, Airtable, Tractive, plus product work inside Eventbrite and Vimeo.
The catch: ORGANIC GROWTH SLOWED TO 3% FROM 6%. That's the number that decides this stock. As long as M&A covers the gap and organic stays positive, the narrative holds. The moment that breaks, the multiple compresses fast.
Bernstein frames it best:
"Bending Spoons needs these deals to work to broaden the TAM while keeping that organic revenue growth in positive territory to capture and keep an easily distracted investor engaged. Bending the definition of what an Internet company looks like in 2026."
Trades $41.63. Top Internet stock YTD, and the momentum tape is clearly behind it. I'd want organic to reaccelerate before chasing — but this is a show-me story that keeps showing. 3p is the line in the sand.
THE DOWNGRADE IS A VALUATION CALL, NOT A FUNDAMENTAL ONE. Phillip Securities cut SHOP to Accumulate from Buy (PT $170 from $160) even as they raised FY26 revenue and profit estimates. That's the tell — the business is fine, the stock just ran ahead of the tape. Market cap at ~$199B, LTM revenue growth of 32.5%, and the market is paying up for it.
The Q2 print was clean. Revenue +34% YoY with merchant solutions +37% and subscription +23%. GMV beat, Payments continues to be the engine. H1 revenue and adjusted profit already at 47% and 40% of full-year forecasts — they're tracking well ahead. The one blemish: margins down 90bps YoI on the mix shift to lower-margin merchant solutions plus rising AI/LLM costs. That's the tax you pay for the agentic commerce optionality.
The Street's collectively constructive but not euphoric — PTs cluster $145-$185 (JPMorgan highest at $185, Cantor and UBS at $145 with a Neutral). Benchmark at $170. Nobody's running for the exits.
Phillip Securities: "Shopify continues to deliver robust growth while remaining well positioned to capitalize on the rise of agentic commerce."
Still, an Accumulate after a beat-and-raise tells you the easy money's been made near-term. Long-term story intact, but the r/r at these levels is mediocre. Better entry points will come.
BMO re-ups Outperform/$135, but the stock sits at $78.22 — DOWN 36% OVER THE PAST YEAR. The bull case lives or dies on ads, and even the bulls are admitting there's no near-term catalyst in UCAN.
The firm hosted Netflix's former head of product innovation, and the readout is mixed. Ads = clearest path to UCAN reacceleration, still early innings. AI could strip 30-40% off content development costs at scale. But YouTube keeps tightening the screws, and Netflix has the wrong mental model on short-form.
"Netflix appears to lack a clear near-term catalyst to help reaccelerate growth in the United States and Canada region."
That's the bear case in one line — straight from an Outperform analyst. Street's fracturing: UBS at $115 Buy, Phillip Securities upgraded to Buy at $110, Rosenblatt at $75 Neutral on Q3'26 revenue slowdown fears. 21 ANALYSTS CUT FORWARD ESTIMATES. Add the bond deal (notes due 2036, refinancing ~$1B maturing debt) — that's balance sheet hygiene, not a growth signal.
Backdrop: Q2 revenue missed UBS by a hair, margins held. GTA VI exclusive content with Take-Two is a nice content hook but not a business model. Net: ads is the only re-rating catalyst, and it's not moving the needle yet. Waiting for ad-tier inflection or the AI cost story to show up in guidance.
TD Cowen holds the line — Buy, $460 PT — but the real signal isn't the rating, it's the AV narrative. Stock sits at $349.28, DOWN 24% YTD, nearly 19% over six months. This is a "ignore the tape, we're playing the 2028 option" reiteration.
The bull case, per AV expert Alex Roy on their Internet Bus Tour: Waymo leads on scale and safety, but the hardware is a problem. That lidar/camera rig is too bulky to ever appeal to car buyers, which caps Waymo's licensing runway to OEMs. Tesla's path is pure software — improve the stack's safety, and let younger consumers do the math:
Roy views Tesla’s efforts as focused on improving the safety level of its software stack. He thinks Waymo and Tesla will likely intersect at some point as younger generation consumers will likely view Tesla’s safety level as good enough, even if it’s not as safe as Waymo.
That's the whole ballgame. Tesla doesn't need to be the safest AV — it needs to be good enough and massively cheaper to scale. Winner-take-most in the US, Waymo and Tesla the two most likely survivors. Zoox, Lucid-Nuro-Uber, Wayve, and Mobileye are the long-shot tracker positions.
But here's the counterweight, and it's in the same note: EU registrations are horrid. Portugal July registrations -68.7% YoY (89 units). Italy -77% (105 units, 0.09% market share). This is NOT a macro story — Jefferies has EU-5 registrations +5.2% YoY in July, and Portugal's light EV market grew 54.3%. Tesla is losing share in a growing EV market. That's company-specific demand destruction, and it's happening now, not in 2028.
Bottom line: TD Cowen is renting Roy's credibility to justify the PT. The AV thesis is real but unprovable this quarter. The EU registration data is real and provable right now. Net-net: positioning call for the long-dated book, but PMs should fade any strength into earnings until delivery numbers stop bleeding.
Wells Fargo gets off the fence — Equal Weight → Overweight, PT $180 (stock at $147.43, basically hugging the $157 52-week high after a 79% six-month rip). Richard Poland's field work convinced him low-teens growth is the base case, not the bull case. The AI thesis: it hits the business model indirectly long before you see it in the P&L. By then, the trade's gone.
Identity is the #2 SPEND PRIORITY BEHIND AI — up from #4 last quarter. Okta's #1 in identity share gains, ahead of Microsoft. Channel data is strong: 47% net above plan, pipeline 12% above expectations. Revenue +12%, gross margins 77% — the financial foundation supports the narrative.
"By the time direct AI contribution becomes obvious, it may be too late for investors to capitalize on the opportunity."
Citizens, KeyBanc, and Scotiabank all recently landed on the same side of the trade (PTs $135-175), leaning on AI security, the ID-JAG standard, and the Anthropic alignment. The Street's converging — and the stock's already pricing a good chunk of it. R/r still works if you believe the indirect AI kicker is real.
The 28% plunge is warranted. ARR missed, guide cut — the market did its job. The real debate: government deal delays are timing, not losses. That's the bull case. The bear case starts with the THIRD LEADERSHIP CHANGE IN UNDER TWO YEARS.
Three analysts cut targets to $12.50-$15 from a $15-$23 cluster. All three kept Buy. The tell: nobody's calling the equity story broken, just the near-term numbers.
The headline sin: Q2 ARR at $507.8M, +16% ORGANIC, missing the $512.1M consensus AND the low end of guidance. Management blamed a handful of large US federal and European government deals slipping. Classic lumpy government motion — a few big bogeys slide and the whole quarter caves. CY26 organic growth guide cut to ~15.5% from ~18.5%. That's a 3-point haircut in one print.
EPS beat though ($0.11 vs $0.07 est). Don't over-read it — that's cost control, not revenue health.
DA Davidson's verdict captures the setup:
"The beating shares are taking today is warranted, though we find shares attractive on reset numbers & valuation."
The bear case: Shiven Ramji walks in as new CEO — the third leadership change in under two years. Leadership churn compounds the visibility problem. If 15.5% is the real growth rate, the multiple deserves a haircut too.
The bull case: deals slide, they don't die. RSI flags oversold. At $10.95, YTD -40%, the stock has already priced the reset. All three Buy ratings say the risk/reward flipped positive at these levels.
Freedom Broker upgrades SERV to Buy with an $8 PT (cut from $18) — the day after the market gutted the stock. That's the tell. They're not defending the quarter; they're buying the balance sheet and the DoorDash option.
Q2 revenue printed $3.2M, +404% YoY, +9% QoQ. Recurring revenue crossed 50% of sales for the first time — DoorDash carrying the mix. Everything else horrid: daily active robots FELL to 792 from 812, daily supply hours dropped to 9,809 from 10,295, gross margin NEGATIVE 271%, net loss widened to $64.1M. Management slashed FY26 guidance to $9-10M from ~$26M on weak Uber Eats volumes — the first Uber-linked delivery decline in 17 quarters. Uber dumped its entire stake in Q2; partnership dead over deployment strategy disagreements.
$240M cash, zero debt. At $5.04, down 52% YTD and hovering near the $4.32 low, the market prices in near-zero equity value. The $8 PT implies ~60% upside — but that's just 8x a FY2028 revenue guess plus cash.
"Freedom Broker's $8.00 price target is based on a blended 8.0x EV/Sales multiple applied to fiscal 2028 revenue forecast plus net cash."
Guggenheim and Oppenheimer both landed at $7 (from $13 and $20, respectively) — everyone converges on the same range. Bull case: $240M cash buys time, DoorDash grows, $10M guide is the floor. Bear case: Uber's gone, utilization falling, margins catastrophic — this is a melting ice cube. Freedom's upgrade says the cash cushion matters more than the operational noise right now. At $5, that's a reasonable debate.
JPMorgan just quit on PLDT. Downgrade to Neutral from Overweight, PT SLASHED to PHP1,050 from PHP1,800 — a 42% haircut. The ARPU reflation thesis in a consolidating wireless market never showed up.
WIRELESS ARPU STUCK AT PHP135 in Q2. Fixed broadband still deflating. No repricing power = no adequate return on capital, and that's a real problem when D/E is 2.71 and the current ratio is 0.44. FY27/FY28 EPS cuts of 19%/28% map the trajectory.
"Competitive pressures limit its ability to reflate ARPUs, which is required to earn adequate returns on capital."
The bull case: 8.9x P/E, 5.6% yield, 52% EBITDA margin, and a capex reduction that already improved free cash flow. The bear case: FCF still doesn't cover the dividend unless capex stays low, and PHI MAY HAVE LOST WIRELESS SHARE in H1 — you don't reflate ARPUs from a position of weakness. Management's H1 print was steady, but steady isn't a catalyst. JPM stepping aside is the right read.
Northland upgrades to Outperform, PT $350 — stock at $321.61, so ~9% upside. That's the conservative end of the Street. Needham's already at $425. The gap tells you the debate: this thing has run 160% in six months and the PTs are still chasing.
Northland's thesis is simple: the AI infrastructure spend peak keeps sliding. They cite $200B in new funding for Anthropic (Google, Broadcom, Apollo, Blackstone, Morgan Stanley — plus crypto miners, which is a wild sentence) and $500B in private capital behind Nvidia last week. That pushes the peak into 2027 and extends the cycle. 20 analysts have revised FY estimates up. The rate of change is the story.
"Astera Labs is well-positioned in signal conditioning for AI infrastructure and will continue to be one of the fastest-growing AI infrastructure chip companies."
That's the moat — signal conditioning is ugly, underappreciated work, and ALAB owns it. Content per rack keeps climbing as they take switching fabric share in both scale-up and scale-out. The Q2 print backs the narrative: $0.80 adj EPS vs $0.69 est (+15.9% beat), revenue $392.4M, +104% YoY. Needham's call hinges on Scorpio X switches ramping hard in 2H26. If that delivers, $425 is the target, not the ceiling. Risk/reward still works at this level, but you're paying for execution on the second half ramp — not the beat that's already in the tape.
UBS is putting a floor under the name into the print — Buy reiterated, $280 PT vs. $226.16 spot. That's ~24% upside, but the target isn't the story. The beat size is.
UBS models July quarter revenue at $94-95B, which is $3-4B ABOVE THE COMPANY'S GUIDE. Same playbook as every quarter — guide low, smash, raise. The real question is October. UBS sees the guide at $107-108B with actuals clearing $110B as Rubin sell-in accelerates toward 500K GPU units/month while Blackwell winds down. That's a $14-15B sequential jump. Feels aggressive until you remember the last year.
Compute supply growth continues to lag demand, partly because the majority of hyperscaler capital expenditure increases this year stem from memory price inflation.
That line is the whole ballgame. Hyperscalers are spending more, but a huge chunk of that incremental capex is getting eaten by memory price inflation — NOT compute. So GPU supply falls further behind demand, backlog builds, and NVDA has pricing power for longer. UBS expects the call to acknowledge another significant backlog increase. Watch that.
The bull case isn't complicated: demand outstrips supply, Rubin transition is on track, and the model gets you to $15 EPS in CY27 and $20 in CY28. At $226, that's 15x CY27 — cheap for 70%+ grower. UBS is effectively saying the derating from the $1T peak is overdone.
The bear case isn't about demand — it's about the financing machine. The $500B AI infrastructure partnership (Apollo, BlackRock, Blackstone, Brookfield, GS, KKR) is a massive demand accelerator but Wolfe flags the long-term risk. NVDA is increasingly rent-seeking into the capital stack, not just selling chips. Bull says it locks in backlog; bear says it concentrates systemic risk and eventually the market will price that in. (Not a this-quarter problem, but the LT risk conversation is starting.)
Chardan boots coverage with a Buy and $7.15 PT — ~80% upside from $3.98. Thesis: ODYS' vision-based predictive maintenance platform flips from pilot programs to production deployments across aerospace/defense, backlog compounds over the next 12-24 months, and revenue hits $50M within five years.
That's a helluva hockey stick off a $1.09M LTM base that's DOWN 78% YOY. The PT math is 5.0x EV/Sales on that $50M, discounted back at 15%. Chardan assigns ZERO value to transportation/industrial/light industrial — those sit as free options on top of the base case.
Bull path exists: the Elbit PO (~$0.3M, on behalf of Israeli MoD) is small but validates the government channel. Benchmark still says Buy — PT cut to $6 from $10 on valuation, not on thesis. Their framing:
"unique approach in the field of condition-based monitoring."
Bear case is the income statement — revenue bleeding out and the Street's modeling more decline this year. COO just got 40k options at $5.09 (call it retention, not a signal). This is a land-grab story on a rounding-error revenue base — size it accordingly.
TSM — Arizona flipped to a 1H26 profit of NT$31.15B vs NT$4.52B in 2025, and JASM turned positive. The overseas-fab dilution bear case is breaking on physical evidence. That removes the geographic discount capping the multiple. Word is 2nm lands FOUR QUARTERS EARLY — P2 equipment installing 2H26, mass production 2H27. Ellwanger's $45M August sale was pre-arranged in March. Noise, not signal.
LRCX — FY26 purchase obligations hit an ATH, +43% vs +26% revenue. Lam is locking in supply for a multi-year WFE boom. The bottleneck narrative flipped from cleanrooms to semicap components. Memory capex is concentrated in dry/plasma etch (SK Hynix M15X, Samsung P4), and HBM adds TSV etch steps. Lam is the dominant etch-content beneficiary.
INTC — Bernstein models 2028 EMIB output, CoWoS-equivalent, as greater than ALL 2025 CoWoS and ~1/3 market share. Strongest quantified case for Intel as a real advanced-packaging third supplier. $20B raise (~5% dilution) clears the funding overhang; +178% YTD means the entry point is the problem. XBM BEOL DRAM uses planar TFTs that won't scale below ~40nm — long-shot patent bet, not a technical advantage. Don't pay up for the stacked-memory talk.
MU — Morgan Stanley raised the revenue trajectory to $43.2B → $70.0B with no plateau. Their phrase: "HOLY STRUCTURAL SHIT." DDR4 contracts +50% QoQ in Q3, DRAM +8% MoM, SLC NAND +50% in both 3Q and 4Q. CBO says customers who skipped LTAs "may not get much allocation." That's supply discipline as a weapon. Tactical buyers are getting squeezed.
SSNLF — ISSCC paper analysis says Samsung overtook SK hynix in HBM4 faster than consensus expects. China sales +208% YoY to KRW 88.6T, overtaking the US; DRAM share to 39.4% from 34.0%. BUT Nvidia is not in the top five customers. HBM4 for Rubin is selling; the volume isn't in the customer list yet.
HXSCL — Flip side of that ISSCC read: market may be under-pricing HBM4 share loss. The KRW 43.2T H1 memory equipment spend (+35.1% YoY) keeps the capex cycle supportive. M15X ramp drives etch demand. Korean HBM increasingly routed to Malaysia for packaging — matters for INTC EMIB vs TSM CoWoS flows.
CXMT — Overtook Tencent as China's most valuable company, $500B+ cap, +500% since IPO. Apple is testing CXMT memory for iPhone/Mac. Caveat: DDR5 die ~40% larger, cost per bit >30% above leaders. Policy buy, not cost-led. A domestic 48GB DDR5 RDIMM launch shows the ecosystem forming end to end.
SMICY — Utilization 93.7% in Q2. China foundry is full. AI demand leaves no slack — pricing power and capex urgency into 2H26.
HHUSY — Utilization 102.8%. That's ABOVE full capacity. China foundry tightness extends to mature/specialty nodes. AI is flooding the whole domestic chain.
KIOXY / PSONF — Phison CEO says 2027 NAND tightness exceeds 2026. That extends the shortage narrative another year. Controllers/modules get a once-in-a-lifetime pricing window. Kioxia is the direct NAND supplier expression; Phison the controller-module play.
KLAC — Direct read-through from AMAT's record quarter with management signaling >30% YoY upside. EUV stochasticity keeps process control intensity high. Inspection is the toll booth on every WFE dollar.
ASML — Stock follows FCF/share; ~€60/share expected in 2028. Current multiple already embeds a strong 2027 recovery. Stochasticity sustains metrology/multi-patterning intensity, not necessarily unit upside. Trading range, not a directional call.
TOELY — Yttrium oxide tightening is a cost/delivery risk on Japanese semicap components. JPM's pricing-power diffusion gives the demand tailwind; rare earths give the supply drag. No public data splits the net effect. Watch procurement commentary — that's the tell.
ULVCF — Guidance miss drove the worst drop in seven years, blamed on production constraints. Delivery miss, not demand miss. If pricing power is diffusing across semicap, part of that drop is recoverable. The next quarter's delivery recovery is the proof.
AJINY — Cutting ABF substrate supply to Chinese customers ~30% under a capacity pretext. De facto export control squeeze on China advanced packaging. Bullish global ABF pricing; a real constraint for mainland substrate buyers.
AMKR / MDTKF — Amkor Korea is helping Intel scale EMIB. MediaTek's TPU is the first high-volume EMIB-T customer in 2028E. That validates Intel's packaging roadmap and gives MediaTek a CoWoS alternative. EMIB is now a two-factory story.
ARM — CEO says all major AI infra players are moving toward Arm-based CPUs. Architectural share confirmation. Royalties scale with AI server unit growth.
COHR — Dropped -14% last week after a +71% bounce, still +77% YTD. Sharpest positioning reversal in the optics group. Demand intact; the move overshot. InP substrate prices jumping >10% in Q4 — biggest hike in history — is a margin tailwind.
LITE — InP shortage is margin-accretive for CW laser makers. Mizuho sees Spectrum-X ramping; Nvidia buying CW/EML capacity with LTAs. Pre-booked demand visibility across 2027. The Nvidia photonics thesis is volume-backed now.
AAOI — InP substrate shortage with record Q4 hikes is a direct margin positive. Supplier says "even with cash, you might not get anything." At +475% YTD the easy double is over. Next leg needs production delivery, not pricing headlines.
AXTI — Purest InP substrate exposure. Prices up a third straight quarter, heading into a fourth with the biggest hike in history. Direct revenue and margin upside. But +3,844% YTD makes it the most crowded trade in photonics. Any supply disappointment hits hard.
CRDO — Mizuho flags VR200 NVL72 ramps and NPO/CPO connectivity as 2027E upside. AI rack connectivity content rising. High-beta way to play the optics transition.
MRVL — TIA/Driver exposure in Nvidia's CPO switch chain. Optical content per switch scales with CPO volume. Volume is the driver, and it's still early.
TSEM — NPO PIC exposure. Nvidia CPO and Trainium NPO configs need photonic ICs. Specialty foundry beneficiary sitting quietly in the middle of the optics supercycle.
CSCO — Beat-and-raised, fell -8% for the week, still +45% YTD. Momentum flush, not a thesis break. Telco orders +30%, three scale-across design wins, >$1B Acacia orders. And it's lowering prices — direct negative for NOK and CIEN.
CIEN — Cisco's price aggression is a share threat. AI optical demand is real, but Ciena needs transport wins to offset. Competition just got cheaper.
NOK — Exiting China: Hangzhou R&D closing, ~1,600 layoffs, headcount 13,700 → 7,200, revenue €2.2B → €913M. 2026 restructuring charges raised €250M → €800M. Cisco price cuts land on a company that's 73% telco. Bad combination.
AVGO — Mizuho sees VR200 NVL72 ramps starting 4Q26E led by xAI and Meta, CoWoS +75% in 2027, CoPoS volume in 2028E. CSP RPO/backlog is $2.3T, up 3.5x y/y. ASIC and networking content on every rack.
DELL — VR200 NVL72 ramp strength is 2027E upside. AI server backlog is the swing factor. Server pricing power remains the debate.
AMZN — AWS projected to consume 5M Trainium 4 units in 2H27 and 12M in 2028, two of three configs adopting NPO. Merchant-scale accelerator buyer and a volume trigger for optics. MS sees a path to $1T revenue in 8-10 years and ~$500B EBIT. AWS at ~$170B annualized, adding 6-8GW compute in 2026/27.
MSFT — 20-year deal backing Constellation's 835MW Crane nuclear restart. Power is the moat. Public cloud revenue at the big three accelerated 23% → 35% → 43% with operating margins expanding from 34% to 39%. That funds the capex cycle without P&L damage. China retreat is 1.5% of revenue — noise.
META — US ARPU +31% YoY to $125, US revenue/hour +27% to $1.33. AI engagement is monetizing. Sopaipilla insurance gap — $14B build, ~$450M property coverage — is a lender credit event waiting to happen. Custom CPU with AMD chatter; MTIA acceleration in 2027 is the margin story.
CRWV — Customer concentration collapsing: Microsoft was 71% a year ago; top three now 36/26/10. Capacity 0.47GW → 1.5GW. Physical AI ~1% of backlog with Caterpillar, Isomorphic, Flow, IMC, Leidos. If the Nvidia backstop headline sells the stock, treat it as a positioning artifact. Demand evidence is intact.
NBIS — Neocloud group may sell off on Nvidia's backstop structure shift. Evidence says positioning artifact, not demand signal. Demand intact.
ORCL — Only ORCL, CRWV, and xAI have built >1GW outside the big four. Oracle has scale but lacks the same secured power pipeline. That makes its AI infra less financeable than AMZN/MSFT/GOOGL/META. Scale without power security is a discount.
VRT — Hyperscalers own ~75% of incremental GW coming online 2026-2028. Cooling and power spend follows. Archetype AI retrofit compounder — when bought at depressed run-rate multiples.
STRL / PWR — PJM capacity costs +262% in 2025; $23.1B datacenter-driven capacity cost increase across three auctions. Grid hardening is decades-long spend. STRL direct beneficiary; PWR scale wins the biggest contracts.
CEG / NEE / ETR — Microsoft 20-year on Crane 835MW. Google 25-year on Duane Arnold 615MW. Entergy's Meta deal covers 2.26GW gas plus transmission, with 5.2GW pending expansion. Nuclear restart risk converting into contracted annuities. Entergy is the template for bilateral AI load.
GEV — Two independent voices say the gas turbine bottleneck clears in three months. If right, the multi-year turbine scarcity thesis is wrong. Don't assume the bottleneck is consensus-safe.
BLK — Sopaipilla JV insurance gap: $14B gigawatt-scale build, $450M property coverage, $645M terrorism coverage. Lenders exposed to billions uninsured. An underinsured gigawatt is a credit event waiting to happen.
TT / ETN — Nvidia DSX-aligned architecture claims up to 15% better efficiency, 80% less copper, 30% lower install costs. Real content-per-rack stories for electrical and thermal infrastructure.
PSTG — Second hyperscaler deal, language "Top 5 hyperscaler," suspected Oracle. The first tease went nowhere for 10 months. Second proof point is the signal.
BABA — Qwen downloads hit 3B in half a year, #1. Buy-side calling it a China AI tipping point; global funds rotating into China AI infra. Mechanism plausible, but downloads aren't production usage. The 35-country US ultimatum on China's AI bloc is the biggest single policy threat to that rotation. July macro was weak across the board — that caps the narrative.
TCEHY — President says compute ordered months ago can now be resold at +30% profit. Real, if opaque, asset value signal. But CXMT overtook Tencent as China's most valuable company, and hardware weight in MSCI China surpassed software. Down ~27% YTD. The market is paying for hardware, not platforms.
XIACY — Down >33% YTD. Consumer hardware is the crowded short end of the China AI rotation. Money rotating into semis/optical.
HNHAF — RMB 8,800 signing bonuses at Zhengzhou for foldable iPhone/iPhone 18. Build cycle real and pulled forward. Positive read to H2 revenue.
AMD — MI500 called the largest generational leap in Instinct history — inference up 2,000x in four years. Direct attack on Nvidia's inference share. Taalas acquisition hardwires model weights into silicon: no HBM, no advanced packaging, no liquid cooling. But model changes require new silicon. Meta custom CPU rumor keeps muddying the foundry narrative.
OABI — Lilly deal up to $370M in milestones plus tiered royalties; 2026 cash guidance raised to $49-53M from $37-41M. Non-dilutive capital and platform validation.
LLY — Small check for Lilly. Secures antibody discovery platform access. Optionality, not earnings.
Z — AI mode users spend 3x+ longer, view 2x+ homes, run ~3x searches, contact agents ~3x. AI features moving core engagement metrics. Monetizes through leads and ads.
ABNB — Hotels still single-digit % of nights but growing ~3x faster than homes. Direct competitive threat to the core supply advantage. Defend or lose.
HOOD — $225.5M fund giving retail access to YC companies. New private-market distribution channel. Expands TAM beyond securities trading.
FTNT — Net cash for 10+ years. Balance sheet standout in security. No new AI catalyst. Clean hold.
SNAP — 1/3 of normal volume by 7am. No news. Liquidity signal, not a trade.
VLO / MPC / PSX / DINO — Russian refinery destruction squeezing distillate. Product cracks strengthening. High-beta to the crack. DINO is the highest-beta expression.
WMT / HD / TGT — Weak retail data already priced. WMT is the first consumer test; TGT the most discretionary. A beat re-ignites the soft-landing narrative and pulls flow from AI. The surprise direction is up.
CAT — New CoreWeave physical AI customer on Vera Rubin infrastructure. Physical AI ~1% of backlog. Early, but industrial AI demand is entering GPU cloud pipelines.
ONDS — ZeroHedge pumping defense/drone M&A narrative. Separate capital pool from AI infra. Retail crowd levering up; not institutional flow.
TTMI — Epiq Design Solutions acquisition, $1.1B all-cash, immediately accretive to adjusted EBITDA margins and non-GAAP EPS in 2028. Adds high-end design services to PCB capability.
NU / BZAI / BLSH / ROC / QXO — No fresh signals in today's feed. No position taken.
All channel checks, gossip, speculative reads. None of this is in models yet. Treat accordingly.