Good morning.
This is a ROTATION tape, not a beta tape. NVDA down 8 OF 10 SESSIONS, software had its best month since 2002, MOST SHORT BASKET +25% IN A STRAIGHT LINE. Futures flat (Nasdaq +0.3%) — mega-cap semis lagging, small-cap software ripping. Optical names ($AAOI, $LITE) parabolic and rolling over. The market's pricing a rotation fundamentals haven't justified. No major earnings overnight; AMZN, CRM, HPE, ESTC, OKTA in pre-market flow.
Asia: Goldman just dropped massive memory upgrades — SK Hynix 2028 op profit to $301B, Samsung to $405B. HBM cycle extends. SoftBank committed €75B to French AI infrastructure (5 GW, Schneider partner) — biggest European AI commitment to date, but energy costs are the binding constraint. META has the most incremental GW coming online in 2026 per GS TMT desk.
Four frames for the day: (1) HBM cycle is consensus-upgrade territory — MU (ARW:1), SNDK, TSM all leveraged, Goldman confirms the tape; (2) AI demand > supply — GPU rental rates holding, xAI's 90-day reclaim deal, token usage exponential, digestion narrative is dead; (3) Positioning stress is extreme — Most Short +25% straight line, software ripping while semis stall, this can run quarters; (4) PC demand destruction is real — DRAM/SSD UP 200-300% since mid-2025, sub-$600 devices facing extinction event, zero-sum allocation is here.
Jensen at GTC Taipei today — Feynman GPU details and Arm N1X PC chip expected. NVDA, ARM, GOOGL, AAPL in the mix.
We'll hit up NVDA, MU, and DELL first, then get to software (ESTC, OKTA, MDB, CRM, ADSK) and the storage complex (SNDK, NTAP, NET, RBRK).
THE VERDICT: DELL PRINTED A MONSTER AND THE STREET RESPONDED IN KIND — BUT THE STOCK ISN'T SCREAMINGLY CHEAP EITHER. Q1 FY27 was a 2-sigma beat-and-raise across every segment that matters: revenue $43.8B (+88% YOY, $10B+ ahead), EPS $4.86 vs. ~$3 expected, ISG +181% YOY, CSG +17% YOY. Mgmt raised FY27 revenue guide by $27B to $167B at the midpoint (+19%) and EPS by $5 to $17.90 (+29%). Six brokers lifted PTs in one week, two upgraded ratings. The print was so clean even Morgan Stanley — who carried an Underweight for god knows how long — threw in the towel. The tape is in the right place near $425, hugging 52-week highs, but the setup from here is "great company, harder entry" rather than "obvious long."
PT range is now $360-$700, and the distribution is ugly. The cluster sits at $440-500 (Bernstein, JPM, Mizuho, Goldman, MS all $448-500). Susquehanna is the $700 outlier — upgrade to Positive, PT up 5x from $138, pegged to FY27 AI server revenue at $60B and 27x FY28 EPS vibes. Melius $565 and Barclays $550 sit above the pack on the bull side. Truist is the lone Hold/$360 — only ~15% below spot — flagging 21x cal 2027 EPS against a 14% FY27-30 EPS CAGR. Stock is right around the bottom of the bull cluster and well above the bear. Not screamingly cheap. You're paying for the AI server growth at a premium multiple and betting it sustains.
The bull case is straightforward and increasingly consensus: AI servers are 1/3 of revenue, going to ~$36-60B run-rate by FY27 (+145% YOY), backlog is $50.3B, demand exceeds supply, and the traditional server business is a sleeper hit. Dell picked up 600bps of share in the 14G→16G/17G refresh as enterprises skipped 15G entirely. PowerStore density is 3x prior gen. Tier 2 cloud providers are giving Dell general-purpose wins. The services mix (40-45% of revenue, ~42% GM) is doing the heavy lifting on corporate margins — AI server GM is structurally lower but the dilution is more than offset. Agentic AI is creating CPU compute demand that benefits Dell's traditional server line, not just the AI business. CSG +17% YOY in a "memory-constrained" quarter is the most surprising line — mgmt is winning on price and content, not just units.
"Dell is executing across nearly all business lines, taking share of a growing AI spend pie, gaining share in traditional enterprise markets, and leveraging pricing power in ways that peers aren't." — Morgan Stanley
"Most of the upside surprise was servers — AI and traditional — driven by agentic AI CPU demand, share gains over competitors struggling to get parts, and some pull-in." — Bernstein
The bear case is essentially "this is incredible but the bar is now here and the multiple reflects most of it." Truist's math is clean: 21x cal 2027 EPS for 14% CAGR through 2030 is a 1.5x PEG on a hardware name in a cyclical refresh cycle. Morgan Stanley, despite upgrading, explicitly said the duration and magnitude of this growth is uncertain and core enterprise checks are "mixed with some degree of pull-in" — they're equalweighting to step aside and watch. The FY27 guide implies a ~10% H2/H1 revenue decline which is either conservatism on component availability or an early tell that the cycle's peak prints in Q1. The Rubin AI platform ramp in early 2027 could push some FY27 backlog into FY28 — a near-term air pocket. Memory/NAND/HDD inflation is a double-edged sword: helps ASPs now, risks demand destruction later. And AI server training share extending into inferencing "with structurally higher margins" is a forward claim that needs to be earned.
"The duration and magnitude of today's record growth is still uncertain... our prior thesis was wrong and we are stepping to the sidelines to evaluate the sustainability of the current spending cycle." — Morgan Stanley
"21x calendar 2027 earnings versus a 14% EPS CAGR through 2030 reflects premium valuation; rising memory and input costs could pressure future demand." — Truist
DELL's quarter is the strongest data point we've gotten on enterprise AI infrastructure spend durability — it's not just the hyperscalers, it's tier 2 clouds AND on-prem enterprises refreshing compute for agentic workloads. That's bullish for the whole stack: GPU systems (NVDA, AVGO), memory (MU — HBM + DRAM pricing power confirmed), networking (ANET, CRDO), power/cooling (VRT, GE VERNOVA). It also pressures the bear case on the broader TMT hardware cycle: durable, not pull-forward, at least through 2H 2026. The SMCI read-through is mixed — Dell is winning share, so SMCI's positioning gets harder (SMCI is the AI pure-play alternative; DELL is the diversified one). For the PC ecosystem, CSG strength + memory price commentary is a tailwind for HPQ and a nuanced read for WDC/SNDK (storage pricing up, PC units pressured). Biggest single-name read: MU is the cleanest derivative — every line item in DELL's report (server units, memory content, AI mix, storage density) supports the memory upcycle.
Verdict: "Show me" stock now. Cloud has decelerated three quarters in a row and ESTC just posted its first-ever sequential cloud revenue decline in a Q4 (the seasonally strongest quarter for consumption). Print wasn't a disaster — revenue beat by 0.8% at the high end, EPS $0.61 vs $0.56 cons — but the cloud miss is the kind of thing that keeps a multiple compressed for a 17% grower. Stock at $61, down meaningfully from where it traded pre-print.
The street lined up with PT trims. The cluster now sits $55-$85 with the stock at $61, so the average target is barely offering upside. Two Buys left, two Neutrals, one Hold. Notably UBS (highest PT) and Stifel (Buy) both kept their ratings — they believe the FY27 guide and RPO acceleration. The Hold/Neutral crew (DA Davidson, TD Cowen, Cantor) is more skeptical on near-term execution.
"Given the deceleration in the cloud business over the past three quarters, Elastic has become a 'show me' story. We do not expect the stock to re-rate materially higher until the company delivers several quarters of execution in the cloud business." — UBS
That's the framing PMs need to internalize. The bid isn't coming back until cloud re-accelerates.
Bull: FY27 guide above street at 14.5% midpoint. RPO acceleration is a leading indicator. AI tailwinds (CISA partnership, search/security/observability all inflecting). 76% GM. First time the 4Q cloud decel is a seasonal pattern disruption argument — could be transitory. Sales-Led growth +16% shows the underlying engine still works.
Bear: Three consecutive quarters of cloud deceleration. First-ever Q4 sequential cloud decline. 17% grower trading at a software multiple that already prices in re-acceleration. Until cloud proves the decel is over, this is a range-bound name with a low ceiling. And 2 of 5 covering analysts now have PTs below the current $61 print.
Stock needs to hold $60 support or it's a re-test of the lows. $65 is the bogey — reclaim it and you get a short squeeze. But with the street's average PT offering minimal upside and a "show me" narrative baked in, catalysts are binary: next cloud print is either the re-acceleration or the rug pull. No reason to chase here. Wait for the Q1 print (which mgmt already telegraphed is the low) — if cloud stabilizes or grows Q/Q, that's the re-rating setup.
"Consensus capital expenditure growth stands at 12% for fiscal year 2027 versus revenue growth of 27%."
Truist bumped AWS estimates for FY27+ to capture revenue from the $100B Anthropic and OpenAI partnerships. Underlying assumption: Trainium3 started shipping early 2026, Trainium4 launches H2 FY27. Custom silicon ramp is the engine.
Citizens' Andrew Boone is on the same page — "still early in estimates moving higher" for both AMZN and GOOGL. Says multiple expansion can persist as long as the cycle does. (Note: that's the standard "until it doesn't" caveat, not a real argument.)
Bear: You're buying +19% YTD sitting on the 52w high. Multiples have already expanded, premium is the base case, not the upside. $100B partnerships are commitments, not realized revenue. Capex still up 12% — not exactly a lean model. If AI demand cools even 6 months, the FY27 capex growth gets real ugly real fast.
(Sidebar: Blue Origin rocket blew up during testing — meant to deploy satellites for Amazon's Leo network. No satellites were aboard, so AMZN-direct impact is limited, but it's a reminder the space biz is messy.)
Verdict: The laggard setup just got a catalyst path. NVDA has been the AI trade's ugly duckling for months — underperforming semis peers into Computex, multiple compressed by ASIC fears. But Vera Rubin is officially in full production, agentic AI TAM is expanding into a $200B CPU opportunity, and the street's most aggressive bull (Tigress $425) is now 50% above where the stock trades. The r/r is asymmetric into year-end.
Goldman's James Schneider distilled the keynote into three investable threads, and honestly it's the cleanest framework we've seen:
1. PC offensive with Microsoft. NVDA is pushing harder into the traditional PC market alongside MSFT, and software partners are signing up. Windows on ARM gets a real tailwind here — this is a TAM expansion story most people aren't pricing.
2. Datacenter performance/cost moat. Goldman thinks NVDA maintains competitive dominance at "all but the largest hyperscalers." Translation: yes, Google/AWS/Meta will keep building custom silicon for their own workloads, but for everyone else — the enterprise, the neoclouds, the sovereign buyers — NVDA is still the answer.
3. Vera Rubin in full production. Systems shipping at scale to AI labs, cloud providers, and hyperscalers. The Vera CPU alone runs 1.8X faster than x86 with 88 Olympus cores and 1.2TB/s memory bandwidth. Rubin ramp tracking on schedule.
"Nvidia is pursuing the traditional PC market more aggressively alongside Microsoft, which could help drive momentum for Windows on ARM given a concerted push with software partners." — James Schneider, Goldman Sachs
Goldman reiterated Buy at a $285 PT. Tigress Financial Partners is the outlier bull — raised PT to $425, STRONG BUY — citing AI infrastructure positioning. Wolfe Research sees 30% CPU market growth driven by AI through 2028, with orchestration CPUs a major vector. Lynx Equity, which has been cautious on NVDA for nearly a year, finally flipped the tactical call:
"On a near-term basis, risk-reward tilts to the upside heading into the Computex keynote."
That matters. Lynx is conceding the catalyst path even while maintaining structural reservations. The stock delivered a 54% return over the past year but has been a clear laggard within AI semis recently — meaning the multiple has done the work for you. PEG sits at 0.29 (P/E 33). On a growth-adjusted basis, this is arguably the cheapest name in the AI complex.
Bull case (steelmanned): Vera Rubin in production + $200B agentic AI CPU TAM + networking optionality + PEG 0.29 + PC market re-entry. The "AI is big enough for multiple solutions" framing from Jensen's trillion-yuan feast dinner (Foxconn, TSMC, Delta, Compal, Pegatron) is the right narrative for a multiple re-rating. Add in the Groq licensing deal ($17B!) and Cadence/Samsung NVLink-C2C ecosystem expansion — the moat is widening, not narrowing.
Bear case (steelmanned): ASIC competition at hyperscalers is real and structural. Google TPU, AWS Trainium, MTIA — all in production, all getting better. Goldman concedes NVDA loses share at the "largest hyperscalers" which is precisely where the capex is concentrated. The stock has underperformed AI peers for months. If Rubin orders are lumpy or the networking segment commentary disappoints, this is a name that sells off on any wobble.
The setup heading into Q2 is straightforward: you've got a lagging stock, a confirmed product ramp, expanding TAM, and a sub-1 PEG. Tigress is $425, Goldman is $285, and the stock is somewhere in between. Not sure we can read too much into a single Computex keynote as a re-rating catalyst, but the Vera Rubin production confirmation is the kind of print that closes the bear thesis argument one data point at a time. We'd want to see hyperscaler capex commentary at the next earnings before getting aggressive, but the asymmetry is real.
Ecosystem reads: Groq raising $650M post-NVDA $17B licensing deal is quietly one of the most important data points in here — NVDA is monetizing inference through partnerships, not just fighting it. That's a different company than the one the bears are pricing.
Stock +18% IN A WEEK into the print Thursday AMC. Already a 48% GROWER LTM, 80% GPM, ~$16B MKT CAP at $78.63. Cyber resilience narrative, 97% subscription mix, consensus rating 1.1 — the name is LOADED into the quarter. Question isn't direction, it's the size of the beat and whether FY27 guide moves.
RBRK reports FQ1 (April-end) Thursday after the close. Subscription ARR is the line that drives the multiple — Rosenblatt modeling 33% sub growth vs the company's 25% guide. Wolfe thinks sub ARR runs >30% this year. Guggenheim sees less upside to ARR vs consensus but still calls for revenue beat. Rosenblatt projects at least $366M (+31% YoY) and notes RBRK has beaten quarterly revenue guide by ~9% on average over the TTM. Real track record.
Bull: Cyber resilience is a board-level conversation now and RBRK is the share-taker — tech differentiation, innovation cadence, expanding portfolio (Identity Resilience, RSC Sovereign, Rubrik Agent Cloud). 3 OF 5 PARTNERS BEAT PLAN in Guggenheim's channel work, pipelines positive across the board. 80% GPM means operating leverage is just starting to show. Rosenblatt explicitly says FY27 guide could prove conservative. If they raise Thursday, this rips.
Bear: +18% in a week, sitting near highs, $16B on a 31% top-line grower — the easy money's been made. Guggenheim flags beats may be smaller as the company scales past IPO. Two of five partners MISSED plan (one just building the practice, so take with a grain of salt). Pro Fair Value flags rich. Wide PT spread ($65 to $110) tells you the conviction isn't uniform.
Buy across the board. PTs cluster: Guggenheim $110 (the bull), Rosenblatt $90, Wolfe $70, BMO $64 (up from $58), Jefferies $65. So the bull case has $30+ of upside, the bear case has roughly $15 of downside from here. The $65-$70 cluster is the "we like it but we're not paying up" group — that's the bogey if the print is just OK.
MEDITECH partnership dropped this week — healthcare EHR cyber resilience. Incremental, not transformational. Adds to the leader narrative.
"Of five partners Guggenheim spoke with this quarter, three exceeded plan and two fell below while pipelines were positive." — Guggenheim
Long into the print has been the right call, full stop. The honest hedge in the sell-side chatter is the Guggenheim caveat on smaller beats and less ARR upside — that's the only bull who's pulling punches. Watch the SUB ARR line above all else. Revenue beat + raise + sub ARR >30% = squeeze. Sub ARR underwhelms even with a revenue beat, and this $16B mkt cap gets heavy fast. Risk/reward is fine for PMs who already have it. For those looking to add, Thursday is a coin-flip — better to wait for the post-print setup.
Bear: 124x P/E. Up 207% in a year — the easy beta's gone. Q1 FY27 has a chiller component shortage that's going to dent volumes (management qualifying new vendors). Near-term air pocket, not structural, but it gives the stock a reason to digest a move that's already priced for a lot of execution.
That's the whole paired trade in one sentence. Sell VRT, buy MOD, bet on share shift in cooling.
Bulls own the narrative. The $200B Anthropic deal, TPU going merchant later this year, and cloud backlog running well ahead of consensus have turned GOOGL into an "accelerate or die" tape — and Truist is betting it accelerates. Bears are stuck arguing multiples at $376 while the bull case targets $515.
Two firms in today, both bullish, but the PT spread tells you something. Truist bumps to $430 (from $415), Buy. Citizens holds $515 PT, Market Outperform. That $85 spread is the bull-bear debate in a single number.
The Truist call is a straight-up cloud call: consensus revenue estimates for Google Cloud are too low given March 31 backlog and recent deals. Their model has cloud growth ACCELERATING from 63% in Q1 to mid-80% by year-end. Consensus is modeling DECELERATION to the 50% range. That's a 30+ point gap on direction — the side of that bet is the whole trade.
Citizens is making the framework call. Hyperscalers remain demand-constrained through 2027, frontier models keep getting better, AI workflows are maturing. Their line worth sitting with:
"It remains early in the demand cycle for estimates to move higher."
Translation: we haven't even gotten to the estimate revision phase yet. The number's going higher before it goes lower.
Backlog is real revenue waiting to be recognized (5-6 year average duration per Truist). The Anthropic deal flows through BOTH TPU sales and Cloud consumption — double count of the same dollar in the best way. TPU going merchant this year opens a brand-new TAM and makes GOOGL a chip company in addition to everything else. Demand-constrained through 2027 means pricing power holds and capex is funded by Search free cash flow. Apollo and Blackstone putting together a $36B loan to Anthropic for Google AI chips is a tell — the financing ecosystem is building around GOOGL silicon.
Stock's already up 21% YTD. Multiples have expanded. The Truist setup requires consensus to catch UP to reality, and a lot of that catch-up is in the price. If cloud actually decels to the 50s the way consensus models, this is a "sell the print" name at $376, not a "buy the dip" name. The $430 PT is only ~14% upside — that's not a great r/r with vol where it is. The $515 case is the real bull thesis but you have to believe consensus is wrong AND multiples hold.
Cloud trajectory direction. Truist says accelerate, Street says decelerate. The next two prints resolve it. Until then, you're buying a story or fading a name — and the momentum is squarely with the buyers.
(Color: Waymo Ojai launching in SF/PHX/LA is a robotaxi datapoint but not moving the stock. ¥576.9B yen notes offering is balance sheet plumbing. The print that matters is cloud.)
VERDICT: Big beat-and-raise quietly buried by a deal nobody asked for. Stock NEAR 52-WEEK LOW ($225 vs $214 low) despite an 18% revenue print — that's a setup, not a thesis break.
The quarter was clean. EPS $2.99 vs $2.84 street. Revenue $1.93B vs $1.89B. 18.4% YOY GROWTH ON BOTH TOP LINE AND BILLINGS — street was modeling 16% and 9.5% respectively. Stifel called out construction as the standout: nearly $600M in revenue growing 20%+. That segment alone is doing what the bears said the whole company couldn't. Gross margins still 92%. Co raised FY27 guide by 0.5-1pp on both rev and billings. Not sure we can read too much into the stock's failure to pop on a clear beat — but the price action is telling you the market is focused on the $3.6B elephant in the room.
$3.6B — largest M&A in company history. Cloud-based maintenance/asset ops platform, $135M ARR growing 50%. Bulls are framing this as a strategic gap-fill (design + make + operate is the long-stated vision). Bears are framing this as a 18-20x sales multiple for a $135M revenue asset — that's $3.6B / $135M = 27x trailing, but on CY27 ramp it gets to ~20x (UBS) or 18x (Stifel) as growth scales. Still rich. RBC CUT PT FROM $335 TO $305 — still Outperform, but that $30 chop is the street pricing in deal dilution and integration risk.
UBS framed the strategic logic well:
"The push into operations addresses a known gap based on prior customer feedback."
Translation: customers have been telling them for years they need the ops layer. Co finally listened. Whether 18-20x was the right price to pay is the question. At 50% growth it pencils if the asset holds — but MaintainX is a different motion (SMB-heavy maintenance software) than ADSK's traditional seat-based AEC/PLM business. Integration is not a free option here.
Bull: Durable teens growth is now durable 18% growth. Construction is hitting on all cylinders. 92% gross margins. MaintainX fills a real product gap. GTM changes and workforce reductions are landing. Stifel's PT of $285 implies ~26% upside from here.
Bear: 46x P/E for a software co growing 18% — that's a multiple deserving of 25%+ growth, not high-teens. $3.6B cash deal at 18-20x sales is rich, even for a 50% grower. MaintainX brings SMB/maintenance exposure that could muddy the AEC premium narrative. Near 52-week low for a reason — momentum is broken, even if fundamentals aren't.
If you believe the core biz is mid-teens durable, this is a contrarian buy at $225 with $285-305 street PTs offering 25-35% upside. If you think the deal is the start of a value-destroying M&A spree, you sit on your hands. The beat was real. The multiple is the argument. We lean slightly bullish here — the setup is too clean to fade on a 20x sales deal that fills an obvious product gap. Sizes accordingly.
Mizuho bumps PT to $425 from $360, keeps Outperform. Stock at $353, basically tagging the 52W high of $356 — so we're up 184% YoY, 223% YTD, and the Street is STILL raising. That's a vibe.
The thesis is getting repetitive at this point, which is either a feature or a bug. Same playbook: handset weakness is "limited to lower-end models" while premium/flagship drives better v9 + CSS mix. Server CPU ramps at the usual cast of characters — Cobalt, Axion, Graviton, Vera, Grace — all contributing. Evercore flagged 140bps of server share gain in Q1, with Intel bleeding from 59% to 55%. That share-grab narrative is the engine, and it's still running.
Mizuho's also pointing to the 2027/2028 pipeline — in-house AGI CPU and a potential ASIC — as the next leg. PT of $425 = 2.7x their unchanged FY28 estimate. So the framework is "this is a 2027+ story" and the multiple is the bet.
412 P/E. Read that again. The bull case requires everything to go right — execution on AGI, the ASIC materializing, the server share gains continuing, the handset mix shift to v9/CSS holding. The bull case is also largely consensus. When the PM asks "what's the bear?" the honest answer is the bear is the multiple. At some point, you can't squeeze a 412 multiple higher on a stock that's already up 223% YTD without a meaningful beat-and-raise cadence. (And the Cerebras approach getting rejected tells you management is also looking for the next leg externally — the in-house pipeline might need help.)
This is a name everyone on the buyside owns. Long ARM is the trade. The asymmetry has compressed — you need either a monstrous print or a pullback to add. Not a sell here, but not a chase either. Bogey if you're long: the next 10% probably needs a fresh catalyst beyond "same thesis, higher PT."
Verdict: Restructuring story is done, AI-voice wedge is the new narrative, but TWLO has ALREADY DONE 62% IN A YEAR so positioning is the harder question here, not the fundamental one. Trading $190.64, right under the $203.71 52-WEEK HIGH, with the street clustering $210-250 PTs (TD Cowen $210 reiterated, Oppenheimer $235, Needham $250). Bogey is that 203 print — break it and the algo chase kicks in, fail it and you're fading a 62% move.
TWLO spent three years getting the house in order — 40% HEADCOUNT CUT from 2021 peak, new leadership installed, margins re-rated, FCF turned durable. That's the boring part and it's largely behind them. The new story is the AI-native customer pipeline, and voice is the wedge.
"Voice services are serving as the primary entry point for AI-native customers... Twilio positions itself as a provider for voice solutions that optimize AI input costs and model selection."
That's the real thesis line. TWLO has been in programmable voice for ~20 YEARS — they own the rails for low-latency voice inference, which is the killer use case for agentic AI. Signal conf launches (Conversation Orchestrator, Conversation Memory, re-architected Segment) extend the platform from comms API into context-aware orchestration. Upsell from voice → messaging → Segment is the GTM motion management is selling.
Q1 was a clean beat — EPS $1.50 VS $1.27 EST, REV $1.41B VS $1.34B EST. Both lines. 18 ANALYSTS revising earnings UP for the upcoming period post-print, which is the buy-side tell that estimates are still too low. InvestingPro health score 3.04/5 ("GREAT") confirms the cleanup is real, not accounting fluff.
Not much on the fundamental side. The risk is positioning — 62% RETURN IN A YEAR with the stock at 93% of its 52-week high means most of the easy money is in. Needham at $250 is the bull case (implied ~30% upside from here), TD Cowen at $210 is the bench mark (only ~10%). That's a tight upside/downside skew for a name that's already worked. The r/r favors ADDING ON PULLBACKS INTO THE $170S rather than chasing into the high.
Trade: Long the name, don't chase it. The Signal conf and Q1 print are buyable catalysts but the $203 52W high is the line in the sand — break it on volume, you chase; fail it, you let it come in.
THE MEMORY TRADE ISN'T DONE. MU PRINTING $970 NEAR HIGHS, +856% TRAILING TWELVE MONTHS, AND NOW SUSQUEHANNA LIFTS PT TO $1,750 FROM $600 — A NEAR 3X BUMP. THE STREET IS REPRICING THE DURATION OF THIS CYCLE, NOT JUST THE NEAR-TERM TAPE.
Consolidation: 4+ firms raised PTs in the last few weeks — Susquehanna $1,750 (Positive), DA Davidson $1,500 (Buy), Barclays $1,175, Mizuho $1,150 (Outperform). The cluster has shifted from a $200-300 bogey range to a $1,150-1,750 zone, with the Street collectively extending visibility into FY27. The thesis is uniform: tight supply through 2027, ASP strength holding, and the structural change in KV Cache offloading that's keeping the majors from pulling forward capex.
"Supply is now expected to remain tight through 2027, sustaining elevated margins and thus warranting valuation re-rating." — Susquehanna
Bull: Memory is a pyramid and the base keeps widening. AI inference workloads (KV Cache) are changing the demand shape — more high-bandwidth memory per accelerator, more HBM stacking, and the offloading dynamics discourage wafer adds. Supply stays constrained, margins stay fat, and the $1T mkt cap print is just the start. (Already happened, btw — first U.S. memory chipmaker to cross it.)
Bear: +856% in a year, PTs tripling, stock at 31x P/E... you're not buying a deep value name anymore. You're paying for perfection through 2027. One bad print, one hyperscaler guide-down on capex, one sign of inventory build — and the air pocket is violent. Not sure we can read too much into the KV Cache structural argument yet, but for now the tape doesn't care.
PM TAKE: If you're underweight into this rally, you're fighting a freight train with a broom. Sizing is the question, not direction. Trailing stops and respect the 52-wk high.
OKTA finally putting together a credible cRPO acceleration story. Stock +8% ABOVE prior 52w high of $107.84, printing $115.94 with a $20B mkt cap — the tape is voting yes on the setup. (For context, this name has been dead money for a while, so the breakout actually means something.)
cRPO +12% YoY — beat guide AND consensus of +10%, AND held the 12% from Q4. Stability matters here as much as the beat. EPS $0.91 vs $0.85E (7% beat), rev $765M vs $752M (1.7% beat). Not a blowout top-line, but the guide is doing the work. Q2 cRPO guide +11% YoY (1pt improvement) and they raised all of FY27. Clean print.
Four firms moved PTs higher — DA Davidson $130 (from $110), Cantor $125 (from $100), Stifel and Jefferies $120 (from $92-100), Mizuho $110. Cluster went from $92-110 to $110-130. That's a meaningful re-rating, not noise.
"The firm said it believes this factor, along with continued sales productivity improvements, meaningful new sales capacity ramping toward full productivity, and improving OIG traction, will lead to current remaining performance obligations growth acceleration." — DA Davidson
The AI Agents angle is the new narrative. Not sure we can read too much into "strong early traction" yet — everyone says that. But the new sales capacity ramping is the real under-the-hood story if it actually shows up in the back half. Watch the Q2 cRPO print — if it goes +12% or better instead of guiding down to 11%, the re-rate has legs. Below that and it's another head fake.
D.A. Davidson sticking with Buy, $45 PT, calling BOX an under-appreciated AI beneficiary. Stock at $26.96, so you're looking at ~67% upside to DA's bogey if you believe them. That's a wide spread vs. the Street — UBS is at $29 Neutral. The gap tells you consensus is skeptical on whether this reaccel is durable.
The setup: Q1 FY27 was a real print, not a fluke. EPS $0.37 vs. $0.36 expected, rev $306M vs. $296.5M. More importantly, growth re-accelerated to LOW TEENS with revenue up 10% CC. Billings +13% CC crushed implied guide of 6-7%. That's the kind of beat where you stop asking "is this a value trap?" and start asking "how much did the market price in?"
DA Davidson's angle is AI + Enterprise Advanced as a compounding mix shift. The note hits on pipeline growth and seat expansions, with EA driving pricing premiums — that's the right kind of operating leverage. They're plugging BOX into their "Best-of-Breed Bison" framework (10/12 criteria), framing it as a sustainable-moat compounder, not a one-quarter story. Stock trades at ~10x CY26 FCF, target is 18x. They're asking for a multiple re-rate on durability.
"D.A. Davidson reiterated a Buy rating and $45 price target on Box, citing the company's position as an artificial intelligence beneficiary... 56% gross profit margin."
UBS at $29 is the wall. Their read: reaccel to low-teens is nice, but is it enough to justify a re-rating in a tape that doesn't reward mid-teens content SaaS growth at premium multiples? UBS only nudged PT up by a buck despite the beat — that tells you they think the guide-up is mostly in. The bear argument: BOX is a cyclical IT budget beneficiary riding AI hype, and the multiple compression risk is real if growth normalizes back to single digits.
The billings beat is the key tell. 13% CC vs. 6-7% implied is a 2x over-deliver — that's not noise, that's pipeline conversion. If they can guide FY27 billings to high-teens, you get the multiple expansion DA is modeling. If next quarter billings normalizes, the $29s win. I'd want to see one more quarter of this behavior before fading UBS hard. R/R looks better long here than it has in 18 months.
Print tomorrow. Sit on your hands if you're long the 16% rip — this is a binary setup and the Street is split right down the middle.
GTLB goes into its Q1/FY27 print at $31.05, up 16% IN A WEEK, and the options market is pricing a 14% MOVE POST-ANNOUNCEMENT. That's a wide range for a name this size. The setup is messy: stock has ripped into the print, bears are already leaning in, and the analyst community is anything but aligned.
Rosenblatt stays the bull — $43 PT, Buy, looking for $233M SUBSCRIPTION REVENUE (+20% YOY) and 13% NON-GAAP OPERATING MARGIN. Their thesis: enterprise dev budgets and software activity remain healthy, even if raw coding headcount faces pressure from AI assistants. The offset — security testing, project management, compliance, ITOM — all flagged as solid. They like the Ultimate tier upgrade story as the Q1 driver.
On the other side, the bears are loud. Cantor cut to $27 (still Neutral), Mizuho cut to $26 (Neutral), both pointing at growth concerns despite AI-driven platform engagement tailwinds. Mizuho is specifically flagging execution risk on the Act 2 restructuring announced May 12. So you have a $26-$43 PT RANGE heading into a print with guidance already reiterated three weeks ago. Wide bogey dispersion = nobody really knows.
"Rosenblatt believes enterprise software development activity levels and budgets remain generally positive. Coding headcount growth faces pressure from AI-assisted coders, but areas such as security testing, project management, compliance, and ITOM remain healthy."
Three things to watch tomorrow beyond the headline numbers: (1) ULTIMATE TIER UPGRADES — that's the revenue mix story Rosenblatt is leaning on, and any softness there kills the bull case fast. (2) DUO AGENT PLATFORM ADOPTION — went GA mid-January, so we should finally get real usage metrics. AI agent platform is the narrative hook for 2026; if Duo traction is anemic, the multiple compresses. (3) ACT 2 RESTRUCTURING COMMENTARY — the May 12 announcement and reaffirmed guide is only three weeks old, so management won't have much new color, but tone matters. Bears are already calling the AI-era streamlining narrative hollow.
LTM revenue growth at 26% is the data point bulls hide behind. Decent, not great — and the deceleration into the 20% YOY print Rosenblatt is modeling is the bear's whole argument in one number.
Not a name to chase into the print after a 16% run. If you're playing the print, the 14% implied move means vol is reasonably priced — not cheap, not rich. The risk/reward on a long feels bogeyed by the rally; the risk/reward on a short feels bogeyed by a low bar (guide was just reiterated, comp is easy). R/R is marginal both ways. FADE THE PRINT, DON'T FRONT-RUN IT — wait for the post-call repositioning when Ultimate tier commentary and Duo metrics give you actual signal. The Conference board stuff with Staples and Ross at BofA Tech later this month is probably a better entry for the longer-term thesis than tomorrow's number.
Valuation catches up, fundamentals don't. That's the read. DA Davidson cut PT to $12 from $13 (kept Neutral) even after UiPath printed a clean revenue beat and guided better on ARR. Q1 FY27: rev $418.4M vs $397.2M est (+5.3%), EPS $0.15 vs $0.16 (small miss), 15% LTM growth, 83% gross margin. Solid quarter. The cut is purely a multiple call — and it's the right one.
Firm raised estimates off the print but still took the target down. Translation: they like the trajectory, hate the entry. AI agent narrative has dragged this thing higher and DA is signaling it's not earned yet.
"We believe growth is adequately priced in today as traction with new AI products remains early in providing more significant upside to revenue."
BofA did the same dance in reverse — raised PT to $13 from $12 while keeping Underperform. Underperform with a raised target is just bear-chasing. PT compression across the Street from $13 down to $12 is the tell: nobody wants to step in front of this name at current levels even with the print in the rearview.
Revenue beat, ARR guide up, 15% growth re-accelerating, 83% GM, raised own estimates — and the Street still can't get to a Buy. Either the AI agent optionality is already in the price (DA's view, and probably right) or the market's giving zero credit for it. Hard to be long here into the prints. Not a short either, but not a name we're fighting for.
BOGEY: $12 PT cluster from the cautious bulls, $13 from the bears. Trade is the range, not the trend.
Verdict: Decent quarter, soft guide, sentiment split. Stock got punished 18% AH for the right reasons.
Q1 net new ARR of $43.9M (+57% YoY headline, but ~15% adjusted for Attivo churn/pushouts) met the bull camp but revenue landed below guide midpoint on backend-loaded bookings and a partner ramp miss. The Q2 net new ARR callback came in light vs prior expectations — that's the print that matters here. FY revenue guide held, op margin guide RAISED ~50bps, ARR +23% — so the operating story is fine, the growth story is just uninspiring at 15% organic.
Street mostly stayed long despite the drawdown. PT cluster post-print sits $20-24: Rosenblatt $20 Buy, Cantor $24 OW, TD Cowen $22 Buy, Citizens $23 MO, BofA upgraded to Buy at $20. DA Davidson stands alone, cutting to $15 Neutral — lowest on the Street, the dissenter.
(Pre-print $15.83, ~$5.4B mkt cap. AH implied ~$13. Mid-cap purgatory.)
"Revenue came in below the midpoint of guidance due to backend loaded bookings and a partner deal with a ramping revenue component that missed expectations."
Neutral-with-a-stale-PT foghorn from UBS while the stock is already at the highs. UBS lifted to $160 from $113 (Neutral maintained) — a 41% PT bump that still only implies ~12% upside from current levels (stock at $143.65, within striking distance of its 52W high). Barclays, for the record, is out at $199 Overweight. That PT spread is doing a lot of work right now.
Real beat, not a pull-in story. Revenue $1.95B (+2% vs UBS), GM 70.5% (vs 70.0%), EPS $2.43 (vs UBS $2.26, guide $2.21-2.31). The important line: all-flash $1.22B, +18% YoY, with management saying they saw minimal pull-in impact and the growth was largely organic. That's the cleanest read you can get on enterprise storage demand right now.
Here's the rub. UBS is modeling F27 product GM <53% in Q1 and full-year GM ~69.2% — that's roughly 200 BPS OF COMPRESSION YoY. They're still Neutral with a PT barely above spot while modeling a meaningful margin step-down. The bull case has to be: mix shift to public cloud + high-margin support revenue offsets product margin pressure, and the all-flash cycle has more legs in core storage where F27 product revenue guided up HSD. The bear case is: you've already had your multiple expansion (+46% in a year, P/E 32.1, P/B 20.8 — not cheap for a storage vendor), and 200bps of GM compression on a 2x P/S base is a tough setup into F27.
"The mix toward public cloud and high-margin support revenue is expected to help offset margin pressure across the product segment." — UBS
Read that line carefully. Translation: "We're modeling the offset works but we're not sure enough to be bullish at the current multiple."
Trade, not a thesis. The setup is messy — fundamentally clean quarter, but you're buying a stock that just ripped 46% into a year where sell-side is modeling real GM compression. UBS Neutral at $160 is effectively a "we missed it" call. If you're long, you're playing for the all-flash cycle to extend and the mix-shift margin defense to hold. If you're not in, hard to chase here into a 200bp GM headwind on a 32x P/E.
Atlas printed 29%, 3P ABOVE the Street, and consumption accelerated Q/Q — that's the tape. Mgmt didn't just hold the line on FY guide, they RAISED IT by more than the Q1 beat, and Q2 Atlas guide of ~26% looks conservative given the Q1 run-rate. MDB is a story where the beats keep getting bigger and the guide keeps ratcheting up. Q1 rev of $688M cleared the $665M consensus by ~3.5%, non-GAAP EPS of $1.32 smoked the $1.19 print, and opex leverage showed up with margins +200bps YoY (150BPS ahead of model).
The Street responded in unison. Six firms lifted PTs into a $350-435 cluster, with Stifel leading the way to $435 FROM $330 (Buy), Cantor at $416 (OW), Needham at $400, BofA at $390, JMP at $366 (MOP), and UBS at $350. Old prior cluster was materially lower — this isn't nibbling, this is a full repricing of the Atlas growth algorithm.
"Consumption trends accelerated quarter-over-quarter and exceeded management's forecasts, driven by strong usage among core workloads, accelerating Frontier Model Lab consumption and growing AI Native applications." — Stifel
That quote is the whole bull case in two sentences. Frontier Model Labs — the OpenAI / Anthropic / xAI tier — ramping Atlas usage is the new variable. If the frontier labs are building inference or training pipelines on top of Mongo, that's a TAM expansion story, not just a database growth story.
Bull: 121% NRR, cRPO +70% YoY, 2,500+ net new logos, and AI Native apps are a greenfield workload. The "database of choice for AI apps" narrative has legs if Frontier Lab usage continues to scale. RPO visibility alone de-risks the next 4-6 quarters.
Bear: Stock's already done the work — +72% TRAILING YEAR — and you're paying for the AI optionality that's not in the guide yet. Atlas decelerating to ~26% in Q2 (from 29%) gives the bears something to chew on even if Stifel sees 3% upside. Multiple expansion from here requires the AI Native app cohort to show up in revenue mix, not just consumption commentary. Not sure we can read too much into the FML comment yet — one quarter of acceleration is a data point, not a trend.
This is a name where the fundamental momentum is clearly positive and the analyst community just caught up. Risk/reward at $400+ levels is tighter than it was at $300, but the path of least resistance is still higher into Q2 prints. The 26% Atlas guide is the floor — if consumption stays at Q1's clip, Q2 will be another beat-and-raise setup. The real question is whether the stock can digest a +72% move and keep grinding, or whether it needs to base.
Mongodb stock price target raised to $435 from $330 at Stifel (Yahoo Finance)
MongoDB Q1 FY27 results, consensus data, and additional analyst price targets (company filings; analyst notes via Investing.com)
Cloudflare prints to the tape at $241.82, basically AT the Barclays $250 PT going into Investor Day. That's the whole setup in one line - the easy money's been made, and tomorrow is a "show me" event. Stock +46% over the past year on a 32% revenue grower that's compounding into a $5B exit run-rate target for FY28. The question isn't whether the story is good. The question is whether the stock gives you anything for owning it from here.
Barclays stays OW, $250 PT unchanged, ~20x FY28 sales of $4.56B. Three things they want to see (or already see):
That's the bull puke right there - 30% top-line + 6pts of margin from headcount = operating leverage that justifies the multiple even at $250.
Bull: Operating leverage has been under-appreciated. Headcount cut is a real margin unlock, not narrative fluff. Act 3 is the next leg of growth and it's already monetizing even at lower GM. Anthropic/Claude integration (Glasswing, Managed Agents) is a real product wedge into the AI agent infrastructure stack. TAM refresh almost certainly shows share gains across multiple verticals.
Bear: Stock is AT the Street PT cluster. Stifel actually cut to $260 from $275 - constructive but trimming on valuation, which tells you where the buyside is. JLens opposing board seats over extremism-related services is governance noise nobody wants to deal with at the highs. Act 3 margin drag is real and shows up in GM first, long before the revenue mix shift helps. And if Investor Day doesn't materially raise the FY28 EBIT margin framework, this is a sell-the-news setup into $250 resistance.
Boring answer: don't fade the name into the event, don't chase it either. Stifel's $260 cut is the tell - even the bulls are running tighter targets. If the day delivers a higher LT EBIT margin framework or a TAM number north of $250B, you get a relief breakout. If they just reiterate the existing 75-77% GM target and the 20% EBIT margin trajectory, this stalls at $250 and you're back to $220. Light position, wait for the print.
Verdict: AI narrative accelerating, monetization story still has to prove out. TD Cowen keeps Buy / $240 (stock $191, so ~26% bogey) post the agentic strategy webinar, and the most interesting number in the print isn't revenue — it's the AGENTIC WORK UNIT CONSUMPTION +111% QOQ vs the prior four quarters of 57/62/34/21%. That's a clean inflection, not noise. Whether it drops through to ARR is the question every PM is asking right now.
The bull case is Headless 360 + Slack-as-OS. Decouples platform from UI, lets agents orchestrate across data/app/agentic/engagement layers. Slackbot is the orchestration hook. If Salesforce can become the connective tissue for enterprise agent workflows, the consumption curve compounds and monetization follows. Agentforce ARR already crossed $1B. CC rev growth of 11.6% beat the Street, op margin 34.8% vs 33.4% expected. Real business, not vaporware.
"The combination of Headless 360 and Slack should drive higher platform utilization, broader ecosystem relevance and longer-term monetization opportunities. TD Cowen maintained its constructive view on Salesforce's AI opportunity, though revenue visibility remains early."
That last clause is the tell. Bull/bear is basically a timing debate, not a thesis debate. PT cluster is wide: Cantor $250 (OW), TD Cowen $240 (Buy), Freedom $230 (Buy, cut from $360 — ouch), BMO $215 (cut from $225). The Freedom cut is the most interesting action — someone with a $360 PT trimmed to $230, which means they still like it but the AI monetization timeline got pushed out in their model. That's the bear case in one data point: the AWU curve is great, but ARR conversion lags and the Street's patience isn't infinite.
Not sure we can read too much into one quarter of 111% AWU growth, but the trend in those four data points is the cleanest acceleration we've seen from CRM in a while. Worth putting on a watchlist if you don't already have a position. Skews toward a Q2/Q3 catalyst trade on the next consumption update.
The most controversial long in TMT right now and the bears keep getting vaporized. SNDK up ~4,150% in a year (not a typo), trading $1,669 against a 52-week high of $1,697, and the street is still raising numbers into a melt-up. Susquehanna just took their PT to $3,250 from $2,000 — that implies ~95% upside from here on a stock that already 40x'd. Read that again.
Susquehanna's move is the latest in a coordinated series of upward revisions. Mizuho now modeling FY27 revenue of $45.3B vs $43.6B consensus, EPS of $184.95 vs $180.14. Barclays went Overweight, PT to $2,300. Cantor at $1,800, Bernstein at $1,700. S&P even upgraded the credit to BB+ on the net cash shift. The PT cluster sits $1,700-$3,250 now — average up meaningfully, low end getting dragged higher. Nobody wants to be the guy with the sell note after a 4,000% rip.
Susquehanna's call is the cleanest articulation of the bull case out there:
"Supply is now expected to remain tight through 2027, sustaining elevated margins... incremental demand from AI inferencing is driving stronger bit demand and supporting further average selling price upside into the second half of 2026."
That's the whole ballgame. KV cache offloading dynamics have made wafer capacity adds uneconomic, so the supply side stays disciplined even as AI inferencing pulls incremental NAND bits through enterprise SSDs. CSPs are the marginal buyer, and Q2→Q3 enterprise SSD demand is inflecting. Gross margins at 56%, revenue +83% LTM — these are not commodity memory numbers, this is a structurally tighter cycle than the last three.
Two things give us pause even as we respect the tape: (1) Taiwan module houses carrying ~quarter of inventory — that's the classic channel-stuff tell and worth watching into the Q2 print; (2) the stock is priced for a 2027-2028 cycle peak already, so any ASP wobble or inventory digestion prints as a 30-40% drawdown, not a 10% one. The 4,153% return is the trade; the asymmetry from here is materially worse than it was six months ago.
Bottom line: Still a melt-up, still a long, but the r/r has compressed hard. We trim into $1,800, hold the core into the Q2 print when we'll see if the channel inventory is real or noise.
Jefferies bumps PT to $56 from $44, Buy — but the print's already running. Sponsor is selling through it, so the easy money's been made.
Jefferies cites Q3 FY26 strength and "evidence of share gains" — data center orders across coverage up 100%+, FPS orders +308%. That's the kind of print that gets you multiple expansion. Backlog extending in duration is the key tell here: not just volume, but visibility. Custom niche biz = high leverage to electrification, and they raised FY28 sales and EBITDA by 15% and 16%, respectively. Margin expansion still on the table. FCF flips meaningfully positive FY27. Bull case writes itself.
BUT — two offerings back-to-back. 42.28M shares at $47, then a 35M secondary, with Neos Partners (sponsor) selling ~23.7M and Forgent itself chipping in ~11.3M. That hit the stock -8.5% AH. When your PE sponsor is distributing 50M+ shares into a +64% YTD tape, that's not a coincidence — they're monetizing the rally. New money has to absorb all that paper, and the float just got meaningfully larger.
"The company has high leverage to the electrification theme through its custom niche business."
Strongest line in the note and it captures the thesis cleanly. But the leverage works both ways once the secondary clears and the comp gets harder.
Positioning read: $56 PT vs $47 offer price = ~19% upside, and the stock's already in the upper third of that range. PMs who missed the 64% YTD move aren't going to chase into a fresh overhang. This is a "let the secondary digest, revisit on weakness" name, not a buy-the-dip-today name. Jefferies' raise likely caps the stock near-term until the order book clears and the comp set-up for FY27 rebuilds.
UBS CUTS VERI TO A $2.50 PT — DOWN FROM $6.00. OOF.
That's the headline. Street isn't underwriting the VDR dream yet, and frankly, after a 31% revenue miss last quarter, who can blame them. Stock at $2.12, dancing on its 52-week low of $1.22. Small-cap AI infra name with a real narrative problem: the story is interesting, the prints are horrid.
UBS initiates Neutral. Likes the transformation, likes the VDR setup — but can't get comfortable with the path to profitability given how lumpy and consumption-based the Data Refinery deals are. Translation: great optionality, zero visibility, and the recent execution hasn't inspired confidence.
"Veritone is better positioned than in the past to scale AI solutions after years of transformation... it is monitoring the path to profitability given the complexity of forecasting the timing of Veritone Data Refinery deals, which tend to be larger in dollar value and almost entirely consumption-based."
The bull case: VDR launched 2024, turns unstructured data into AI-ready assets. TAM is real — $2.6B IN 2024 → $17B BY 2032 training data market. UBS models 25% revenue CAGR 2025-2028, with FY26 GROWTH OF 49%. Balance sheet flexibility has improved. On a 1.5x 2028E EV/Sales (2-year forward), you're in the lower end of the ad tech comp range, and well below the 3-year historical band of 0.6x-5.4x. So the multiple compression is mostly the market pricing in execution risk, not TAM skepticism.
The bear case: REVENUE PRINTED $20.26M VS $29.31M EXPECTED — 31% SHORT. EPS beat was nice (-$0.128 vs -$0.145), but you don't beat on the cost line when you miss the top line by 30%. And VDR pipeline visibility is 2-3 months ahead of delivery. For a sub-$3 stock with a 2.5-month forward window, you can see why the multiple has compressed.
This is a show-me story now. The AI data infra narrative is intact, TAM math checks out, and the multiple is broken — but broken cheap isn't the same as broken low. We'd want to see a clean VDR print before underwriting the re-rating. $1.22 52-week low is the obvious technical bogey; a breach there and it's a different conversation. Until then, Neutral feels right. Not a short, not a long — a watchlist name for the next inflection.
Stock ripped +19% IN THE PAST WEEK and there's real meat here. Enovix owns FEOC-compliant cell and pack capacity (read: DoD-friendly, China-free supply chain) and just printed a Q1 beat — revenue $7.6M vs $6.95M Street, EPS -$0.14 vs -$0.16, with Korea defense/industrial shipments doing the work. LTM revenue growth is 50%, and Northland's still Outperform at a $16 PT (~87% upside from $8.54) calling a revenue inflection on silicon-anode across drones, phones, and eyewear.
Cantor held $25 Overweight — lead smartphone customer aligned on a silicon-specific qualification framework, another customer acknowledged the legacy 0.7C cycle-life test is the wrong benchmark for silicon chemistry. That's real commercial de-risking, not hand-waving. But Benchmark just CUT to $15 from $25, maintained Buy, citing valuation. So you're looking at a $10-WIDE PT SPREAD with the stock already up huge — easy money's been made, the next leg needs execution, not narrative.
Drone + FEOC compliance = structural, not fashion. Smartphone qualifications are moving in the right direction. But chasing a stock up 19% into a wide PT spread on light coverage is terrible r/r. Position-building on weakness, not FOMO into the rip.
Neutral, but the setup is interesting. Goldman initiates with a $17 PT on Octave Intelligence, stock at $17.20 — basically a fair value call right at the print. The selloff has already done some of the work: shares down 21% in the past week, sitting just above the 52-week low of $16.65. So the question isn't whether OCTV is cheap, it's whether the new standalone story has legs.
The thesis is a familiar spinout playbook. OCTV sells software into industrial verticals to design, build, and maintain critical infrastructure — think mission-critical, not sexy consumer. Now free from its former parent, the bull case is two-fold: (1) macro tailwind from an uptick in infrastructure spend, possibly juiced by faster innovation cycles at industrial customers, and (2) better unit economics from an uplevelled cross-sell motion and a SaaS transition that the parent probably never let them execute properly.
"Goldman Sachs expects the next 12 months to be a period of transition for the company as it executes on these more focused investments. The firm said it will look for proof points that new initiatives are translating into better growth."
That last line is the whole ballgame. "Proof points" is analyst-speak for "we're not buying the story yet, show us." NTM is a transition year — that means the stock is going to live and die on incremental data: cross-sell attach rates, SaaS mix shift, deal velocity into the industrial customer base. Either the standalone narrative starts showing up in the numbers, or this thing drifts lower from the 52-week low and Goldman gets to upgrade into a real story.
The bull case: $4.62B market cap, infrastructure spending cycle, focused execution, SaaS economics. If they nail the first 2-3 quarters as a standalone, the multiple re-rates fast.
The bear case: It's a spinout that just got dumped by the parent for a reason. The 21% weekly decline suggests someone (or someones) knows something, or it's just a forced-seller dynamic from the separation. Industrial software is also a grind — long sales cycles, lumpy bookings. The "transition year" framing gives management a 12-month hall pass to disappoint.
Bottom line: Not a buy here, not a sell either. This goes on the watchlist for the Q2 print — first standalone quarter. Need to see the SaaS transition land and cross-sell motion show up in the numbers before we get constructive. If it takes out the 52-week low on volume, the thesis is dead and the PMs move on.
The verdict: ACN's a melting ice cube, not a value stock. The 39% drawdown and 12X NTM P/E look like a gift, but Truist's tapping out and the reason matters — they're not worried about the quarter, they're worried about the model. Downgrade Buy → Hold, $210 PT (only ~12% upside from $187), citing pressured budgets, AI pure-play competition, and the big one: revenue cannibalization from the very technology ACN's deploying.
The numbers that anchor the bull case: 12X NTM P/E, 11% FCF YIELD, ~$36 in "fair value" per the platform. That's a real yield, not a fake one. The numbers that anchor the bear case: stock DOWN 39% YOY and 12X is what you pay for a business model under structural threat, not a cyclical low.
"Accenture has a significant portion of its business exposed to AI-driven revenue cannibalization."
That's the line. That's the whole call. Headcount-based pricing → token-based pricing = utilization goes to zero on the replaced workstream. ACN is selling the rope. Every agent deployed for a client is a consultant hour that doesn't get billed.
The partnership parade — Mitsubishi Chemical JV (Rix Business Partners in Tokyo), HUMAIN in Saudi, Aera Tech investment, OpenAI federal, ServiceNow agentic AI program — is all strategically logical, all real, all ZERO help on the pricing model problem in the near term. Truist's framing is the right one: "well positioned to reshape its business" but "limited near-term upside as it transforms." That's a show-me rating, not a trust-the-process rating.
Peer read-throughs are the overhang nobody's pricing. If INFY or CTSH guides softer on AI cannibalization, ACN's the next shoe. (Not sure INFY's clean enough to be the canary here — TCS might be the real tell.)
Verdict: Tactical squeeze candidate into the print, but Mizuho's $15 anchor keeps a hard floor on the long thesis. Stock at $27.84, +302% YTD (ridiculous), -12% on the week. The setup is classic — momentum chasers versus a sell-side anchor that's refusing to chase.
Mizuho reiterated Underperform, $15 PT, and flat out said they're not moving until they see order momentum. The stock sits 85% above that target. They're not stupid — they already model 10% BESS share in data centers, and even the bull case (the Siemens/NVIDIA DSX win) only moves the needle $2-4/share, which Mizuho has already told you to expect.
Siemens dropped an NVIDIA DSX Vera Rubin NVL72 reference design — a ~103MW AI factory blueprint — with Fluence SmartStack embedded directly into the electrical architecture, BOM, and microgrid controls. This is the punchline:
"Fluence is the only named battery energy storage system partner across seven infrastructure OEM designs."
Sole-named. Out of 7 OEMs. That's a real win on paper. The 2-3 hour battery duration spec is also above Mizuho's prior ~1hr assumption — explains the $2-4/share kicker they flagged.
Longs own the tape into any further AI-data-center-BESS headlines. The catalyst path is real but back-half weighted. Bears have a clean $15 PT to anchor against. R/R for a fresh long here is meh — you're paying for the win that's already partly in the model. Mizuho's patience is the tell; we should match it.
Parabolic name hits its first real speed bump — and the tape barely flinched. BMO stuck with Market Perform and a $279 PT (stock $275), flagging that the Green Chile lateral pipeline supplying gas to Oracle's Project Jupiter may not qualify under FERC's blanket certificate program. That's the kind of thing that should matter when the underlying project requires 2.45 GW of Bloom fuel cells — a real revenue stream, not a theoretical one. But BE is up ~1,462% over the past year, so anyone who's ridden this probably has a hard time adding here.
BMO's base case: no hit to 2026 guidance. Fine. But the second-order question is whether pipeline hiccups push fuel cell deliveries right, which would compress the story into a tighter window and raise the bar on execution. With the stock trading basically at BMO's PT, the asymmetry isn't great.
"third party analysis suggesting the Green Chile lateral pipeline to Project Jupiter has not met necessary requirements to qualify under FERC's blanket certificate program" — BMO Capital
The rest of the Street is decidedly more constructive post-Q1. Cluster of raises: BTIG $295, Evercore $295, Barclays $254 (Equalweight), TD Cowen $235, Roth/MKM $225. Strong Q1 beat, 130% YoY growth, raised FY26 guide. Data center demand story intact, Oracle partnership expanded.
Not fighting the trend at 1,462% in a year. If you don't own it, this FERC wrinkle isn't an entry — it's a reason to wait for a real pullback. If you do own it, sizing is the conversation. The narrative's not broken, but the easy money is.
Needham lifts to the street-high (range $258-$310) after mgmt meetings — conviction up on growth trajectory, margin opportunity, and positioning in A&D, semi cap, and emerging data center apps. Stock at $268, so ~15% to Needham's bogey on a name that's already up 104% OVER THE TRAILING YEAR. Stifel also reiterated Buy at $280 post-Wisconsin site visit — not just one analyst getting excited, the cohort is converging. This is PT raise chasing momentum, not calling a top.
Fiscal Q2 2026 print was solid — EPS $2.05 VS $1.88 EST, REV $1.164B VS $1.13B EST — yet shares dumped after-hours. That's positioning, not fundamentals. Means expectations are locked and loaded going into the back half. Mid-teens+ rev growth is the bull case; macro visibility is the only real flag Needham called out.
"Management indicated the year is shaping up to be one of the best in the company's history. New program ramps, share gains, and improving end-market demand are converging to deliver mid-teens plus revenue growth."
Verdict: Own on weakness, don't chase. On a stock that's already done 100%+ in a year, you want to be a buyer on the next 5-7% pullback, not paying $310. R/R is better after a cooldown than at the highs.
Buy-rated at $42 PT but the path there is ugly. C-H cut to $42 from $48 after a Q2 print that underwhelmed AND a Q3 guide that underwhelmed. Stock at $32.07, DOWN 34% IN A WEEK after a 95% RIP over the last year. Classic buy-side question: is this a setup or a falling knife? Lean setup, but you're paying for FY27 — not buying it today.
Q2 was a clean miss on both lines. Adj EPS $0.42 vs $0.53 consensus — $0.11 light. Revenue $209.9M vs $216.7M est, DOWN 0.5% YoY. Q3 guide: $207-215M revenue (midpoint $211M) and $0.39-0.45 EPS — both below the $218.5M Street number. So you're getting two quarters of "less than" before any green shoots. The drivers were the usual photomask cyclical pain cocktail: IC softness from elevated fab utilization, memory supply constraints, and geopolitical noise pushing out design releases.
"Management noted early signs of tape-out recovery beginning in May."
That line is the whole bull case in one sentence. Tape-outs lead masks. If May is the inflection, the Q3 print could be the last of the bad ones.
Three things actually working underneath the noise:
1. Allen, TX ramp — facility entered qualification mask production, initial revenue late FY26. Migration of mainstream work out of Boise frees capacity for higher-ASP advanced node masks. Meaningful contribution 2027+. This is the multiple expansion story. 2. FPD — $62M quarter, UP 13% YoY, one of the strongest on record. AMOLED strength in China, Korea seasonal reaccel. They also installed a new FPD mask writer — described as the most advanced globally and first of its kind deployed. Real differentiator. 3. Valuation — 11.96x P/E with ~$11 net cash per share. C-H's $42 PT is 16x FY27E EPS of $1.98 plus that net cash. So you're underwriting a recovery that hasn't started yet.
"Near-term headwinds... early signs of tape-out recovery beginning in May."
PMs looking at a 34% drawdown on a name that was up 95% YTD — that's a tape event, not a fundamental re-rating. If you're long, you're sized for volatility and waiting on the July/Aug tape-out data points. If you're flat, $32 is a level, not a catalyst. Bogey is whether Q3 guide holds or gets cut again in 30 days.
Verdict: Mizuho stays Outperform, $52 PT, and honestly the thesis doesn't need a lot of dressing. NiSource just stacked Alphabet as a new long-term energy customer on top of an expanded Amazon deal, taking total signed data center capacity to ~4GW. This is a real number with real counterparties — not a press release with "MoUs." That's the story.
The headline move: NI signed a long-term energy supply agreement with Alphabet (new) and expanded an existing contract with Amazon, which accelerates site energization. Aggregate signed capacity is now ~4GW with more in active negotiations. Mizuho flags that customer savings have expanded too, which they view as the unlock for continued regulatory support — important because every utility data center deal lives or dies at the state commission. Management just raised long-term consolidated adjusted EPS CAGR to 9-10% THROUGH 2033 (up from 8-9%) following Q1 2026, with the GenCo merchant generation unit expected to contribute $0.25-0.35/SHARE BY 2030. That's a real earnings stream, not a fluffy aspirational slide.
"Mizuho views NiSource's growing data center pipeline as a meaningful growth lever positioning the company well versus utility peers on earnings visibility."
Q1 print was a $1.06 EPS BEAT vs $1.05 consensus — fine. Revenue was $2.28B VS $2.6B EXPECTED, A -12% MISS which is horrid on the surface but for a rate-regulated utility the top line is a function of pass-throughs and fuel, not the earnings story. Nobody's selling this stock on revenue. NI also tapped the market for $1.25B IN SENIOR NOTES ($500M of 4.750% due '31 and $750M of 5.300% due '36) — the kind of stuff a capex-heavy utility is doing constantly to fund the data center buildout. Quarterly dividend held at $0.30, payable Aug 20.
Barclays Lenschow kicks off coverage at OW, $350 PT — the high end of the Street cluster. Stock at $297.80, so ~17% upside to new PT. The angle isn't new (software mix shift, Red Hat franchise, regulated sticky base) but Lenschow frames it as a "stable compounder with a Quantum option" — and that framing matters for a name that trades like an old-world IT services shop. Almost 50% of revenue and the majority of profits are software now, and the mix is still tilting higher. Hard to argue with.
Lenschow's $350 sits $50-60 above the existing PT cluster (RBC $300, Stifel $290). Two reads here: (1) Barclays is more constructive on the software mix math than the incumbent bulls, or (2) the Street is catching up and Barclays is just first to mark it. PEG of 0.28 is the kind of number that makes PMs look twice — but you need to underwrite the growth rate, and mid-single-digit organic is a step down from the 9.7% LTM print (which was juiced by Red Hat comps and consulting recovery). The "stable compounder" framing implies Barclays is underwriting deceleration, not acceleration. That's the bet.
"This approach makes sense for large customers that will never be 100% in the cloud." — Barclays
That line is the whole bull case in one sentence. It's a positioning trade against the "everyone goes hyperscaler" narrative. The setup works as long as hybrid/on-prem is a real, durable posture for Fortune 500 — and so far every regulated industry survey we see says it is.
$297.80 with two of three PTs below spot is awkward. Stifel and RBC are effectively "in the green" for anyone who bought on their calls — those targets are bogeys, not catalysts. $350 from Barclays is the only fresh upside number on the board, and 17% to a Street-high PT in a slow-grind name like this is decent but not screaming mispricing. PMs holding: nothing to do. PMs not holding: this is a "starter on weakness" setup, not a chase at $298.
$1B Chips Act award + $1B IBM match into a new quantum wafer company is the optionality. Not in numbers today, not in numbers for a long time. But it gives Lenschow cover to frame this as more than a value/compounder story — and that's what gets a multiple to expand in a name that historically trades at 12-15x. The DoW contract mod (now ~$155.5M total) is bread-and-butter, not a catalyst. Ignore it.
Verdict: estimates are too low, PTs are ripping, and the buy-side hasn't caught up yet.
UBS just walked DY PT to $611 from $444 (Buy) — that's a 38% raise, not a tweak. Cantor already at $654 from $436. Two firms, two massive upward revisions in the same week. When street PTs are getting hiked by 35-50% in one shot, it usually means models were anchored to a demand picture that doesn't exist anymore. Read the Q1 print: EPS $4.42 vs $2.72 est (a 62.5% beat), revenue $1.965B vs $1.67B est (17% beat). That's not a weather tailwind. That's a backlog unlocking.
UBS cited "robust and growing market opportunity" and upside to estimates as the year progresses — classic language for "we were modeling the wrong growth curve." The FY27/FY28 EBITDA hikes tell the real story: $1,096M / $1,271M vs prior $992M / $1,126M. Roughly two-thirds of the EBITDA bump is in Building Systems, not the legacy telecom contracting segment. That's the underappreciated angle here — DY isn't just a fiber dig story anymore, the diversification is starting to print.
"These results underscore strong underlying demand, as noted by the company's management, which highlighted that favorable weather conditions were not the sole driver of the performance."
Management explicitly telling you to not dismiss the beat as transitory. We take that at face value.
NTI acquisition — not yet closed, not in estimates. Street models have dry powder on the M&A front. If NTI closes and the Street pencils in accretion, $611 and $654 are likely sandbagged.
FTTH capex cycle is still in early innings for the Tier 2/3 markets. Not sure we can read too much into one quarter, but the magnitude of the beat (17% on the top line) suggests something structural — either market share gains or pull-forward, and DY management is saying it's neither weather nor one-time. Long DY into the print-and-raise cadence. Bogey is Cantor at $654; consensus is going to migrate north over the next 2-4 weeks as the rest of the street updates.
AMD — MI300x secondary market forming at $32K eBay and $1.99/GPU/hr rental is the real story. Demand is hot but buyers prefer rental over purchase at 2x retail, which screams price sensitivity vs NVDA's supply tightness. Confirms AMD is a distant #2 in the high-margin hyperscaler game. China = ~20% of revenue and Huawei's HiONE optical engine is the canary.
INTC — 18A yields improving and Panther Lake ramping at 7x volume on ~50% yields is the first public signal Intel Foundry is for real. Bear case on yields is breaking — the volume ramp disproves the "fail" narrative. External supply locked to Apple/Google/NVDA/AWS strategic slots; not a CoWoS reliever. PC segment still a drag — entry-level CPU prices UP 15% into a collapsing market. Foundry is a 2027-28 story, Q3 is still ugly.
HPE — +30% after-hours on AI server beat and raised guidance. This is the "next Dell" trade — market re-rating server OEMs as AI integrators. Juniper networking beat confirms networking is the next bottleneck. Stock is up on a momentum chase / short squeeze, not just fundamentals. Key Q: is the AI server revenue sustainable or a one-time pull-forward? If margins stay low this move is fragile.
AAPL — Smart glasses locked for end-2027 launch targeting $200-$500 mid-tier eyewear. Multi-year cycle catalyst with Apple Watch-like disruption potential in a $200B+ TAM. MacBook Neo shipments revised 5M → 10M on memory inflation — Apple gaining STRUCTURAL share as low-end PC makers get squeezed. Walled garden strategy still works in AI integration era.
META — Most incremental GW of training compute coming online in 2026 among AI labs. Including neocloud capacity, potentially more than OpenAI + Anthropic COMBINED projected 6GW by end-2026. MSL team starting to produce competitive work. Strategy is capex-heavy brute force, not algorithmic breakthrough. Bull case = enough compute closes the model gap. Bear case = capex becomes value-destructive arms race.
GPRO — Going concern warning explicitly citing AI-driven memory chip shortage. Most direct evidence that AI demand is starving non-AI sectors. Memory vendors allocating to HBM and high-cap NAND for AI, consumer electronics can't source at viable prices. Canary in the coal mine for ALL memory-dependent consumer tech. For GoPro it's existential.
TSM — NVDA now >20% of TSMC revenue, confirmed by Jensen himself. Concentration is a double-edged sword — record revenue on NVDA demand, but any softening flows directly to utilization. CoWoS remains the binding constraint on the silicon track. Toll booth on AI compute and pricing power is real. Geopolitical Taiwan risk currently priced as ~zero.
ANTH — Confidentially files S-1, $47B ARR vs $9B at end-2025 = 5X IN 6 MONTHS. $965B private valuation implies ~20x revenue multiple. First public test of the AI lab business model — S-1 disclosures on gross margins and customer concentration will be CRITICAL. Claude 4.8 Opus beating GPT-5.5 on GBA Eval is a narrow benchmark win. If profitable at scale, validates the whole AI lab category.
Memory inflation destroying PC TAM — DRAM/SSD prices up 200-300% since mid-2025. Component hoarding spreading downstream. "Extinction event" coming for sub-$600 devices. Zero-sum allocation is now visible: AI is consuming supply at the direct expense of consumer electronics. Distribution signal for memory pricing power.
Goldman memory earnings revisions — Massive upward revisions for 2027-28 on SK Hynix and Samsung. Sell-side conviction that HBM cycle extends well beyond current consensus. Distribution-only: margin capture by memory vendors at the expense of downstream buyers. Not a chain scale-mover.
SoftBank €75B France deal — 3.1 GW initial phase by 2031, targeting 5 GW total. €45B over 5 years in Hauts-de-France. Schneider Electric as industrial partner. Largest European AI infrastructure commitment to date. European energy costs remain the binding constraint on execution — projects will migrate to lower-cost power regions within Europe.
SPV credit risk surfacing — Multiple commentators flagging special purpose vehicle structures at the heart of AI infrastructure financing. If securitized/leveraged like pre-2008 structures, this is systemic. Currently at "people are starting to notice" stage, not "data confirms" stage. Flagged for monitoring.
Vista's Robert Smith quote — Insurance claims: $8M manual, $3-4M with general LLM, $200K with proprietary systems carrying domain context. 15-20X inference cost reduction from proprietary context, not frontier benchmarks. Reinforces that data structure and orchestration are the enterprise unlock.
META MSL as frontier challenger — Word is META's Massive Scale Learning team is starting to produce competitive work. If META releases a model competing with GPT-5/Claude 4, it redistributes value within AI labs. Strategy = capex brute force, not algorithmic breakthrough.
Pretraining costs not at $1B yet — Per one informed estimate, labs have not used 300T+ tokens or paid $4/hr for GB300 on long-term rentals. Efficiency feedback loop may be compressing compute-per-frontier-model faster than scale expands. Mild negative for aggregate silicon demand from training side — though inference could offset.
OpenAI Sora team pivoting to robotics — Sora video generation team reportedly became the robotics team. Combined with OpenAI Robotics "making rapid progress" and actively hiring. Resource reallocation from generative media to physical-world AI. Directional bet that embodied AI has more commercial upside.
Glass core substrates 2028 — Being evaluated for future high-end packages, initial production potentially 2028 for high-performance apps. Larger package sizes, finer interconnects vs organic substrates. 12-18 month forward signal for advanced packaging layer. Not yet at commitment stage.
xAI short-term capacity deal — Struck favorable short-term deal with optionality to reclaim within 90 days. Suggests neoclouds are structuring for flexibility as compute becomes more available. Channel check on neocloud bargaining dynamics.
GPU rental rates holding firm — Stable or increasing across the board. xAI deal and token usage growth across tools are the strongest demand confirmation signals in the feed. Directly argues against the digestion narrative.
"AI harness wars" 2027 framing — Emerging narrative that user memory, context, and data portability between AI platforms will be the defining competitive battle of 2027. Open-source tooling around self-hosted memory and skill files gaining traction on GitHub. Early software ecosystem signal, not yet investable.
Positioning stress extreme — GS TMT momentum pair had one of its largest daily pullbacks in recent history. Software ripped 20%+ in May while mega-cap semis stalled. Most Short basket up 25%+ in a straight line. Hedge fund IT exposure at levels where buyers may be exhausted. Software had best month since 2002 while NVDA fell 8 of 10 sessions. This is positioning, not fundamentals — but positioning can move prices for quarters.
DeepSWE benchmark noise — Coding evaluation standard circulating with OpenAI models reportedly dominating. One evaluation framed Claude Opus as a "cute chatbot companion" relative to OpenAI's coding performance. Model competition noise at the benchmark layer.