Good morning. Tape green at the open, semis leading, and the only question for the day is whether the AI capex cycle is extending or the positioning is crowded. Answer: both, and that tension IS the trade.
Lead signal: SoftBank's €75B / 5 GW France commitment — biggest single European AI infrastructure announcement to date, Schneider Electric as named partner, build cycle visibility now extends to 2031. That's a scale-mover. Stack Goldman's +24% revision to SK Hynix 2028 OP (now ~$301B) and +23% to Samsung (~$405B) — those aren't tweaks, that's margin capture confirmation. A "HBM: High-Bandwidth Mistake" skeptic piece is already circulating because of course it is. Memory is the most contested tape of the day.
On positioning: software had its best month since 2002, semis stalled, GS most-shorted basket up 25% in a straight line, GS TMT momentum pair swung 1,020 BPS in a single session. That's flow, not fundamentals. Crowded in both directions.
Three frames for the day. (1) Capex envelope extending vs. positioning crowded — SoftBank 2031 anchor, Goldman HBM upgrades on the bull side, AMD MI300X printing at $32K on eBay and rentable at $1.99/GPU-HR as the bear counter (share split widening, not narrowing). (2) Foundry mix shift — Jensen confirmed NVDA is now 20%+ of TSMC revenue, overtaking Apple, while Lip-Bu Tan's Panther Lake "7X VOLUME" implies Intel 18A yields have moved meaningfully. Two real signals the foundry map is redrawing. (3) Enterprise ROI proof point — Vista's $8M → $200K claims cost reduction is the only concrete ROI data point in the feed; if it replicates, it extends capex willingness at the application layer and is the durable bid under the hardware trade.
Asia: SK Hynix and Samsung both bid on the Goldman revisions. Brent +1.8% at the open on US-Iran — energy is the binding constraint on European DC economics; Dunkirk site selection is nuclear-power arbitrage, full stop.
We'll hit NVDA, MU, and CRWD first, then get to optical (LITE, AAOI) where the parabolic run is starting to roll over. After that, the long-cycle foundry names — TSM, AVGO, INTC.
Beat-and-raise, but the tape didn't love it. Stock trading ~$74-77, off on the day despite numbers that were clean across the board. Go figure.
F1Q27 prints: Revenue +39% YoY, subscription revenue +41%, subscription ARR $1.57B (+32% YoY, beat the 31% Street by ~150bps). Net new subscription ARR $103M, +16% YoY — crushed the $85M consensus. Identity hit $50M ARR (+38% QoQ, +150% over 6 months). FCF margin 19% in the quarter, guiding 18% for the year. Op income $25M (6.4% margin) vs consensus of -$19M. EPS $0.16 vs -$0.53 expected. GM 80.1%. FY27 ARR guide midpoint raised to $1,858M (+27.1% growth) from $1,834M (+25.4%). Revenue guide $1.638-1.648B (~25% growth). FCF guide $293-303M.
10 of 11 analysts moved PTs higher, one cut. Cluster sits $87-$95 post-raises, with Guggenheim the outlier bull at $110 and Jefferies the dissenter at $90 (down from $100). Wolfe had the biggest raise ($70 → $95, +36%). Street PT range $70-$126, consensus Strong Buy.
The PT grid post-print:
| Firm | New PT | Old PT | Move |
|---|---|---|---|
| Guggenheim | $110 | $110 | reiterating |
| Rosenblatt | $95 | $90 | +$5 |
| Wolfe | $95 | $70 | +$25 |
| Scotiabank | $95 | $70 | +$25 |
| BTIG | $91 | $76 | +$15 |
| Stephens | $90 | $75 | +$15 |
| Mizuho | $90 | $80 | +$10 |
| Truist | $90 | $80 | +$10 |
| Jefferies | $90 | $100 | -$10 |
| KeyBanc | $88 | $70 | +$18 |
| BMO | $87 | $73 | +$14 |
Valuation math from the bulls: ~9.5-9.6x FY28 revenue, 46x FCF for KeyBanc, 7x CY27 EV/ARR per Mizuho. Premium to software comps but justified if you believe the growth durability.
BULL (Guggenheim, Scotiabank, Truist, Wolfe): Execution is best-in-class. ARR beat, guide raise exceeded the beat amount, margins inflecting. Identity at $50M ARR is a real product, not a slide-deck number. Non-Cloud ARR grew sequentially for the FIRST TIME IN 13 QUARTERS — that's the legacy displacement story starting to inflect. Hardware cost exposure is a nothingburger per management. Mythos/Agent Cloud creating real pipeline (BMO cites CRWD/PANW comps post-launch). Trades at discount to high-growth software peers on FCF — room for multiple expansion as FY27 plays out.
"Rubrik has a best-in-class product in the mission-critical area of cyber resilience and could benefit from the coming release of Mythos, which is already driving an increase in customer conversations." — Scotiabank
BEAR (Jefferies, the undertones elsewhere): Cloud Net New ARR grew just 5% YoY. Strip out the migration tailwind and it's only ~16% (in line with total). That conversion benefit is DWINDLING, and it shows. Cloud is the engine — if it's decelerating while total ARR holds up via legacy/non-cloud, the composition is getting worse, not better. Stock had rallied 34% in the prior month, so expectations were already high. Memory product hasn't moved the needle yet. 9.5x revenue for 25% growth isn't cheap. And the stock is DOWN on the print — tape is telling you something the model isn't.
The print was strong. Guide raise was strong. Identity is real. Mythos has a pulse. Margin trajectory is real (op income going from -$19M expected to +$25M actual is a 44-point swing on a $400M base).
But the stock reaction tells you the cloud deceleration is the dominant signal for the marginal buyer. Cloud Net New at 5% is bad optics even if management frames it as a mix story. And it's hard to chase a name that's 34% in a month when the next leg requires Mythos to actually convert (analyst meeting NEXT WEEK — could be a catalyst either way).
Trade it as a quality compounder with a real Identity/Mythos call option sitting at ~9.5x FY28 rev. Not screaming buy here, but the setup into the analyst day is interesting if you've been waiting. Jefferies cutting while others raise is a tell on the internal debate — cloud guys vs. thesis guys. Thesis is winning on numbers, but cloud tape is what'll move the stock near-term.
The sleeper: USAGE REVENUE +29%, fastest in nearly FOUR YEARS. That's on-platform monetization working — not just seat adds. Max program LOCATIONS 2X'D in Q1, mgmt guides to 2x again in Q2. >10% of jobs automated among ramped customers. Enterprise cohort ($100K+ ARR) = 60%+ of total billings, still the fastest-growing segment.
Mgmt raised FY27 revenue guide to 18.1% (from 16.5%), incremental margin to 29% (from 25%). Guide raise EXCEEDED the Q1 beat. That's the setup — this isn't a "beat and modestly raise" quarter, it's a "beat and telegraph acceleration."
Bull (KeyBanc, Piper, Freedom, TD Cowen): Growth REACCELERATING, not just holding. Max 2x'ing — AI isn't vaporware, it's revenue flow. Margin EXPANDING while investing in AI — operating leverage intact. 6x forward sales for a 20%+ grower is cheap. Enterprise mix keeps moving up the stack. Mgmt directly rebutted the disintermediation bear on the call (more on that below).
Bear (undertones in the data): Needham's the cleanest articulation: real Max revenue doesn't materially hit until FY28. ~75bps of platform rev came from transitory factors (extra biz day, weather) — strip those out and you're closer to 23%. Q1 is the smallest quarter for trades — easiest comp of the year. Still GAAP unprofitable. The $67 low vs $125 high = nearly 2x PT spread means real dispersion on the name.
1. USAGE REVENUE +29% — Piper called it out as the standout. Highest growth in ~4 years. Means on-platform monetization (Max, GTV-based pricing, voice agents, roofing, commercial) is flowing through, not just seat growth. This is what the bull case needed to see.
2. MAX GOING FROM DEMO TO MEASURABLE — Needham: "membership increased more than 2x Q/Q, expected to double again in Q2." Needham's note also flags "material revenue contributions from Max expected to fall more in FY28" — so we're watching the leading indicator, not the harvested revenue.
3. MGMT NARRATIVE SHIFT — Freedom captured the key repositioning: ServiceTitan pitched itself as "the execution, orchestration, and interaction layer of the trades." Direct rebuttal to the "general-purpose AI disintermediates vertical SaaS" bear. Their frame: the work happens HERE, the AI comes to us. Voice AI is now a "second monetization pillar" per Freedom.
"Squeaky clean quarter with another solid beat and raise. Business is firing on all cylinders." > — KeyBanc
"Subscription revenue beat of 3.5% marked the highest in ServiceTitan's history." > — TD Cowen
The tape: good quarter, punished anyway. Net new ARR of $256M (+32% YoY) beat the high end of guide and consensus — but the beat was only ~$6M (2%), well below the 10%+ pattern CRWD has delivered in 4 of the last 5 quarters. Stock got hit 11% after hours to $664 from a $747 close, bouncing back to $719 by next session. The crime wasn't missing; the crime wasn't being great enough after a 40% run-up in the prior month. Classic "show me" tape on a name that was priced for perfection.
| Metric | Read | |
|---|---|---|
| Net new ARR | $256M (+32% YoY) | Beat, but small |
| Revenue growth | 23-25.6% YoY | In-line to solid |
| Total ARR | $5.51B (+24% YoY) | Tracking |
| FY27 ARR guide | 24-25% (from 23-24%) | +520bps raise |
| Implied net new ARR growth | ~28% YoY | Acceleration |
| FCF margin | 34% | Rule of 59+ intact |
| AI D&R | 2.5x QoQ acceleration, $50M pipeline in 2Q | The new growth lever |
| Falcon Flex | $1.9B, +~100% YoY | Land-and-expand working |
| DSO | Down 5 days YoY, lowest in 13Q | Collections = real demand |
PTs raised across the board into a $675-$790 range, cluster of $725-$780. Bull camp anchored by UBS at $790, Benchmark $780, DA Davidson $765, RBC $755, Piper $750, BMO $745. Hold camp: Berenberg (downgraded to Hold at $720) and Canaccord ($675). Only two dissenters in 10+ names — that's a deeply constructive tape from the sell-side even with the stock down 11%.
The Berenberg downgrade is the read-through to watch. Valuation is the bear case. CRWD trades at 30x forward EV/sales — 460% premium to the 5.2x sector average, 45% above its own 5-yr average of 20x, and 35% above a peer basket (NET, DDOG, CDNS, PANW, SNOW) despite a near-identical 22% 3-yr revenue CAGR. Only Palantir (39x) screens more expensive. Not arguing the quality — arguing the price.
Bull case — CRWD is the best pure-play on the Mythos/AI cybersecurity spending inflection. Record Q2 pipeline, 520bps guide raise, AI D&R going from 0 to $50M pipeline in two quarters, EDR accelerating for the third straight quarter, and the Falcon Flex land-and-expand motion nearly doubling YoY. UBS summed it up best:
"CRWD had the best argument across cyber companies on why Mythos demand accrues disproportionately to them, even if that demand is taking longer to show up than investors had expected." — UBS
The TD Cowen read is similar: the pullback is transitory because the Street "wanted more following the recent run-up" — not because the fundamentals broke. DSO improving to a 13-quarter low means the $256M is real collected revenue, not paper ARR.
Bear case — the beat got smaller because April was a tough close month across cyber, and the Mythos demand is "taking longer to show up" than bulls expected. Worse, the stock gave back $80B+ of market cap on a guide-raise quarter. That tells you positioning was fat and any softness in execution gets punished. Berenberg's framing is the one PMs need to wrestle with:
"At its current share price, the market is pricing in CrowdStrike's growth and duration. We struggle to find meaningful upside from current levels despite raising our price target." — Berenberg
At 30x sales into a sector where the average is 5x, you're paying for a 3-yr forward acceleration story that needs to land without a hiccup. One more 2% beat quarter and the multiple compresses hard.
1. Mythos as a discrete demand catalyst — this is the first quarter where the AI-driven cybersecurity spending narrative has a name attached. CRWD is positioning as the primary beneficiary, and management is leaning into it hard with the $50M AI D&R pipeline as proof. 2. The guide raise by 520bps — bigger than the beat, which is the right shape (forward-looking). Implies net new ARR growth accelerates to ~28% in FY27 from the 24% this year. 3. The "good isn't good enough" tape — first time in a while a CRWD beat has gotten punished. Market is signaling it wants the bar higher, OR the run-up was excessive. Both can be true.
This is the first real "Mythos accrues to cyber" data point. If CRWD's Q2 prints the acceleration the pipeline implies, PANW, S, FTNT, ZS, OKTA all get a tailwind as the AI-spend narrative broadens. If Q2 disappoints, the whole complex de-rates together. The 11% after-hours move on a guide-raise quarter is also a warning shot for any cyber name that has run up into prints with the same setup — positioning matters more than fundamentals at these multiples. Next catalyst: Q2 prints in late August, and that's when we find out if the Mythos story is real ARR or just a pipeline mirage.
VERDICT: Print was a clean beat across the board. Stock got HAMMERED anyway. The setup now is way more interesting than it was at $720+ before the print.
Let that sink in. Q2 rev $1.57B (+40% YoY, +10% QoQ) — beat consensus by $65-70M. Adj GM 44.9% (90bps above guide). Adj OM 19.5% (150bps above guide). EPS $1.64 vs $1.46 consensus (12% beat). FY26 rev guide raised to 32-34% from 28%. Backlog $7.7B, book-to-bill 1.4x. Direct Cloud +69% YoY, Telco +27% YoY. And CIEN got clipped ~14-20% on the day after DOUBLING in the prior 90 days. That's a positioning story, not a fundamentals story.
Five houses touched it post-print. Targets cluster in a wide range:
BULL CASE (Rosenblatt, Stifel, Needham, RJ): The pullback is a consolidation, not a thesis break. Rosenblatt's framework is 45x FY28 EPS of ~$16 — implying a 25%+ TAM CAGR through 2029 to a $50B optical market, with CIEN taking share inside the datacenter. The new disclosures are what matter: (1) industry's first hyperscaler multi-rail order on RLS hyper-rail, "hundreds of millions over multiple years" with meaningful rev starting 2027, (2) CPO/CPE moving inside the datacenter in 2H27/2028, (3) 3.2T coherent pluggables. Scale-across — the architecture that lets multiple AI training DCs operate as one unit — is the new product vector. This isn't a DCI company anymore; it's an AI-fab interconnect company. Stifel on the setup:
"Ciena reported a backlog of $7.7 billion."
That's real visibility. Needham's cloud strength call — Direct Cloud +69% — says the mix is shifting in the right direction.
BEAR CASE (UBS, Neutral): Multiple already does a lot of the work. UBS expects CIEN to compress from mid-80s P/E to ~40x CY27 — "more in line with optical vendors like Lumentum, Corning and Coherent." That's the re-rating trade. Rosenblatt's article noted a P/E of 338. Even UBS's more conservative read has the stock at 206x. The Neutral isn't a "this is broken" call — it's a "you're paying for perfection, and GM trajectory vs optical peers is a question mark given revenue mix." UBS's own take:
"investor expectations were misaligned with the company's ability to drive revenue and gross margin upside given its revenue mix including systems and pluggables along with supply chain constraints"
Translation: the bar got too high too fast. EPS growing mid-40s% CAGR next two years with leverage coming from opex (not GM) means the Street is underwriting execution, not tailwinds.
Lumentum (LITE), Coherent (COHR), Corning (GLW) are the direct comps UBS anchors CIEN to. If the optical cohort re-rates lower on AI digestion fears, CIEN pulls with it. But if the market starts pricing CIEN as an AI-fab play (closer to MRVL/AVGO/NVDA ecosystem enablers) rather than a telco optical vendor, multiple holds. The next leg depends on whether the market believes the multi-rail + CPO narrative is real or roadmap. Watch the next quarter for multi-rail rev disclosure — that's the tell. Currently supplies are gated.
Stock had run +639% YoY and +100% in 90 days pre-print. Was due. The fact that the bid held at $520 after a 20% air pocket on a clear beat tells you the long thesis isn't broken — just the price. At $520 you're paying roughly 40x forward earnings per UBS's own framework, in line with optical peers despite a meaningfully better growth profile. The risk is multiple compression, not earnings. For PMs: this is a name where the question is your time horizon, not your directional view. Bulls want the multi-rail rev to print in 2H26/early FY27. Bears want to fade into the multiple compression as the 90-day rate-of-change normalizes. Both can work depending on your book.
Print was a beat. Guide was in-line with the whisper. Stock got flushed 13% AH anyway. That was the entry. $412 now, almost 20% off whatever-the-high-was, and the setup is cleaner than it's been in months because the Street is finally getting honest about the framing. This is no longer a "does AI work" stock — it's a "how do we underwrite FY27 numbers" stock. Different game.
Q2 FY26: REV $22.2B (vs $22.1E), EPS $2.44 (vs $2.39E). AI SEMI REV $10.8B, +143% Y/Y. Both halves of the P&L (silicon + software) growing mid-20s to mid-30s.
Q3 guide is the headline number: REV $29.4B, +89% Y/Y. AI SEMI $16B. EPS $3.25. LTM rev growth now 32%. Gross margin steps DOWN to 74% (~310 bps QoQ) on higher semi mix — not a thesis-breaker but something the GM-conscious PMs will flag.
The two numbers that actually matter for FY27 modeling:
The Street's split into a clean two-tier setup. Bullish cluster sitting $500-$640, neutral/skeptic cluster sitting $400-$437. Stock at $412 is trading basically at the skeptic price — interesting.
Buyers (4 firms, $500-$640 PTs): BNP Exane raised to $640 from $600, KeyBanc at $575, Bernstein at $550, TD Cowen reiterated at $500. Erste Group also just upgraded to Buy on the AI momentum. All Outperform/Buy.
Neutrals (2 firms, $400-$437 PTs): Macquarie downgraded to Neutral at $437 (Google insourcing with MediaTek is the explicit trigger). DA Davidson at $400, raised but still Neutral, citing "guide fell short of heightened investor expectations."
So you've got consensus skew bullish at +25-55% upside vs the skeptics calling it dead money. Not a hard binary — the dispersion IS the trade.
TD Cowen framed the bull case the cleanest: upside to FY27+ numbers based on "GW-level calculations." That's a model-driven bull case — when your demand is denominated in gigawatts of buildout, top-down modeling stops working and you go bottoms-up on the compute pipeline. That's a regime change for how this stock gets valued.
Bull: AI bookings of $30B in a single quarter with sales of $10.8B is a 2.8x book-to-bill on the most in-demand product cycle in semis history. The 20GW Blackstone/Apollo financing deal is a moat nobody else has — it's not just selling chips, it's financing the data center. 76% GM, 67% op margin, infrastructure software growing 24% QoQ to $8.9B — this is a HYBRID story (best silicon + best software franchise) that the market is still valuing as just a chip stock.
"The Group is growing much faster than its peers in the sector. The stock should benefit from the company's above-average prospects and therefore continue its upward trend." — Hans Engel, Erste Group
Bear: Google insourcing with MediaTek is a real customer-concentration risk — TPU is a meaningful piece of the XPU story and if the biggest hyperscaler pulls in-house, the multiple compresses. Q3 GM guide of 74% is the FIRST material margin stepdown in the AI era — "higher semi mix" today becomes "margin dilution trend" tomorrow if it doesn't reverse. And management's reluctance to lean into upside (TD Cowen called it out explicitly) is a yellow flag — when your CEO won't preempt a raise, ask why.
The 13% AH flush washed out the overbought momentum chasers. What's left is a stock priced at FY26 multiples with FY27 AI sales of $100B+ as the visible destination. The bull PT cluster ($500-$640) is basically saying 25-55% upside is the base case; the skeptics at $400-$437 are saying the cycle peaks here. The 20GW deal + the bookings number make the bear case hard to defend on fundamentals, but the GM stepdown + Google insourcing make the multiple expansion case harder than the bulls want to admit.
Trade it for what it is: a compounder with hypergrowth optics and real cycle risk. Not a clean long, not a clean short. Pair-friendly if you can find the right hedge. The path of least resistance is probably higher into the next guide — but the GM line is the one to watch every quarter from here.
Mixed print sold off. Cheap for a reason.
Beat and raised — $1.09 EPS vs $1.00, rev $830.2M vs $823.23M, FY27 guide up a measly $6M. 9% growth, 50bps beat vs high end. Stock down 5% AH after being up 27% off the Feb low. So you had a quarter where DOCU did basically what it had to do and the stock STILL couldn't hold. That tells you something.
All three covering analysts stuck on the sidelines. WFC actually LOWERED to $55 from $60 (EW). Jefferies raised to $50 from $45 (Hold). Needham maintained Hold. Two PTs below current ($50.94), one above. The Wells Fargo cut is the headline action here — they read this as a mixed quarter with "modest growth improvement but ultimately not much change."
The WFC note is the most honest read:
"Investors continue to debate whether the second half presents room for further improvement or if growth starts to fade as the company laps the Intelligent Agreement Management launch."
That's the whole debate in one sentence. The 2H setup is the coin flip.
Bull — IAM ramp is real (12.6% of ARR, up 180bps in one Q), net new large deals hit a record, stock trades at ~10x CY27 P/E which Jefferies calls the cheapest in their mid-cap coverage, 80% gross margins, 11% FCF yield, $50 into a name with that cash conversion is not crazy if growth re-accelerates. Conference demand was strong.
Bear — Three quarters of "modest improvement" and you're still at 9%. Guide raise was $6M — that's rounding error. Dollar net retention stuck at 102% with no inflection. IAM is the new product easy comp — the lap risk WFC flagged is real. Needham basically said the quiet part: awaiting "more tangible evidence of growth acceleration and improved sales and marketing efficiencies." All three analysts on the sidelines, two of three PTs below where it trades.
Cheap stays cheap for a reason until it doesn't. The setup is: FY27 guide is conservative enough that prints should clear, IAM is the real catalyst if mix keeps shifting, and 2H is the proving ground. If you're long here, the r/r is reasonable at 10x with 11% FCF yield — but the path to a multiple re-rating requires a growth re-acceleration print, not a "modestly better" print. This is a show-me story for another 2 quarters. Not a sell, not a buy, not yet.
WFC's $55 is the PT to watch — if IAM momentum continues and 2Q clears the 1.8ppt headwind, that's a tighter bogey. The $50 from Jefferies is effectively "this is fair value with zero credit for re-acceleration."
PT cluster now $475-510: JPM $475, TD Cowen $490, Cantor $510. Cantor still the bull anchor, JPM catching up to the reframe.
Subtext: the bull case has been the same for three years, the STREET was just refusing to pay for it. JPM capitulating is a flow event, not new info for the longs who lived through 2022-24.
Bear: P/E 386. Optimus is real but zero revenue. Cybercab regulatory path is a minefield. Musk is Musk. And at $1.57T you're paying for Optimus + Cybercab + auto stabilization all hitting. Asymmetry is gone — this is a quality compounder now, not an asymmetric long. The r/r that made TSLA special is just... not there at this mkt cap.
Verdict: the Street is split 2-to-1 bull, but the bear camp is loud and not stupid. PT range into the June 10 print is the widest in mega-cap land — $190 (RBC) to $400 (Guggenheim) — and the stock sits at $236 which is roughly the bottom quartile of that range. That's the trade. You're either fading the bears or leaning into a name where the LFCF is a horror show but the RPO is a once-in-a-decade backlog.
Five days out from the F4Q26 print and ORCL is doing what ORCL does pre-print — grinding higher into the setup, the AI tailwind narrative fully intact. Stock at $236, market cap just shy of $680B, trading at 42.6x P/E. The news flow into the print has been almost uniformly positive on the demand side, ugly on the funding side. Same coin, two sides.
The Street has coalesced into a clear two-tiered view post the $45-50B raise and OpenAI's $120B round. Guggenheim, UBS, Mizuho, Scotiabank are all firmly in the bull camp with PTs of $285-400. RBC is the lone holdout at $190 — and notably their hike from $160 was a multiple expansion call, not a fundamentals upgrade. They're not saying ORCL got better, they're saying the comp set re-rated.
Capex is the number that matters more than revenue. Guggenheim modeling ~$75B for FY27, their own estimate $85B for FY28. LFCF was -$24.7B LTM. This is a company burning cash on a scale that would have been unthinkable for an enterprise software vendor 18 months ago. The bet is that the $553B RPO monetizes faster than the capex compounds. RPO growth is the single most important line item on the call.
Bull case (Guggenheim "Decade Stock", Mizuho $320, UBS $285, Scotiabank $290) — the $553B RPO is the moat, the $45-50B raise de-risks the build, OpenAI's funding validates the demand curve, and partner checks into F4Q were strong (largest partner beat, three others met, with Middle East + Industry Solutions + Cloud Database cited as the drivers). Guggenheim is a Best Idea and called it a "Decade Stock" on tech and execution. They're explicitly OK with the FCF trough in FY27 because RPO gives them visibility into the FY29 inflection.
Bear case (RBC $190) — the Stargate structure just changed under their feet. OpenAI abandoned the first-party JV model in favor of bilateral capacity leases, which "could result in meaningfully different ownership and financing demands for Oracle than originally anticipated." Translation: ORCL may have to fund more of the build alone than the original deal implied. Pair that with 30,000 jobs cut (18% RIF) to redirect opex to data centers, negative LFCF, and a P/E of 42x, and the bear says you're paying growth multiples for a company in the middle of a multi-year FCF air pocket. RBC's framework: watch the multi-year capex guide, the Stargate financing impact, GPU deployment pace, and customer diversification within the IaaS backlog. Note RBC is $46 BELOW the current price.
1. FY27 capex guide — street at ~$60B, ORCL likely guides $75B+. Higher = more pain, but also more conviction on demand. 2. RPO growth — needs to extend the trajectory; $553B was last print, any deceleration gets punished. 3. IaaS revenue/ARR — Guggenheim needs a similar print to the last two quarters to validate the FY27 ramp. 4. Stargate financing structure — any color on how the JV-to-lease shift affects ORCL's balance sheet. 5. Cloud gross margin — Guggenheim flags risk to consensus here on IaaS mix and depreciation. Operating margin less at risk.
"Decade Stock" based on technology and execution. — Guggenheim, Best Idea note
The $190-to-$400 PT spread is doing the talking. You're not buying ORCL on fundamentals at $236, you're buying the RPO conversion story. Print matters.
Street keeps leaning in. Six firms lifted PTs post-Summit/Investor Day, clustering now $282-325 with UBS the bull outlier at $370. Spot $245, $86B mcap — the print is in the green, but the setup is the FY27 guide raise (400BPS to 31% PRODUCT REVENUE GROWTH) plus a fundamentally faster go-to-market. The debate is no longer "is the AI story real" — it's whether 31% re-accel from a $5B base already prices in the next 18 months.
LTM rev $5.03B at 31% YoY. Net new customers running +36% YoY. Migrations 1.9x. Use cases 1.7x. But the real tell is sales cycle: usage ramp to production now ~7 MONTHS, down from ~12 — roughly 40% faster. That's the conversion flywheel. If usage ramp is shortening AND NNR customers are accelerating, you're not getting one quarter of beat-and-raise, you're getting a structural consumption curve shift. FY27 guide raise of 400bps wasn't a hedge — it was management admitting the funnel is bigger than they thought.
Cluster sits at $282-325, with UBS at $370 staking out the high. Consolidated read: CoCo (their agentic/orchestration layer), Cortex Code, and CoWork are doing what every data infra platform has promised for three years — actually moving customers from POC to production at scale. The 400bps guide raise validates the consumption durability.
Best line of the bunch, from Truist: SNOW is "building out a large, high-quality AI opportunity" as the governance/connectivity/orchestration layer for enterprise AI. That positioning is the bull case in one sentence — SNOW as the system of record + system of action for AI workloads, not just a query engine.
Bull — 31% growth on a $5B base is hard to find in software. Sales cycle compression + agentic tools = consumption compounds. Governance layer for AI is a TAM expansion, not a TAM substitution. Databricks punting the IPO removes the near-term supply event AND means SNOW is the only pure-play public expression of this category.
Bear — At $86B, you're paying for 35-40% growth three years out. Easy to be disappointed on a single quarter. Competitive set is expanding (Databricks, hyperscalers all building native AI orchestration). And consensus 1.49 with this much analyst enthusiasm — the room for incremental positive revisions is narrowing vs. the room for an air pocket on a miss.
Constructive but not a chase here. The guide raise + cycle compression is real, the narrative upgrade (AI governance layer) is durable, but you're 15-50% from the PT cluster with the stock already working. We'd be adding on weakness into the mid-$230s, fading $260+ into prints. The asymmetric setup was last quarter.
C3.ai is a broken stock with a $70M lifeline from the founder. That's the trade. Tom Siebel just bought ~$70M of stock at $11.16 in conjunction with his return as CEO — about as loud an insider signal as you get in software. Everything else around this name is ugly. Revenue down 53% YOY, four straight quarters of declines, op margins worse than -100%, and the only "beat" was a couple of pennies on EPS. Stock at ~$10.60, still down 59% over the past year and trading closer to its $7.67 52-week low than the $30.11 high.
The PT cluster says it all — range is $6 TO $12 and nobody's at Buy:
Bull case: Siebel putting $70M of his own money where his mouth is. FY27 guide midpoint of $225M cleared the $200-210M Street bar. Return to growth expected in H2 FY27 (admittedly on an easy comp — H2 FY26 was down ~50% YOY). Cost structure improving, headcount down, founder back in the seat. At ~5x cal 2027 revenue, the multiple is cheap if you believe the bottom holds.
Bear case: Four consecutive quarters of YOY revenue declines isn't a stumble, it's a structural break. Bookings missed. FY28 outcome range "remains wide" per UBS. The "improving pipeline and deal velocity" language is exactly what you'd expect from a company that needs a narrative shift more than it needs a quarter. Management itself called execution issues "completely unacceptable" — not a great look from the CEO suite.
"Fiscal 2027 will likely be a noisy year given the restructuring, go-to-market changes and the return of Tom Siebel as chief executive officer." — UBS
That word — "noisy" — is analyst-speak for "we don't have conviction in the numbers and neither should you." UBS is explicitly sitting on the sidelines waiting for visibility into growth drivers before getting involved.
5x cal 2027 revenue is a low multiple, but for a company in revenue freefall with no near-term catalyst except an insider bid, the market's pricing in continued decay for a reason. The $70M Siebel buy is the only reason this isn't a zero — and the smart trade is to let him prove it. Watch the bookings trajectory in the next couple of prints; that's the real read on whether "improving pipeline" is real or marketing. Not a long here until we see the sequential bookings inflect.
Verdict: Broken stock, broken narrative, broken chart. Print confirmed what the tape was telling you for 6 months.
ARR $845M, +29% YOY — matched Street ($842M) but missed the more optimistic buy-side bogey of $856M. That's a ~300bps DECELERATION QOQ, and on a stock down 37% over 6 months, "in-line" doesn't cut it. Revenue $202M, +28% YOY was actually a slight beat ($198M expected) but nobody cares when ARR is the metric that matters. NRR compressed to 113% from 117% — four points of deterioration, and that's the line item that tells you customers are tightening wallets on upsells. Stock at ~$9.80 after hours, hovering near 52-week low of $7.67, well below the $27.99 high. Earnings June 10.
Five firms cut PTs in a tight band. The cluster is now $13-19 from a prior $18-25 range. BMO, RBC, and Mizuho all landed at $13; MS and BTIG at $14; Piper at $18; TD Cowen the high outlier at $19. Ratings are universally held — Outperform/Buy/Overweight across the board — but the price targets got crunched. Net effect: even the most bullish sell-side target ($19) implies ~90% upside, while the cluster around $13-14 is only 30-40%. Buy-siders are not going to be anchored to those numbers; they're going to be anchored to ARR trajectory and NRR.
"The timing of the CFO exit could be viewed unfavorably by some investors following the deceleration in several SaaS key metrics." — TD Cowen
That's analyst-speak for "this is going to be a problem for the stock." CFO exits after growth deceleration reads as either (a) the numbers are worse than reported, or (b) the board lost confidence, or (c) both. Neither interpretation is bullish.
Bull case: New logo ARR +60% YOY is genuinely strong — that's the engine. AI products showing early pipeline traction, sales rep ramping expected to drive a 2H re-acceleration, and the company is taking share in a dynamic SSE/SASE market that's structurally growing. TAM is real, product is competitive, and 29% ARR growth isn't dead. BMO frames it as "net new ARR roughly flat YOY in F2Q, modestly growing in 2H" — implying the bottom is near. CFO exit could just be a normal career move.
Bear case: NRR compression is the headline. -4pp YOY means existing customers are expanding less, and that's the metric that compounds against you in a SaaS model. Growth is decelerating, profitability is a mirage (net loss $679M), and the CFO leaving right before the print is a credibility hit you don't need when the narrative is already fractured. The $13-14 cluster PTs from the more bearish houses imply these analysts think the deceleration continues. Multiple compression + growth compression = no place to hide.
NTSK is a name where the chart is doing the fundamental work — 37% off the highs, near 52-week lows, CFO out the door. If you're long, you're betting on a 2H re-acceleration that nobody can confirm until it prints. If you're short, you're fading a stock with a $13-19 PT floor and 30-90% "upside" to targets that are themselves being cut. June 10 is the next catalyst. Until then, this is a trading-range stock at best, falling-knife stock at worst. We'd be looking for a clean print with NRR stabilization and a credible CFO replacement story before we'd underwrite the bull case. Not there yet.
Verdict: The relative strength name in a wrecked tape, but the setup into WWDC is binary. AAPL only down ~1% while the Nasdaq gets crushed -4.2% and semis get absolutely obliterated (MRVL -17%, MU -13%, AVGO -8%). After a 56% run over the past year and trading at 37.6x, the print at $307-311 is near the 52-week high of $316.94. Not a spot to be adding aggressively. But in a tape where everything else is leaking, you don't fight this name.
Two-firm snapshot ahead of WWDC (kicks off June 8) and the spread is wide — $84 of PT, or roughly 28%. That's a lot of disagreement for a mega-cap with this much analyst coverage.
"BofA views WWDC as an important marker for Apple's ability to convert its installed base, silicon roadmap, privacy architecture, and App Store distribution into a differentiated agentic AI platform."
That's the bull case in one line. Whether the market buys it Monday is the question.
Stock is up 56% YTD into an event where the smart money knows the comp is hard and the US App Store is rolling over. The bull needs Siri/AI to land HARD on Monday to justify $380 from here. The bear doesn't need anything to happen — the multiple does the work. With semis down double digits and the Nasdaq breaking, AAPL's relative strength is real but the asymmetry isn't great at 37.6x. Tactical: fade the pop into WWDC if Siri underwhelms. Strategic: this is still the cleanest balance sheet in mega-cap tech and the buyer of first resort on any 5-7% pullback.
Verdict: Print was clean, stock said no thanks, AI is the new narrative lever. Q1 beat (rev $478.8M vs $455.2M, EPS $0.17 vs $0.13), FY27 guide bumped to 24% from 21.7%, and both covering shops walked up PTs — yet IOT sold off in the cash session and kept selling after hours. The bull camp calls that a buying opportunity; the tape disagrees for now. Worth a closer look but not a chase.
Only two covering shops on the tape today, both lifting targets:
Bull: 30% growth at $1.6B+ revenue is genuinely rare. GM at 76.8%, ARR mix shifting toward emerging products and international, large-customer multiproduct expansion still working. RBC says the "growth and profitability at scale remains impressive" — the AI product cycle (dashcams, asset tracking, operational intelligence) is a fresh TAM expansion that the bear case hasn't discounted in.
Bear: The stock printed red on a beat-and-raise. That's the tape telling you something the press release isn't. ~13x sales is rich for a hardware-adjacent IoT name if the multiple compresses, and Piper flags the AI-vs-software rotation risk (even while dismissing it) — when narrative flips, growth-at-any-price names get hit first. Guide being "conservative" is the same line that turned into a haircut last cycle.
End-of-month user conference / analyst day. Operational AI Intelligence is the headline — Piper explicitly says it doesn't see the AI vs software debate weighing on IOT, RBC sees "additional potential use cases" with scaled adoption. If the conference fails to convert AI chatter into bookings acceleration, the post-print selloff makes sense in hindsight. If it lands, $42-44 is in play and the bear narrative breaks. Positioning still looks light here (RBC target implies ~20% upside, Piper ~14%) — not a crowded long.
Verdict: Still the AI infrastructure tape-leader, but the easy money's behind us — this is now about landing the customer wins that justify the multiple.
Rosenblatt out today post-AIPCon 10, reiterates Buy, $225 PT (above Street consensus). The thesis hasn't changed — Karp's shop is the picks-and-shovels play for enterprise AI deployment, and the customer logos keep stacking. Worth flagging: 21 analysts revised estimates UP for the upcoming period. That's not a vibe, that's a tape tell.
The new wins are the actual story here, not the price target. Customer chatter at AIPCon ran the full vertical: Kirkland & Ellis (legal/PE platform for fundraising — basically a private markets ontology play), McCarthy Building (multi-year, AI into construction ops — boring, profitable, exactly the kind of unsexy deployment that compounds), USDA, Hertz, Nscale, Accenture, Parts Town. And the Google Cloud expansion is the sleeper — two-way data federation between BigQuery and Foundry means PLTR is now a native citizen of the GCP stack, not a third-party bolt-on. That's a channel story.
"We view Palantir as one of the most important components of the enterprise AI value chain."
Rosenblatt's framing, and honestly it's the right one. Revenue +68% LTM, gross margins 84% — this is software economics at scale, not a services company in disguise (though the bears will still scream "services" until they're blue in the face). Baird also parked at Outperform with a $200 PT for context on the Street range.
The bear case, steelmanned: valuation is doing a lot of the work. Every customer win is already partially priced. The gov exposure creates headline risk (one bad DoD contract cycle and the multiple compresses 5 turns overnight). And "enterprise AI" is becoming a crowded table — Snowflake, Databricks, ServiceNow all want a slice of the same workflow ontology layer PLTR built first.
The bull case: the customer list tells you what the product actually is. Kirkland & Ellis running a PE fundraising platform on Foundry, McCarthy putting AI into construction project ops — these aren't pilots, these are mission-critical workloads. Stickiness compounds. Every logo becomes a beachhead for 3-5 more in the same vertical.
PM takeaway: own the name, don't chase it. The $225 bogey is a print-the-news level — real upside requires the next quarter to print a >50% beat-and-raise with US commercial rev acceleration. If you don't have it, you trim into strength. If you do, you let it run.
GenAI monetization stops being a PowerPoint. That's the readthrough from Conversations 2026 — Meta's Business Agent is out of beta, globally available across WhatsApp, Messenger, and Instagram, and already in use by 1M+ businesses handling CS, recommendations, and closing sales. UBS keeps Buy with a $865 PT (stock at $638, so ~35% upside) and frames this as the start of the largest of Meta's five new GenAI revenue streams. Translation for the room: the "Zuck is burning capex with no ROI" bear case just got harder to defend.
The underrated piece is the 2027+ revision setup. Investors have already digested higher opex and capex from 1Q26, so anything that moves the revenue line higher starting next year is asymmetric. 26% LTM rev growth and 82% gross margins give them the firepower to absorb the spend while they wait for the monetization to ramp — that's the structural advantage nobody gives Meta credit for vs. the pure-play AI names. Mizuho and Canaccord both doubled down on the same theme (Outperform/Buy reiterations), so this is a consensus narrative in formation, not a one-shop view.
"The development should alleviate investor concerns about Meta's ability to deliver return on invested capital on its GenAI capital expenditures." — UBS
(Also worth flagging: Hatch consumer AI at up to $200/month — that's a ChatGPT Pro/Claude Max comp, and if even a sliver of Meta's engagement base converts, the TAM math gets stupid fast. Not modeling it yet, but it's on the watchlist.)
Verdict: Cheap on the tape, but ARR re-accel is the only thing that rerates this — and June 11 is the print. RBC stays Outperform $350 (basically the high PT on the Street) arguing ARR is "within striking distance" of re-accelerating Q/Q. That's the line. Everything else is noise until that gets confirmed.
The setup's actually pretty clean. Stock down 26% YTD against IGV down 5% — so it's done its job punishing holders. P/E at 15.14 on a $104B mkt cap name with 89% gross margins is silly low for a software franchise of this quality. The SaaS complex has stabilized since the end of March, up 8%, and ADBE hasn't really participated in that bounce. So either it's a laggard catching a bid, or the market's correctly pricing in structural ARR decay. June 11 settles it.
The bull case (which is what RBC is buying) is that Experience Cloud is doing the work. Three businesses now over $1B in ARR growing 20%+, with AEM and Agentic Web called out specifically. That's wallet share expansion inside the enterprise, which is what Adobe needs to tell a growth story again. Piper's sitting on Neutral $280 citing the 9.9% Q2 rev guide — fine, but that's not where the upside comes from. Upside is the ARR trajectory inflection.
"ARR is within striking distance of re-accelerating growth on a quarter-over-quarter basis." — RBC Capital
DB initiates Buy $350, Stifel reiterates $350, Needham ratchets to $330. Three brokers, tight cluster. Street locked in constructive here.
"Stifel reiterated a Buy rating on Onto Innovation... due to the accelerated adoption of the company's Dragonfly G5 inspection tool by key customers. The tool's qualification process was notably faster than anticipated, highlighting strong demand."
That's the kind of color PMs want — faster qual cycles mean revenue pulls forward. Not just vibes.
The buy-side takeaway: MSFT is quietly weaning itself off frontier labs, and the Street is starting to notice. TD Cowen came out of Build with a $540 PT reiteration, and the more interesting read isn't the price target — it's the seven self-built models MSFT rolled out. Fine-tuning focused, cost-optimized, designed to reduce reliance on the OpenAIs and Anthropics of the world. That's a narrative shift, not just a product update.
Why it matters: this retroactively explains the GPU buildout. TD Cowen explicitly called it out — the GPU capacity MSFT has been hoarding in recent quarters wasn't just for Azure resale or OpenAI inference. A chunk was going to internal R&D. Now we know what for. PMs who were confused about MSFT's capex intensity vs. the revenue ramp on AI should be paying attention here. The "frontier model dependency discount" that some bears have carried in their heads is starting to look stale.
Wall Street is clustering Buy with PTs $502-$575 — Jefferies high end at $575, JMP at $550, Cantor at $502, TD Cowen $540. Cantor specifically called out the pivot from "Copilot assistants to a comprehensive enterprise agent platform" which is the right framing. Jefferies likes the model-agnostic posture. Nobody on the sell-side is breaking ranks here.
"The launches reduce Microsoft's reliance on frontier labs for AI capabilities... the developments explain why GPU capacity had been allocated to internal research and development in recent quarters." — TD Cowen
The bear case (steelmanned): DIY models at the fine-tuning layer don't compete with GPT-5/Claude 4-class frontier reasoning. MSFT still needs OpenAI for the top of the stack. The "agent platform" narrative is still 12-18 months from showing up in enterprise wallet share. And at $3.2T market cap, even great execution is priced in. Not sure we can call the frontier lab decoupling a done deal after one conference.
Positioning lens: MSFT has been a beta-to-mag-7 trade for most of 2026 — when NVDA sells off, MSFT catches a bid as the "safer" AI exposure. The Build narrative gives fundamental bulls something new to hang their hat on beyond "Azure is growing." Pinecone-OneLake integration is a small but real data plumbing win for the agent thesis.
Watch for: next earnings — specifically any disclosure on internal vs. external model mix in Azure, and whether the DIY models start cannibalizing OpenAI rev share. That's the line item that will tell us if the strategy is working or if it's just a defensive moat-digging exercise.
Verdict: the easy 293% is gone, but the next leg has a credible setup. Evercore just rolled the dice with an upgrade to Outperform, $155 PT — and the rest of the Street is in a noisy cluster around it. FORM is no longer a hidden probe-card name; it's a consensus AI-test play now, and the question is whether you chase at 46x NTM P/E or wait for a pullback into the teens.
The print justifies the bid. Q1 EPS of $0.56 smoked the $0.44 estimate (27% beat), revenue of $226.1M came in 11% ahead on the $204M street number, and gross margins hit 49% — A 510 BPS QOQ EXPANSION. That margin print is the cleanest evidence that probe-card pricing power is real, not a function of mix tailwind.
EPOS is the engine. Evercore's working assumption is a 41% TWO-YEAR CAGR for the EPOS business, and the entire 38x 2028 EPS of $5 valuation framework (discounted back to 2026) hinges on that holding. PEG prints at 1.12x — rich, but not bubble territory if growth delivers.
Street is split, and that's the trade. Bull camp is Evercore at $155 and Craig-Hallum, who just took the stock to Buy with a $175 PT, modeling $1.6B annual revenue. Cautious side is Stifel at $135 (Hold) and Northland at $118 (Market Perform) — both raised PTs but neither is willing to underwrite the AI multiple. $175 bull, $118 bear = A $57 SPREAD that tells you nobody knows what this thing earns in a normalized cycle.
"FormFactor's strong first-quarter results and significant gross margin improvement" — Stifel, raising PT to $135
The trade: 293% in a year kills the easy money, but the gross margin print and EPOS trajectory say this isn't a melt-up — there's a fundamental story backing it. Problem is positioning. Anyone who bought under $40 has life-changing gains and will be sellers into any rip. Size accordingly. The 38x 2028 framework means you need to believe 2027-28 estimates are sticky, because any cut and that multiple compresses fast.
Verdict: The story works. The stock doesn't. Northland out with a Market Perform after the Clearwater Forest demo, and that's the tell — the firm literally just said INTC is catching TSMC on process and AMD in server, then said "but we can't own it here." When the bull thesis is working and the sell-side still won't get constructive, you're paying for the comeback. INTC +386% OVER THE PAST YEAR to $37.19, kissing the $38.99 52-week high. That's not a coiled spring — that's a stock that has already re-rated.
Clearwater Forest matters. Northland's read is that INTC is closing the gap on TSMC's process lead and AMD's server share, and the foundry packaging biz is the real sleeper here — INTC talking "billions" in annual packaging foundry revenue, with real customers attached. That's the under-appreciated angle for most of the buy-side. Packaging capacity buildout is a direct read-through to Veeco (VECO) and Camtek (CAMT) on the equipment side, and Alphawave IP on the heterogeneous integration piece. Own the picks and shovels if you don't want to fight the INTC tape.
Valuation. That's it. They're not disputing the trajectory — they're disputing the entry.
"Intel's stock has surged 386% over the past year to $37.19, trading near its 52-week high of $38.99."
Translation for PMs: the easy money's been made. The next leg requires either (1) a cleaner read on gross margin recovery as 18A ramps, or (2) a foundry customer win big enough to re-anchor the multiple. Neither is a Q2 catalyst.
Bull: Packaging foundry is a $10B+ TAM that INTC basically owns. 18A external customer pipeline fills in. Server share stabilizes with Granite Rapids / Clearwater Forest cycle. Multiple rerates to 2x+ sales as foundry mix shifts.
Bear: Stock up 4x in a year into a still-dilutive margin profile. Foundry "billions" is a 2027+ story. Competition in packaging from TSMC and Samsung isn't standing still. $37 on 2026 EPS that probably prints sub-$1 is a rich multiple on hope.
Positioning read: Longs are crowded, shorts are dust. R/R skews neutral here. Better risk/reward is in the packaging supply chain (VECO, CAMT) where the INTC re-rating is the gift that keeps giving without the multiple problem.
Love the story, hate the chart. If you're underwriting the foundry/AI thesis, get paid through the equipment names. If you must own INTC, size it for the next leg — not the last one. Wait for a 10-15% pullback into the 18A yield updates and Clearwater ship dates for a better R/R.
Two fresh initiations, both bullish. Stock's already done the heavy lifting — +78% YTD to $6.52. Question for PMs: chase or is this the start of a real re-rating?
Craig-Hallum opened with a Buy and $18 PT, but the headline isn't the PT — it's the long-dated framing: potential for a valuation exceeding $100 BILLION. Current market cap is $13.4B, so that's roughly 7.5x from here. Northland is the other initiate, Outperform at $11 PT. Street targets span $3.59 to $15 — CH's $18 sits well above the prior high, so the street will need to catch up to validate the move.
Bulls say: Aurora Driver is the most commercially advanced on-highway trucking product, R&D is heavy, the team is legit (industry pioneers out of the AV racing world), and if autonomous long-haul works commercially the freight opportunity is massive. That's the whole trade.
Bears say: no revenue, ongoing operating losses, Q1 was a $0.10 LOSS (beat by a penny, but still a loss), and dilution is the story until the model flips. The $3.59 low-end target exists for a reason — this is a name that's been "5 years away" for a long time.
"Aurora Driver is the most commercially advanced on-highway trucking product currently available." — Craig-Hallum
At $6.52 you're paying for a 2027+ story. CH's $18 is the "if it works" number; $3.59 is the "if it doesn't." Wide cone. Catalyst path is commercial scaling milestones and proof of unit economics — nothing else moves this name. Time horizon is everything. Not sure we can read too much into a single quarter's penny beat, but the initiations matter more than the print.
Cellebrite = quietly turning into a real compounder, and the print still doesn't reflect it. TD Cowen's Eyal came out of their TMT confab with mgmt (CFO Barter + IR Kramer) still at Buy, $23 PT against a $14.29 print — that's ~60% upside in a name nobody's talking about. The setup is straightforward: Q1 EPS PRINTED $0.12 VS $0.06 STREET (2X THE BAR), revenue $128.3M also slightly ahead, and 84% GROSS MARGINS that would make any SaaS PM jealous. This isn't a show-me story anymore.
What we like here: digital forensics is a niche with LIMITED COMPETITION, AI is making the TAM bigger (more endpoints, more data to extract, more cases to solve), and CLBT just dropped Genesis AI as the platform anchor. The SaaS transition is the real catalyst — recurring revenue mix keeps grinding higher, and every incremental dollar drops at those gross margins. 18% LTM REVENUE GROWTH isn't explosive but it's plenty for a profitable, 84% GM franchise that's still in early days of the platform shift.
Mgmt sounded constructive on demand at the conf — no signs of budget pressure, federal/law enforcement pipeline holding up, enterprise expanding use cases beyond the core investigative workflow.
"Cellebrite's leading position in the digital intelligence market, noting limited competition and industry tailwinds from rising AI adoption. The company's financial strength supports this positioning, with an impressive gross profit margin of 84%." — TD Cowen
Under-owned, under-followed, just delivered a clean 2x EPS beat. Not a homerun name but a decent risk/reward add at these levels. Size accordingly — this is a compounder, not a squeeze.
Needham bumped PT to $53 from $40, Buy intact. The print was strong — but the tape is already pricing in a lot of perfection here. PL is up 630% OVER THE PAST YEAR. Bumping PTs after a name has already 7x'd is a lagging indicator, not a leading one.
Q1 FY27 revenue $94.2M vs $90M consensus (+4.6%). EPS -$0.03 vs -$0.04 est. Revenue growth accelerated to 42% YOY — fastest in 12 quarters. Two standouts driving it: EMEA +88% YOY (eight-figure European contract landed early in the quarter) and Defense & Intelligence +68% YOY. Backlog +72% YOY TO $906M. Gross margin 56%. Company hit Rule of 40 and says it's sustainable.
Q2 guide 3.6% ABOVE CONSENSUS on revenue. FY27 midpoint held in line, but GM guide raised. Needham nudged FY27/FY28 estimates up modestly.
"First-quarter revenue growth accelerated to 42% year-over-year, marking the highest growth rate in 12 quarters."
The acceleration narrative is the real story. Three quarters ago this was a decelerating name with a defense/government exposure overhang. Now you've got EMEA exploding, D&I re-accelerating, AND Rule of 40 being delivered simultaneously. The 12-quarter acceleration stat is hard to argue with — that's not a quarter-end pull-forward, that's a real mix shift.
THE BEAR CASE though — and PMs need to sit with this: still unprofitable. After a 630% run, the multiple is doing all the work. Needham going to $53 implies maybe 20%+ upside from here, which is fine for a quality compounder but not a moonshot. Positioning in the name is almost certainly crowded after that return profile — any deceleration in the D&I or EMEA lines and this gets chopped hard. Watch the Q2 print for whether EMEA's 88% growth was lumpy contract timing or a real run-rate. That's the make-or-break data point.
Verdict: Real product catalyst, but a lot of that $1B FCF story is already in the stock. Up 68% Y/Y, sitting 6% off the 52-week high ($133.46), trading at $125.59. Piper's $148 PT — only 18% upside. The Street is increasingly constructive (MS at $170, Citizens at $170, Piper $148), but the easy money's been made. R/R from here is muddier.
First major UI overhaul in 10+ years, rolling out to 100M+ households. Piper modeling the Marquee home screen ad unit at $500M+ in revenue at ~90% gross margins. That's a real margin asset, but the key uncertainty — which Piper itself flags — is whether this is incremental or shifted from existing in-stream surfaces. If $450M is cannibalized from current inventory, gross profit benefit is ~$140M flowing to FCF. Material, but it's not net new dollars to Roku.
"The firm estimates the Marquee home screen ad unit could generate over $500 million in revenue at approximately 90% gross margins."
(That's the bull case math. The bear case is that $500M includes a lot of reshuffled deck chairs.)
Political ads + World Cup (all 104 matches on FOX One via The Roku Channel at $19.99/mo) + USA 250 = 2026 has a stacked calendar. Piper's thesis is that home screen monetization offsets the 2027 growth hangover. $1B FCF vs $970M consensus is barely a beat — the market needs to believe the home screen ramps fast enough to fill the post-election air pocket.
Near 52-week highs, not chasing. If it pulls back to the $115-118 zone on any 2027 air-pocket anxiety, gets more interesting. Above $133, you're paying for perfection.
The dip is a gift, not a warning. APP OFF ~9% OVER THE PAST WEEK TO $559 on CloudX anxiety, but Oppenheimer just came out of a meeting with CloudX CEO Jim Payne and effectively said the threat is overhyped. CloudX isn't trying to replace MAX or AdMob — it's an alternative supply path. The structural moat around AppDiscovery, ROAS optimization, and MAX holds.
The 1Q26 numbers got lost in the CloudX noise. Revenue +59% YoY, BEAT GUIDE BY 5%, and April monthly revenue exceeded any Q4 month. On a $188B mkt cap name. Q2 guide of +4-6% reads soft but it's a seasonally tough comp and PMs are looking through it. Five PT hikes post-print clustered $580-$700 (Piper $665, Jefferies $700 Buy, GS $585 Neutral, Wolfe $580 Outperform, MS Overweight). The $700 Jefferies number is the bull-case bogey.
"AppLovin's moat around AppDiscovery, return on ad spend optimization, and MAX remains intact." — Oppenheimer post-CloudX mgmt meeting
CloudX adoption path runs through studio-level fragmentation within large publishers — specifically games that aren't dependent on ROAS campaigns. SDK friction is dropping fast with agentic coding tools, and publisher dissatisfaction with MAX/AdMob is real. Anchor demand from Meta and Liftoff gives CloudX a credible demand side. CloudX doesn't need to kill APP to hurt the multiple — it just needs to make the TAM expansion story harder to underwrite.
Morgan Stanley flagged the most interesting angle: 99% of APP's ads don't generate a conversion yet. That's the TAM expansion lever. PEG of 0.43 is stupid cheap for a 59% grower with 88% GROSS MARGINS — APP has the firepower to defend (mediation fee cuts, publisher rebates, exclusivity) if CloudX actually gains traction.
NVDA — CROSSED 20% OF TSMC REVENUE. Vera Rubin ramp is the structural story and pricing power is intact ($1.99/GPU/hr rentals for AMD vs NVDA still supply-constrained). But the stock is DOWN 8 OF 10 SESSIONS and the software rotation is real — this is positioning stress, not demand. The bogey here is whether $1T market cap holds through earnings; if GS most-shorted basket is up 25%, the squeeze risk is asymmetric to the upside.
AMD — MI300X ON EBAY AT $32K AND RENTABLE AT $1.99/HR. That is the data point that matters. NVDA premium is widening, not narrowing. AMD has no near-term catalyst to close the gap and the secondary market is screaming weak utilization. Bears are right on the share-loss narrative. Wait for a real product cycle inflection (MI400?) before fading this.
TSM — NVDA IS NOW 20% OF REVENUE. Customer concentration went from theoretical to confirmed. Intel 18A yields improving is the only real threat to the monopoly pricing power, but that's a 2027+ story. Glass core substrates extending the CoWoS roadmap is structural-positive but not a near-term catalyst. The real question is whether TSM can defend gross margins if NVDA ever decides to push back on wafer pricing.
GOOGL — $190B ANNUAL CAPEX AND $80B EQUITY FINANCING BEING DISCUSSED. That's the number. Internal FCF can't fund the buildout, which means dilution risk is real and margin compression is coming. The $920M/MO SPACEX LEASE ($30B over 32 months) confirms demand is strong but they're outsourcing capacity they can't build fast enough. Down 3 weeks straight on the rotation, but White House multi-vendor AI mandate is a quiet positive for gov deals.
DELL — +33% ON FRIDAY. LARGEST GAIN IN COMPANY HISTORY. First public Vera Rubin NVL72 delivery to CoreWeave is a flex moment. The HDD shortage call from the COO is a read-through to storage names (STX, WDC). AI server revenue is accelerating, not decelerating, and Dell is the prime beneficiary in the integrator layer.
MU — $1,600 PT SET. Goldman raised 2028 SK Hynix OP by 24% — that's a MASSIVE single-estimate revision. HBM cycle is accelerating, not peaking. DRAM/SSD contract prices up 200-300% since mid-2025. Earnings 6/24 is the catalyst. This is the cleanest memory long in the space, and any chip selloff is a buy.
MRVL — S&P 500 INCLUSION REPLACING POOL. Passive bid is the catalyst. Fundamental story (AI networking, custom silicon exposure) is intact. Down with semis on positioning rollover but the index inclusion provides a floor. Trade this as a buy-the-dip into the inclusion date.
VRT — GUIDING $13.75B FY25, +34% GROWTH. SoftBank €75B / 5GW France commitment is a direct read-through. This is the cleanest pure-play on AI DC power and cooling demand. Field Services hiring accelerating = visibility is extending. Multi-year setup is intact.
LITE — DOWN 5 STRAIGHT. Optical names have gone parabolic and are rolling over. GS TMT note flagged the technical setup. InP TAM expansion ($1.9B → $22.75B by 2030) is the long-term bull case, but near-term positioning unwind is the dominant force. Not a chase here — wait for 40-week moving average tests.
AAOI — Same playbook as LITE. Parabolic move rolling over, technical risk elevated. InP supply shortfall thesis is intact ($0.06B → $2.1B revenue by 2030 is a 35x). But the near-term tape is not your friend. Patience.
GLW — DOWN 4 OF 6 WEEKS. Optical/photonics rollover is the story. Technical weakness in a parabolic uptrend is dangerous. Not seeing a catalyst that reverses this near-term. Stay sidelined.
CRWV — FIRST VERA RUBIN NVL72 RACK DEPLOYED. This is the proof point that CoreWeave is the leading neocloud with early-access hardware. Validates the neocloud model and positions them ahead of competitors for next-gen workloads. Bullish.
LSCC — CEO TALKING $1T+ HYPERSCALER CAPEX IN 2027. Even if it's optimistic, the direction is consistent with everything else we're seeing. Upper bound for the AI infrastructure capex envelope just got raised. LSCC benefits on the FPGA side. Bullish.
STM — Major article expected, no details. Could be a catalyst on AI power semis exposure. Low-conviction trade into the print, but worth watching.
CRM — AI-native rivals can't rip out the legacy database. Stickiness is real. The market is overpricing the AI disruption risk to incumbent enterprise software. Salesforce data moat is the underappreciated angle here.
NOW — NOW ASSIST REVENUE TARGET BUMPED FROM $1B TO $1.5B FOR 2026. That's a 50% raise on the AI monetization line. Concrete proof enterprise AI software is actually generating revenue, not just demo-ware. This is the cleanest enterprise AI software read in the space.
RDDT — AI disintermediation risk is the structural bear case and traffic data would be the canary. No new data points today. The data licensing revenue partially hedges the advertising model. Neutral until we see traffic trends.
QCOM — Snapdragon C for $300+ laptops. Competes directly with NVDA N1X in the PC AI market. Product launch at Computex is the near-term catalyst. Neutral — needs to show traction with the OEM partners to matter.
WDC — SANDISK Q1: $5.95B REVENUE (+251% YOY), 78.3% GROSS MARGIN VS 22.5% LAST YEAR. Growth is ENTIRELY PRICE-DRIVEN, not volume. That is a memory super-cycle in full swing. HDD shortage flagged as next constraint. This is a melt-up name, not a value name.
FLEX — S&P 500 INCLUSION (REPLACING CAMPBELL'S). Passive bid catalyst. EMS exposure to AI servers and devices is a real but secondary story. Trade the inclusion, not the fundamentals.
Hearing the "HBM: High-Bandwidth Mistake" article is circulating on the buy-side this week. Thesis is that HBM value capture is unsustainable as customers (NVDA, AMD) build custom memory controllers and/or HBM supply finally catches up. Word is this is a late-cycle bear flag worth tracking — if the thesis gains traction, it puts a ceiling on MU and SK Hynix multiples regardless of the 2027-28 OP revisions.
Channel checks from neocloud sources continue to confirm: "demand is ahead of buildout, universally from every CSP, hyperscaler, neocloud." Rental rates per GPU-hr are "stable or increasing" per multiple contacts. This is the base case for cycle extension.
Vista Equity's Robert Smith citing insurance claims cost reduction from $8M (manual) to $200K (proprietary AI system) is the most concrete enterprise ROI data point we've seen. If this replicates at scale, it extends capex willingness at the hyperscaler layer. Hearing other PE portfolio companies are seeing similar order-of-magnitude cost reductions in legal, compliance, and customer service workflows.
Word is META may have the most incremental GW coming online of any AI lab in 2026 — potentially exceeding the ~6GW total projected for OpenAI and Anthropic combined by year-end. MTIA (Meta's custom silicon) allegedly faces less inference demand competition since Meta doesn't yet have an exploding agent business. Single-analyst view, unconfirmed. What would confirm: META releasing a model that benchmarks competitively with GPT-5/Claude Opus. What would disprove: MTIA utilization rates coming in below expectations.
Heard xAI struck a "pretty favorable short-term deal" on some capacity with optionality to reclaim it (90 days at any point). Subleasing signal — xAI views its compute needs as growing and wants flexibility to scale back up. This is a read on xAI's internal view of their trajectory.
Hearing concerns about SPV financing structures in the AI buildout. If capex is being financed through special purpose vehicles rather than balance sheet commitments, the financing is more fragile to demand disappointment. Narrative is circulating; no specific data yet. Worth monitoring.
Per one analyst (unconfirmed): labs have NOT yet hit $1B per pretraining run. $4/hr for GB300 represents long-term rental rates, not spot. If we haven't breached $1B/run yet, the compute scaling envelope has more room before hitting an economic ceiling — the $1B threshold is where only 3-4 labs can participate in frontier training.
OpenAI Sora team has reportedly transitioned to robotics. OpenAI Robotics is actively hiring. Read: video generation deprioritized, physical world applications prioritized. No revenue impact near-term, but signals where compute and talent are going structurally. Longer time horizon, potentially larger TAM.
Enterprise AI bottleneck is being confirmed across multiple sources: "The models are smart enough already. What is missing is the company-specific context locked in senior people's heads." Whoever cracks proprietary context extraction at enterprise scale has a direct path to the $200K cost tier. This is the data structure unlock problem — model capability is NOT the binding constraint.
Intel 18A yield improvement is being read as a potential inflection. Per one source referencing Lip-Bu Tan, Panther Lake shipping at "7x volume this Q" implies yields are no longer at 50%. Internal 18A demand rising; external supply limited to Apple, Google, Nvidia, AWS. If true, this is the first real threat to TSMC's monopoly in 5+ years. Watch for external customer tape-outs.