Good morning. Risk is still an AI barbell, but the financing tail is starting to wag the dog. Futures look like another semi-led open: NVDA FIRST RECORD HIGH SINCE MAY, MARKET CAP PUSHING $6T, while SMTC BROKE $200 TO AN ATH and LITE/COHR keep the optics bid alive. The % board is thin in the feed, but leadership is obvious and breadth is horrid: equal-weight S&P DOWN 7 STRAIGHT WEEKS, 52-WEEK LOWS OVER HIGHS FOR 14 STRAIGHT SESSIONS. The 10Y BACKED UP MORE THAN 10BP FROM MORNING LOWS and refused to rally even after a weak jobs print, which is the real bogey for every long-duration AI multiple. Earnings are not the main event: TSLA beat Q3 deliveries, AEHR is the test print on the radar, and I would rather buy an AEHR air-pocket than chase a vertical test move. Asia handed us a foundry/China tape: INTC bulls are pushing the 14A-within-5%-of-A14 framing at 2.5X P/B versus TSM, while ASML has a live headline problem after reports China stockpiled ~343 immersion DUV systems and Washington re-upped full-export-ban chatter. Oil dipped on the G7 100M-barrel release, but the “covers less than one-third of a 2.8M bpd draw” math keeps energy from bailing out the rates tape. Theme one: NVHBM is a margin knife-fight — MU says NVHBM still lifts ASP/margins, skeptics say NVDA’s custom base die plus TSMC manufacturing moves the HBM4E economics toward NVDA, so watch who keeps the uplift. Theme two: optics has a real regulatory-relief leg, not just vibes — FCC transceiver rules screened “better than feared,” SMTC is the credibility winner, LITE has an ASP lever if Hurlston pulls it, and CRDO stays in the connectivity conversation. Theme three: AI circularity is now a rates-market story — hyperscaler off-BS liabilities hit $3.8T, UP $700B IN Q3, META alone carries $600B OFF-BS, AMZN/AVGO SPV headlines add fuel, and Q3 AI-ecosystem DV01 ran ~25% ABOVE ALL TREASURY ISSUANCE. Theme four: the “cheap compute” map is splitting — INTC/TSM is the foundry-value debate, STX/WDC lose some duopoly premium if Toshiba ramps, and TTMI/ON/TER/ONTO are second-derivative ways PMs are expressing US capacity and test intensity. Keep hedges honest: narrow leadership plus rising DV01 means rallies can gap on NVDA/SMTC and still die in equal-weight land. We'll hit up NVDA, MU and SMTC first, then get to optics, test and the AI-financing complex...
STREET SCRAMBLES TO CATCH UP AFTER A 20% WEEK — AND MOST STILL DON'T BELIEVE. Q4 beat across the board, bookings $22.2B (+5% lc), 37 mega-deals >$100M, FY27 guide of 3-6% lc growth — and the stock ripped +20% in five sessions to ~$210, dragging EIGHT firms into PT raises. Here's the tell: the stock now trades ABOVE TD Cowen's fresh $193 target and right at the new $210-240 cluster. Only two firms (RBC, Evercore at $250) actually rate it a buy. Everyone else chased the number, not the thesis.
Bull (RBC $240, Evercore $250): This is share gain in a bad tape, not a good tape flattered. Multi-year managed services bookings give visibility, the ecosystem/platform positioning makes ACN the arms dealer for enterprise AI deployment, and 400 new AI logos is demand BROADENING, not pilot purgatory. New work is even emerging in AI safety. RBC explicitly says ACN keeps gaining share "in an uncertain demand environment."
Bear (BMO $235, Jefferies $220, TD Cowen $193 — all Holds): The other side of rev/employee +6% is AI eating the billable hour. BMO put it plainly:
"AI is also generating client productivity benefits, and we continue to question the longer-term net impact on IT services growth." - BMO Capital
Jefferies concedes the quarter "challenges the pessimistic outlook for IT Services" but refuses to call a definitive turnaround. TD Cowen flagged "one-time assistance" in the Q4 close, says AI displacement fears are merely "set aside for now," and expects the stock RANGEBOUND until IA Day settles the valuation framework debate.
New: the sheer breadth of the beat (Consulting reaccelerating to 7% matters — that was the dead weight), the AI partner bookings tripling, and Wells going out to FY29 with $85.3B rev / $17.46 EPS (first real three-year anchor on the street). Known: AI demand narrative, managed services mix shift, buy-the-Hold-rating skepticism. The 20% move re-rated the stock before the street re-rated the model — TD Cowen's new $193 sitting below the tape tells you how far behind consensus PTs still are.
Positive for the IT services complex (CTSH, INFY, IBM consulting) — a genuine demand print, not cost-out heroics, with bookings leading revenue. But the bigger debate this quarter tees up is the one that matters for the whole AI trade: is services AI a beneficiary or a victim? ACN's answer so far is "both" — bookings triple while revenue per employee climbs. IA Day is now the catalyst that decides whether $210 is a floor or the top of the range. We lean constructive on the print, but after +20% in a week with a Hold-heavy street, the r/r is a lot cleaner on a pullback than a chase.
CRWV at $88.57 — nearly cut in half from the $153.20 high — and the Street still loves it. That's the setup. Cantor reiterates OW/$176 with the core bull stat: 10X DEMAND FOR EVERY MW OF CAPACITY, 115% LTM REV GROWTH TO $7.59B. Pricing power is real and accelerating — 25% hike in July, another 10% since. None of it hits 2026 numbers. Shows up 2027, bigger 2028. Layer that on top of margin trends already improving, and Cantor's call is the market is mispricing what this thing becomes, not what it is today.
"Cantor Fitzgerald believes the market misunderstands what CoreWeave is today and what it will be in the future, creating an attractive entry point."
The broader tape agrees. Post-Fully Connected event, Rosenblatt at Buy/$250, Truist Buy/$165, Freedom Buy/$151, William Blair initiating Outperform off valuation (3.1x EV/S, 5.2x EV/EBITDA on 2027E). Lone dissenter is Mizuho at Neutral/$115.
Buy-side bogey: demand is not the debate. 10x demand per MW plus back-to-back price hikes tells you everything about the order book. The debate is capex, financing, and whether the 8GW-by-decade-end target is fundable without blowing up the balance sheet — and note the stock is down 40%+ from highs even WITH the bullish chorus, so the market is voting on that exact fear. Mizuho's $115 Neutral is the honest bear: execution and capital intensity, not demand. At 5.2x 2027 EV/EBITDA, r/r skews long if you believe the build-out. But the stock needs a proof point on funding, not another demand datapoint.
The Street's turning cautious ahead of the Q3 print, and the stock's already priced in a lot of pain — APP at $281 sits a hair above its 52-WEEK LOW OF $275 after a 58% drawdown over the past year. Piper reiterated Neutral at $325 (cut from $385 after management talks), flagging rising inventory costs from slower supply growth, more competition, and conversion rates creeping toward 1.4%+ that make the growth math harder from here. Their Network Effects data shows Q3 ad supply growth decelerating.
"AppLovin-specific data suggests lower advertising supply growth in the third quarter." — Piper Sandler
The bull case isn't dead, just cheaper. Morgan Stanley still OW at $450 (cut from $650), Evercore at $510 (from $630) after trimming e-comm advertiser spend assumptions — so even the bulls are marking down, not defending. Piper's 2027 revenue bogey of +23% y/y vs. Street at +28% tells you where the estimate risk sits into the print. Unity lawsuit overhang adds another question mark. Valuation's more interesting after the selloff (Piper concedes as much), but with supply growth slowing and estimates still drifting down, not sure you need to catch this one before the print.
Rosenblatt nudges its PT to $123 from $115 but stays NEUTRAL — and that tension is the story. The firm loves the strategic setup (Physical AI M&A trend, onsemi manufacturing synergies) yet won't pound the table with the stock already +56% over the past year and sitting at $106.15, just ~16% from the new target. Classic "right thesis, wrong entry" positioning.
The bull case here isn't quarter-to-quarter, it's structural: analog/mixed-signal players buying Physical AI assets (wireless connectivity, sensors, edge NPUs, sensor-fusion software), with potential synergies from running select SYNA products through ON's internal fabs. The catch — the transaction doesn't close until MID-2027. That's a long time to hold a synergy story.
"In our view, these transactions underscore a broader strategic trend: analog and mixed-signal companies are pursuing smaller, more agile companies with Physical AI capabilities." — Rosenblatt's Kevin Cassidy
Net: the Physical AI narrative keeps getting validated, but with the close 18+ months out and the stock having already re-rated hard, r/r looks balanced-to-meek here. Watch for any update on the ON integration timeline — that's the real catalyst, not the PT bumps.
Stock's down 14% on the week and Cantor's not blinking. They reiterated OW with a $540 PT — over 50% upside from $351.73 — and the argument is simple: the street's FY28 core numbers are too conservative if AI-native Atlas revenue keeps accelerating. That's the entire bull case in one sentence.
The data supports it, at least directionally. Subscription ARR hit $2.89B, +25% Y/Y, third straight year of acceleration. AI-native customers are still small (3-5% of Atlas revenue) but growing 165% Y/Y — that's the classic consumption inflection setup where the base business looks decelerating while the new cohort quietly compounds underneath. Cantor's read: the company's framework is a floor, with enterprise AI, app modernization, and three new products (Atlas Infinite, Agent Engine) all outside the base case. Plus a $1B incremental buyback to lean on the drawdown.
"Fiscal 2028 core forecasts by Wall Street appear too conservative, assuming AI Atlas revenue continues to accelerate in line with trends reported by MongoDB and partner discussions."
Now the steelman for the bears — and it's not nothing. Everyone else is trimming. DA Davidson cut to $415, Stifel to $385 (transition strategy concerns), BMO to $420 — and BMO's cut came after the CEO departure, which is the overhang nobody can model. A decelerating core in FQ2 FY27 with a leadership vacuum is exactly the setup where "the framework is a floor" gets stress-tested.
Valuation is the tell, though. 12x CY27 EV/revenue vs. consumption peers at 15-18x, and 30 analysts have revised earnings UP. The positioning here feels washed out on the CEO news more than fundamentals. R/r favors the Cantor side if you can stomach the transition risk — but the print calendar is your friend. Wait for the next consumption datapoint to confirm the AI-native cohort is actually bending the core curve before sizing up.
DA Davidson pounds the table at $222 (~42% upside) on a fresh Tier 1/2 P&C carrier win — one that lands in Guidewire's TOP 10 by ARR once ramped. Shares sit at $156, down 23% YTD, so the street's bulls are effectively saying the market has over-penalized the name. The catch: 12-18 month implementation means this is NOT material to FY27 guidance, so this is a 2028 story the bears can shrug at today.
The bigger tell is the positioning split. DA Davidson and Citizens ($220, Market Outperform) lean into FY27 guide upside and the FQ4 beat ($0.99 vs $0.93, ARR $1.24B +19% cc, net new ARR $95M — $5M above the high end). Wells Fargo stays Underweight at $195. Stock sold off on that print anyway — classic "good quarter, hated setup" price action that usually means expectations, not fundamentals, are the problem.
"The new customer is evidence of Guidewire's continued momentum in the market for Tier 1 and Tier 2 property and casualty insurance carriers."
Catalyst watch: INVESTOR DAY OCT 27 at the Las Vegas user conference. That's where the FY27 bogey gets set and where this new win gets sized. Until then, r/r favors the bulls if you can stomach a name the market keeps selling into strength.
PT cuts on a good news day. BMO trimmed to $350 from $379 (kept Outperform) on the heels of CEG's 690MW, 20-YEAR PPA WITH AMAZON out of Calvert Cliffs — and Argus took its number down to $325 from $350 on rate-driven valuation math. Neither is a thesis break. Both are multiple compression on a stock ALREADY DOWN 36% FROM ITS 52-WK HIGH ($412.70 → $262).
The deal itself is the story. ~$85/MWh implied pricing, capacity the swing factor, worth roughly $0.70 EPS and $225M FCF at full run rate per BMO. That's 1.6GW OF NUCLEAR PPAS SINCE Q1 — the hyperscaler contracting machine keeps grinding, and each deal at premiums to forward pricing extends the duration of the earnings base.
"What Is Another 690MW Amongst Friends? Another Step Toward Creating Value."
Context matters: Q2 printed an EPS BEAT ($2.55 vs $2.41) with a REVENUE MISS ($7.5B vs $7.94B bogey), and management RAISED FY guidance on commercial execution and higher PJM capacity prices. The r/r debate here isn't fundamentals — it's what multiple you pay for contracted nuclear in a higher-rate tape. PT cluster now $325-350 against a $262 print. Bulls own the contracting momentum; bears own the duration risk. We'd lean with the former but size for the latter.
MU ripping through the $1T+ market cap conversation and the street is chasing. Baird reiterated Outperform, $1,520 PT, against a stock at $1,097 — the core call is DRAM supply "much tighter" in FY27-28 than FY26, which should force a rerating as the market starts paying for the out-years, not just the current print.
The backdrop does the heavy lifting. MU guided Q1 rev to $61.5B and EPS $38.2, smashed FQ4, and 11 analysts have already revised up. The PT ladder tells the story: DA Davidson at $2,100, CLSA $1,700, UBS $1,625, Freedom $1,600, Bernstein the relative skeptic at $1,300 — everyone singing the same song, just at different volumes. Tight memory supply through 2028, AI demand absorbing every bit.
"Micron expects DRAM supply to be 'much tighter' in fiscal years 2027-2028 compared to fiscal 2026, which should support a stock rerating."
The one caveat worth flagging: Baird models 20% bit growth in CY27 stepping up to 30%+ in CY28 as new capacity (MU's and competitors') comes online. That's the bear seed — the same supply discipline driving today's pricing power gets tested the moment that capacity ramps. And margin expansion is expected to be modest from here, with 40-50% of hyperscalers locked into supply agreements that cap upside pricing flexibility.
For now, r/r still favors the bulls. The shortage narrative has legs into 2028, estimates keep moving up, and the stock hasn't priced the DA Davidson version of the world. Watch the capacity commentary every print — that's the first place the story breaks.
Fresh blood on the tape — FBN initiates at OUTPERFORM with a $325 PT, the street high, with the stock at $276 and sitting just under its $292 52-week high after a +103% YTD run. The pitch is the unified-platform story (one agent, 26 products, 1,000+ integrations vs. competitors' M&A-assembled stacks), ~80% gross margins, and a $1.2B annual R&D engine — plus the track record: five straight quarters beating the top end of its own guide, with Q2 +36% y/y to $1.12B and a $42M consensus beat.
The more interesting nugget for positioning: Stifel's OpenAI spend analysis shows DDOG billings from that one customer hitting the LOW $100M RANGE in Q2. That's the AI-native tailwind quantified — observability attach to the biggest compute buyer on the planet. Consensus PT cluster now sits $260-325 (Macquarie the low, FBN the high), so the initiation is bullish but not contrarian. At 103% YTD and near highs, the r/r on chasing here is thinner than the initiation headline suggests — this is a name you own on pullbacks, not one you chase into a fresh-coverage pop. Not sure a single-shop initiation changes the narrative much when the whole street's already there.
ON — Synaptics deal sweetened on our side: $123/shr in CASH, ~$5.7B EV vs. ~$7B under the old all-stock terms. The amendment came after an unsolicited competing bidder showed up, and the SYNA board still unanimously stuck with ON. Both stocks popped after-hours — market reads this as ON getting the asset for less and removing stock-dilution overhang in one move.
The catch: the street isn't chasing it. Stifel reiterated HOLD at $75 with the stock trading at $80, and the recent analyst cadence has been PT cuts, not hikes — Stifel $90→$75, Cantor $110→$90, Mizuho $110→$95 (kept Outperform), UBS Neutral citing double-digit rev growth potential. That's a $75-95 bogey cluster against a stock up 64% over the past year. Translation: the deal math got better, but the multiple already reflects a lot of good news. Cash structure does help — no arb spread pressure, cleaner accretion story. But with the stock above the Street's low-end anchor, r/r here is about waiting for a pullback, not adding into the pop. Not sure the competing-bidder drama tells us much beyond someone else also liking the asset at these prices.
Wainwright reiterating Buy / $25 PT vs a $10.64 stock — that's a >2x bogey, and the street cluster ($15-25, everyone Buy/OW except Needham trimming to $20 on a "less robust pipeline") tells you the AI-HPC pivot has real sponsorship even if the stock is down 7.6% on the week. The catalyst: Bitdeer locked 70%+ offtake on its 21.7MW Malaysia A201 facility ahead of Q1'27 energization — $1.7B over five years, $340M ARR, more than doubling contracted AI Cloud revenue. Wainwright's SOTP framework holding $25 steady suggests they think the market is still pricing this as a miner, not an AI infra story.
"The GPU ramp in Malaysia is becoming too large for investors to overlook."
The r/r here is a positioning question. Numbers are big ($420M combined ARR across A201 + A102 contracts), but nothing energizes until 2027 — you're underwriting execution and power delivery 15 months out with zero near-term proof points. Bulls own the re-rating optionality at ~$1.6B market cap; bears point to the stock fading the news and Needham's pipeline caution as the tell. 128% LTM revenue growth with 49% projected for FY26 does the heavy lifting for the bull case. Not sure the market believes the 2027 timeline yet — this trades on trust, and trust is thin in BTC-miner-turned-AI names.
TD Cowen lifts its PT to $29 from $26 (Buy maintained) and the thesis is simple: the demand environment for next-gen endpoint is accelerating and consensus hasn't caught up. Stock sits at $24.67, basically AT its 52-week high, up 64% YTD and 85% over six months — so the market has been sniffing this out already.
The key data point: sales cycles have compressed ~15% in recent weeks, with more inbound calls from new logo prospects. That matters because FY27 net new ARR consensus only models 6% y/y growth — TD Cowen thinks that's too low. They also flag "Re-Flexing" as a potential 20%+ ARR uplift lever as cohorts mature. New PT implies ~6x EV/FY28 revenue, 47x FY28 FCF.
"Sales cycle compression signals that the fiscal 2027 net new annual recurring revenue consensus could be too conservative."
Worth noting the rest of the Street is more muted — Cantor and Scotia at $26, Canaccord $25, UBS $24, DA Davidson Neutral at $20 — so TD Cowen is the high-water mark here. The bears' point is fair: Q2 was a clean beat-and-raise, but FCF took a severance hit and some investors wanted more given competitor prints. Still, when the cycle data starts moving this fast, the r/r usually favors the side betting on acceleration over the side betting on the bogey. Not sure we can read too much into one firm's channel checks, but 15% compression is a real number, not vibes.
Bernstein going all-in on NVT as its TOP PICK across entire coverage into the print. Outperform, $229 PT vs $165.77 — 23% upside. The call isn't about the quarter; it's about the market still underpricing liquid cooling earnings power now that RUBIN shipments have started.
The thesis: liquid cooling grows FASTER than gigawatts added over the next 3-5 years as it takes share of deployments, and the VERTIV multiple discount is no longer justified on growth and margin outlook. Bernstein expects convergence "over the next couple of quarters."
Q3 bogeys are modest — revenue $1.43B vs consensus $1.42B, EPS $1.39 vs $1.38 — but the orders setup is the real story. Street expects weak ~10% y/y orders growth off tough comps, and Bernstein flags some demand may have slipped into Q3 after soft Q2 orders. Investors are braced for a 10%+ beat on guidance.
Context: stock +68% over the past year, revenue growing 46% LTM — so this is a momentum name where expectations matter. Street positioning mixed elsewhere: UBS at Buy/$225 post the $1.75B Maverick Power acquisition (funded with $800M in 6.15% notes), Wells Fargo initiated Equal Weight at $182. R/r looks fine if you believe the Vertiv convergence story; the risk is a good-but-not-great print against elevated beat expectations.
TTAN +18% OFF THE POST-Q2 LOWS, and Stifel thinks next week's Pantheon conference in Orlando keeps the momentum going. Firm reiterates Buy/$100 heading in, expecting notable platform and packaging updates — particularly around Max, the product that's been both the bull case and the headache.
Quick recap on why we're here: Q2 print was a 2.7% revenue beat vs. the ~5% bogey Stifel wanted, and new commentary on Max-related revenue recognition headwinds sent the stock off a cliff. Street responded with a cluster of PT cuts — Stifel $125→$100, TD Cowen to $100, Truist $110→$100, BMO the outlier at $90 — but most kept Buys. That's a "near-term mechanics, not thesis" message from the Street.
"Stifel expects management to expand the vision and introduce new on-ramps to Max at Pantheon next week."
The setup for the event: investors want clarity on the 12-18 month growth trajectory as net new business skews harder toward Max, and Stifel expects new on-ramps to accelerate that adoption. One caveat — no formal analyst/investor day at Pantheon, so don't expect a guide reset or framework update. This is product vision, not numbers.
Stock at $67.34, $6.5B market cap. R/r here hinges on whether Pantheon reframes Max from "revenue rec headwind" to "durable TAM expander." Stifel's betting it does. If management delivers on the on-ramps, that $100 cluster of PTs looks conservative against a $67 print. If it's just vision talk with no adoption mechanics, the 18% bounce has probably priced the optimism already.
Profitability story intact, topline still doing just enough. PRGS printed Q3 FY26 with EPS of $1.69 vs. $1.52 street — a clean beat — on revenue of $246.0M vs. $246.7M bogey (a hair light). Lower opex and stronger margins did the work. Non-GAAP op margin at 43%.
DA Davidson sticks at Buy / $55, and the real nugget is DOMO. Management's FY27 revenue contribution color came in BETTER THAN EXPECTED, and initial profitability commentary "sets a conservative floor" for upside. That's the kind of phrasing that usually means numbers drift higher from here.
"DA Davidson said it has confidence management will be able to bring DOMO margins in line with the overall target margin profile of 40%."
The M&A-integration execution risk is the bear case — DOMO margins need to travel to 40% and cross-sell has to actually show up, not just get talked about. But at ~18x earnings with a 0.16 PEG (per the article's data), the market isn't paying for much of that success. R/r leans long if you believe the integration. Not a crowded name — worth keeping on the screen as DOMO numbers start flowing into FY27 guides.
Quiet list: HPE, ONTO, ALGM, SWKS, DT, LR, BRZE, FH, SAIL, CRWD, OKTA, BE, AI, AUR, SPOT, CBRS — nothing on the tape worth a bullet today. CRWD gets an incidental mention via PANW (27-second breakout stat), but no signal of its own.