Good morning.
Futures flat to slightly green — S&P +0.2%, NDX +0.1%. MU +8% GETTING CLOSE TO $1T off Goldman’s aggressive memory upgrades through 2028. ESTC -14% on a miss (revenue guide light, consumption model still lumpy). ADSK +6% on billings beat — Autodesk finally executing on the subscription transition after years of "next quarter." Asia mixed: KOSPI +1.5% on the Samsung/Hynix upgrades, Nikkei flat.
Three themes frame the tape today:
ONE: The memory cycle is structural, not cyclical. Goldman raising Hynix OP by 21% for 2027 and 24% for 2028 isn't a restocking call — it's a structural HBM intensity thesis. If Goldman is right, the GPU/horsepower ratio we're all modeling is too low. MU is the pure play here, but TSMC is the derivative beneficiary since HBM stacks through CoWoS. The risk: positioning is maxed out (GS TMT Most Short pair up 25% YTD, hedge funds out of IT buyers). This could be the fundamental validation that clears the positioning fog — or the excuse to sell into strength if the broader narrative cracks.
TWO: The European AI buildout is becoming a real GW pipeline. SoftBank's €75B ($87B) France commitment — 5GW total, 3.1GW initial phase by 2031 in Dunkirk — is the first large-scale non-US DC announcement we've seen. Schneider Electric as the industrial partner. Europe's power cost disadvantage is real (US-Iran tensions compounding grid prices), but France's nuclear-heavy grid makes it the logical winner. This validates the multi-year CapEx trajectory for DC power/cooling names — Schneider is the direct beneficiary, but the broader read-through is that hyperscaler demand is not US-centric. The 2031 timeline is a reminder: grid constraints are real, but they aren't stopping the buildout. They're extending it.
THREE: The secondary GPU market confirms demand durability, not saturation. AMD MI300x at $32K on eBay (~2x retail without warranty). Rental at $1.99/GPU/hr. MF's channel check: "rental rates stable or increasing." This is the cleanest rebuttal to the "demand is rolling over" narrative. Secondary markets form when demand exceeds supply and the primary channel can't allocate fast enough. The $50K NVDA crash call circulating on Twitter is dangerous narrative — no data supports it. The actual risk is mechanical: crowded positioning meets a catalyst that doesn't arrive as expected. Watch the Goldman TMT pair today — if it stabilizes, the positioning flush is a buyable dip.
We'll hit up NVDA, MU, and TSM first, then get to the memory and semicap names, then close with SoftBank/Schneider for the infrastructure thread.
VERDICT: THE ATLAS RE-ACCELERATION IS REAL, BUT THE STOCK PRICES IT. Q1 FY27 was a clean beat-and-raise — revenue $688M (+25% YoY, $23M above consensus), Atlas growth of 29% (above the 26% guide and slightly ahead of Q4’s 29.2%). Management raised FY27 revenue guide by $60M to $2.92-2.96B and guided Q2 Atlas to ~26% (Street had 22%). The 72% rally over the past year says the market already sniffed this. Now it’s about proving the AI tailwind isn’t just a one-quarter sugar hit.
PT range $315 (Macquarie, Neutral) to $435 (Stifel, Buy). Median around $390-400. Consensus is bullish with a cautious tail: five Buy/Overweight (Stifel, BofA, Cantor, Mizuho, Morgan Stanley), two Neutrals (UBS, Macquarie), one confusingly ambiguous Needham (raised to $400 but maintaining an Underperform? Likely an error — reader should treat as Buy given Conviction List mention). The bulls are leaning on Atlas acceleration + AI consumption, while bears point to the stock’s pre-run and AI monetization still being “early.”
Key data points from the quarter:
New: The exact magnitude of the Atlas upside (3% beat on their own guide) and the fact that core workloads drove the beat, not just AI (UBS hammered this). CJ Desai’s customer meetings boosting confidence on product roadmap is incrementally positive. The margin expansion was above forecasts — Rule of 40 in play.
Already known: MongoDB is a share-gainer in next-gen databases. The long-term AI thesis (Frontier Labs, agentic workloads) is well understood. The stock’s 72% run reflected expectations of a strong Q1.
BULL CASE: Atlas growth re-accelerated (29% vs prior Q 29.2% — slight uptick) and management raised guidance by more than the beat. The AI consumption story is real: Frontier Model Labs and AI-native apps are driving usage. NRR at 121% implies existing customers are spending more. Margins expanding 200bps YoY. Rule of 40 achieved. MDB is moving from a “technical decision” to a “strategic platform choice.” Stifel sees at least 3% upside to the Q2 guide based on consumption trends. Needham explicitly called out “early stages of monetizing AI” but saw enough to put it on Conviction List Buy.
“MongoDB is raising its position from a technical decision to a strategic platform choice for customers.” – Needham
BEAR CASE: The 72% rally already prices in this trajectory. UBS was blunt: the upside was driven by core business, not AI — and in contrast to Snowflake, AI pull-through remains early-stage. Q2 Atlas guide of 26% implies deceleration from Q1’s 29%, against tougher comps. Macquarie keeps Neutral, noting “efforts to sell AI to enterprises remain largely nascent.” Valuation is stretched — Macquarie’s Fair Value suggests overvalued, and UBS sees limited follow-through after the pre-earnings move. Needham’s confusing rating (Underperform?) signals some on the street think “buy the rumor, sell the news.”
“MongoDB was clear that the upside was driven primarily by the core business, as any AI pull-through remains early-stage (in contrast to Snowflake).” – UBS
This is a database layer being pulled into the AI stack as workloads move from experimentation to production. Cantor Fitzgerald frames it as a “later-cycle winner” and a beneficiary of the token-based economy. Morgan Stanley noted MDB’s cloud growth aligns with Snowflake and Datadog — consistent margin expansion across the infrastructure cohort. The Q1 beat across SNOW (strong guide), DDOG (consumption up), and now MDB reinforces the narrative that AI workloads are beginning to hit the data infrastructure layer, albeit unevenly.
The divergence: Snowflake saw clear AI-driven acceleration; MDB’s beat was core business + early AI. That’s a nuance PMs should watch — if MDB’s AI contribution is smaller, the multiple (currently ~37x base case FCF) is harder to defend unless the core re-acceleration sustains. For now, the setup is good, but the easy money is made.
Verdict: S printed a headline that looks like a mess (revenue miss, weak Q2 guide, 8% RIF) but the underlying ARR and margin story is stronger than the tape suggests. Stock down ~18% after-hours, which feels excessive for a name that just accelerated net new ARR +55% YoY and raised operating margin guidance. The bull case rests on "demand signals are better than the reaction" — the bear case is "execution remains inconsistent and the AI tailwinds are taking longer than expected." I’m siding with the bulls here, but the range of PTs ($15–$24) tells you there’s no consensus on the r/r.
Revenue $276.7M (+21% YoY) — MISSED the midpoint of guidance and consensus by ~$500K. That tiny miss combined with a Q2 guide 110-130bps below Street consensus is what triggered the selloff. But look past the top line:
PT range: $15 (DA Davidson, Neutral) to $24 (Cantor, Jefferies). The cluster of Buy ratings sits at $20–$24. UBS ($16, Neutral) and Raymond James ($18, Market Perform) are the skeptics. Street consensus is roughly $20 — call it 3.9-5x EV/CY27 revenue.
BULL CASE (Cantor, Rosenblatt, BofA, Jefferies): The quarter was better than the headline. Net new ARR acceleration is a leading indicator — deals were pulled forward from Q2 into Q1, but management reiterated full-year net new ARR improvement. The RIF is proactive, not reactive: they’re cutting excess capacity and reinvesting into AI Security, Purple AI, Data, Cloud. Operating margin expansion is on a clear trajectory to 10%+ by year-end. The demand signal for AI endpoint security is real — non-endpoint products now 50% of ARR, Prompt Security nearly doubled QoQ. The pullback is an overreaction. "Demand signals are better than the headline reaction suggests" — Cantor.
"We see the pullback as an overreaction. Revenue is increasingly back-end loaded as larger deals ramp." — Rosenblatt
BEAR CASE (Raymond James, DA Davidson, UBS): Revenue miss + weak Q2 guide = execution still inconsistent. The RIF is twice the size of the last restructuring in 2023 and includes go-to-market personnel — that’s a red flag, not a proactive move. Calculated billings growth of 14% missed the Street-low estimate of 16%. The stock ran 50% from April 10 to earnings — that run-up was pricing in perfection, and the guide didn’t deliver. AI tailwinds are taking longer to materialize. The new CFO (60 days in) is still learning the ropes. "Evolving macro and geopolitical effects led to an unfavorable invoice cycle" — that’s management-speak for "we can’t control the timing of deals."
"The guidance suggested that AI tailwinds are taking longer to materialize than anticipated." — UBS
"The company reported an in-line but solid quarter. The pullback offers an attractive entry point." — BofA (upgraded to Buy)
"First-quarter net new ARR growth exceeded 50% after the company pulled forward several large deals. We maintain Buy on the emerging need to secure agentic AI." — TD Cowen
"Management does not expect any impact on go-to-market operations and does not appear to be embedding the changes into guidance." — Jefferies (on the RIF)
"Working capital metrics suggested a back-end weighted quarter. The restructuring involves changes to team structure and go-to-market strategy." — Raymond James (downgrade)
S is a microcosm of the AI security vs. growth deceleration debate in cybersecurity. Endpoint EDR is maturing, but the agentic AI wave is creating a new demand vector for identity, data, and cloud security. S’s non-endpoint acceleration (now 50% of ARR) validates the platform thesis. Peers like CRWD and CHKP are watching — if S can sustain 20%+ growth while expanding margins, it validates the entire mid-cap security space. The RIF across multiple vendors (S, ZS, OKTA) suggests the industry is shifting from land-grab to operational discipline. Short-term pain for long-term gain — if you believe the demand signals.
Bottom line: The stock is down on a $500K revenue miss and a prudent guide. The ARR acceleration and margin trajectory are real. Not sure we can read too much into calculated billings volatility. I’d be a buyer into weakness, but the $15–$20 range feels like the battleground until Q2 prints.
Verdict: DELL IS THE AI INFRASTRUCTURE TRADE DONE RIGHT. This quarter rewrote the narrative. Not just an AI server story — traditional compute, storage, and PC all fired at once. The market's re-rating from high-teens to 25x is rational, but the real debate is whether $60B in AI revenue is peak or just the first inning.
STREET VIEW: $440-$700 CLUSTER, CONSENSUS BUY
Price targets exploded post-print: UBS (Neutral) at $440, Evercore (Outperform) at $450, Piper (Overweight) at $497, BofA and JPM (Buy/Overweight) at $500, Barclays (Overweight) at $550, Melius (Buy) at $565, Susquehanna (upgraded to Positive) at $700. That's a 60% range from low to high. The median is ~$500 — roughly 28x the midpoint of the new FY27 EPS guide ($17.90).
The collective thesis is shockingly consistent: this isn't a one-quarter beat, it's a structural re-rating of Dell as a dominant AI compute platform.
"Beat every line in our model, including server revenue in both traditional and AI segments, with accelerating growth in high-margin storage." - Melius
The bull case is a compounder thesis. AI servers now 36% of revenue (~$60B FY27, +145% YoY) but the real torque is in attach — storage (up 8.5% YoY, accelerating), traditional servers (up 92% YoY on the 14th-to-16th gen refresh cycle), and PC margins (9.7% OPM vs 7.7% consensus). Dell is supply-constrained, not demand-constrained. They have $51B backlog and $24B in new AI orders in a single quarter. Susquehanna argues the AI server mix shift to enterprise inferencing brings structurally higher margins as training share converts. Scale + services (40-45% rev, 42% GM) absorb the AI gross margin dilution — corporate OPM expanding 50bps to 9.3% despite mix. The PEG is 0.97x — cheap if this earnings growth sustains into FY28. Bulls say the Rubin ramp in early 2027 pushes backlog into FY28, extending double-digit revenue growth.
"Second-half EPS of $8.24 vs first-half $9.66... risks from demand pull-in and elevated valuation." - UBS (Neutral, sole holdout)
The bear case isn't about demand quality — it's about timing and multiple. Dell acknowledged pull-forwards and price increases. FY27 guidance implies ~10% H2 revenue decline vs H1. That's a demand cliff if supply chains normalize faster than expected. The Street is using the supply-constrained guide as a baseline — upside requires better GPU allocation, not incremental demand. UBS flags CSG margin reversion in H2 (pricing pull-forward normalizing). At $317, DELL trades at 37x trailing EPS — the PEG is low only because FY27 growth is so high. Bears argue that $60B in AI revenue may be the peak before hyperscalers optimize their own silicon or post-Rubin digestion. The $700 PT from Susquehanna implies 39x FY28 EPS — aggressive for a hardware cyclical.
This is the cleanest positive read-through to HPE (enterprise server refresh + AI), SMCI (AI server, but Dell's margin execution tops), and ANET (networking for AI clusters). The bearish read-through: if the supply constraint is real for Dell, it means GPU supply (NVDA) is gradually improving — which is a positive for the whole ecosystem. No read-through to PC pure-plays (HPQ) because Dell's PC margin beat is idiosyncratic (pricing power + share gains). The "skip 15th gen" narrative is a powerful catalyst for enterprise refresh cycles across the sector.
Verdict: SNOW just delivered the quarter that changes the narrative. This isn't another "AI is coming" story — the data is here. 34% product growth, a 400bps raise to full-year guidance, and Cortex CoCo scaling to 7,100 accounts in three months since GA. The stock ripping 39% post-earnings is the market pricing in what analysts are now calling a "step change." The question isn't whether the AI thesis is real anymore — it's whether the multiple can support the next leg.
This is the quarter where "data infrastructure for AI" became real for SNOW. CoCo (Cortex Code) isn't a pilot program. It's the largest driver of a full-year guidance raise. Management called it out specifically. HSBC's upgrade from $176 to $289 captures the street's belated realization: AI is accelerating both direct AI revenue AND core data platform consumption. Customers are moving workloads to Snowflake faster to support governed AI use cases. The flywheel is spinning.
PRODUCT REVENUE PRINTED $1.334B, UP 33.9% Y/Y — 5.3% ABOVE CONSENSUS. The largest beat in six quarters and highest growth in eight. Operating income beat by 35.2%. Record sequential dollar growth. Net retention rate inflected up 100bps QoQ to 126%. Customers spending >$1M grew 28.8% Y/Y to 779.
Management raised FY product revenue guidance by 400bps to ~31% growth. That's the biggest upward revision in recent memory. Q2 guidance also came in above expectations.
CoCo is the headline, but the core business is accelerating too. RPO grew 37.7% Y/Y to $9.21B. Accounts using Snowflake AI hit 13,600, up 45.5% sequentially. Benchmark noted Snowflake Intelligence and Cortex Code are scaling faster than any products in company history.
Consensus is now $275-$300, with one outlier at $320. The old cluster was $200-$255. The shift is uniform:
| Firm | PT | Rating | Key Callout |
|---|---|---|---|
| Monness, Crespi, Hardt | $320 | Buy | Highest on street |
| TD Cowen | $300 | Buy / Top Pick | "Step change in AI narrative" |
| Stifel | $300 | Buy | Stronger than expected AI consumption |
| Freedom Broker | $300 | Buy | Operating leverage from AI tool rollout |
| Piper Sandler | $295 | Overweight | Product growth acceleration + full-year raise |
| HSBC | $289 | Upgrade to Buy (from Hold) | CoCo is "clearest evidence of AI monetization" |
| KeyBanc | $285 | Overweight | Cortex Code wasn't in initial guidance |
| Scotiabank | $285 | Sector Outperform | Product revenue exceeded investor expectations |
| Cantor Fitzgerald | $282 | Overweight | 13x EV/CY27 sales (in-line with medians) |
| Truist | $275 | Buy | Core + AI workload acceleration |
| Benchmark | $270 | Buy | Record sequential dollar growth |
13+ firms raised targets in a 24-hour window. The uniformity of the move matters. Nobody's arguing the numbers.
BULL CASE (they won the quarter): This is the inflection the Street has been waiting for. AI is no longer a narrative — it's driving consumption of core data, analytics, and application workloads. CoCo at 7,100 accounts in <4 months tells you the product-market fit is real. The $6B AWS partnership and Natoma acquisition (agentic AI) extend the runway. Management's guide raise isn't conservative — but they've earned credibility with this beat. Trading at ~13x forward sales is in-line with historical medians, not overly aggressive if growth sustains 30%+. Monness called it: "well positioned to benefit from long-term secular tech trends."
"CoCo marks the clearest evidence of Snowflake's AI monetization opportunity. Since GA in February, it has scaled to more than 7,100 accounts and contributed meaningful AI revenue." — HSBC
"These developments mark a step change in the AI narrative for the company." — TD Cowen
BEAR CASE (don't get carried away): The stock is up 39% in a week. That's pricing in perfection. RPO and billings came in weaker than expected — customers are still cautious on long-term commitments. Competition from Databricks, Google BigQuery, and hyperscaler-native solutions isn't going away. The AI tailwind is real, but it's also pulling forward demand. Can growth sustain 30%+ when comps get harder? Monness' own note acknowledged "competition remains dynamic, software is in transition, and the macroeconomic environment is treacherous." The $320 PT at ~15x sales is already pricing multiple expansion that may not materialize if growth decelerates.
Verdict: Beat-and-raise quarter with an AI-agent option. Analysts uniformly raising PTs to a $110-$130 range (from prior $100-$110 cluster). The re-rating story is live – identity security as a platform trade, not just a utility.
cRPO grew 12% y/y – 200bps ABOVE guidance and consensus (everyone had 10%). NRR ACCELERATED TO 107% (wasn't a talking point last quarter). Revenue $765M vs $752M est. EPS $0.91 vs $0.85. Gross margin held at 77%. The guide: Q2 cRPO 11% y/y (above consensus 10%), raised full-year revenue, OI, FCF, and EPS. This is a textbook beat-and-raise.
New products – Okta Identity Governance – drove 25% of bookings. AI Agents (both Okta for AI Agents and Auth0 for AI Agents) are early but showing strong pipeline and larger average deal sizes.
Bull case: This is a growth re-acceleration story. cRPO is inflecting – guide implies acceleration, Jefferies notes a typical 2pp beat would put cRPO at ~13% y/y. New sales capacity is ramping, OIG traction is real, and AI agents add option value that's NOT in the numbers yet (Cantor: "sooner than previously expected"). At 4.8x EV/CY27 revenue (Jefferies), the stock isn't priced for acceleration. DA Davidson: "re-rating higher."
Bear case: Not much substance in this coverage – the only real pushback is that AI agents are early-stage and still immaterial. Mizuho calls it "early stage." Jefferies says guidance "assumes minimal contribution from agentic tech." So the bull case depends on something that hasn't shipped in volume yet. The stock is already trading above its prior 52-week high (now $115.94, up 8% from $107.84) – forward returns might be more muted until we see real AI agent revenue.
"Our price target increase reflects continued momentum in emerging product areas such as Identity Governance, along with growing traction in Okta for AI Agents and Auth0 for AI Agents, repositioning Okta from a core access vendor to a broader identity security platform." > – Cantor Fitzgerald
"Guidance assumes minimal contribution from agentic technology. A typical 2 percentage point beat would imply acceleration to approximately 13% year-over-year growth." > – Jefferies
"We believe these factors (AI Agents, sales productivity, new capacity, OIG traction) will drive a re-rating higher in shares." > – DA Davidson
Okta is the purest play on identity security as a platform – and identity is the new perimeter. AI agents create a new identity vector: how do you authenticate a non-human entity? Okta is betting they own that. The read-through for the space: if Okta’s AI agent pipeline materializes, it lifts the entire identity security category. Peers like CYBR (identity-centric) and S (SentinelOne – endpoint + AI agent security?) see a tailwind. The bigger narrative: security vendors with platform expansion optionality are getting re-rated (see: CRWD, PANW). Okta is in that club now.
Verdict: Mixed signals but the bull case has ammo. Stock up 48% in the past year, trading at $91.84 – just 5% off a 52-week high. The Hanwha deal is the headline, but supply chain risk is the overhang. Not a binary – just a risk/reward that’s getting tighter.
Price targets range from $96 (BofA, Neutral) to $120 (Rosenblatt, Buy), with Stifel at $106 (Buy) and Northland at $101 (Outperform). Consensus skews bullish – 4 out of 5 analysts say Buy or Outperform – but the lone downgrade from Summit Insights (Hold) is the one that matters if you’re long. The cluster: $96–$120, with Rosenblatt the most aggressive. Consensus EPS for this year is $0.76 (from -$1.62 LTM).
“Ambarella integrates all accelerated computing functions including fusion, perception, AI acceleration, CPUs, encoding, and other system functions into a single chip, contrasting with data center approaches that typically use a collection of discrete SoCs from different vendors.” – Stifel
Incremental this week:
AMBA is a pure-play on edge inference – the idea that AI workload moves from the cloud to the camera, robot, or car. That puts it in the same conversation as NVDA (Jetson), INTC (Mobileye), MRVL (custom ASICs for auto/edge). But AMBA’s niche is lower power, higher integration. The Hanwha deal is a proof point that enterprise security is a real edge AI use case – not just hype. Peers to watch: ALL (Allegion, security), HIK (Hikvision, but geopolitical), ROK (robotics).
The bear case is supply chain – and that’s thematic too. If Samsung foundry tightness hits second-half shipments, it’s not just AMBA – it’s every fabless semi with TSMC/Samsung dependency. Watch for commentary from NVDA, AMD, MRVL on 2H supply constraints.
Bottom line: AMBA is a position for the thesis, not the quarter. The Hanwha deal is a multi-year catalyst, but the 2H26 risk is real. If you’re long, you’re betting the channel build is pre-emptive, not a canary.
ESTC is a “show me” story after cloud revenue declined QoQ for the first time in a Q4. The range of PTs tells you everything: $55 (TD Cowen) to $85 (UBS). Five analysts updated – two Buys, three Holds/Neutrals. Consensus? Notional PT ~$65, but the distribution is bimodal. The stock sits at ~$58, so it’s pricing in the neutral case.
What’s new vs. what’s old:
UBS holds the high bar at $85 (3.2x EV/CY27 revenue) and Stifel at $65 (3x). They’re leaning on acceleration in H2 FY27 and strong RPO momentum – multi-year deals suggest real commitment from enterprise customers. Management pointed to sales capacity adds in H2 and traction with CISA partnership + AI product uptake.
“For the stock to re-rate higher in coming quarters, SaaS results must improve beyond the current focus on Sales-Led Subscription Growth.” – Stifel
That’s the bull case in one sentence. They’re betting that cloud re-acceleration is real, just delayed.
TD Cowen ($55), Cantor ($59), and DA Davidson ($60) sit on the sidelines. The core issue: cloud deceleration for three straight quarters. Cantor flags that SLS is guided to keep decelerating in Q1 FY27. Without cloud momentum, the stock stays range-bound.
“Given the deceleration in the cloud business over the past three quarters, Elastic has become a ‘show me’ story.” – UBS (the bull firm, echoing the bear critique)
DA Davidson notes the revenue beat was “lower than typical” for ESTC. The bar was already low – and they still didn’t clear it cleanly.
ESTC is a second-half bet with no near-term catalyst. The RPO numbers are impressive, but the market wants to see cloud revenue turn positive QoQ before re-rating. The bull-bear divide is simple: do you trust the H2 guidance or need proof? The stock at $58 already discounts a lot of caution. If cloud inflects in Q1 or Q2 FY27, the Buy targets ($65-$85) become reachable. If it doesn’t, the stock grinds lower into the $50s.
Read-through: ESTC’s cloud miss is a microcosm of the broader enterprise software digestion phase – customers are still optimizing existing cloud spend before expanding. Peers like DDOG, MDB, and SNOW will face similar scrutiny on cloud growth rates next quarter.
Verdict: CRM beat the quarter but didn't change the narrative. That's the problem.
Headline beat on revenue (constant currency +11.6% vs. 10.8% consensus) and margin (34.8% vs. 33.4% est). Agentforce crossed $1B in ARR. Consumption metrics accelerated. Guidance? Deceleration in the legacy app franchise. CRPO in-line for a second straight quarter. The bull case needs second-half acceleration that management is promising but hasn't proven yet.
Stock down ~33% YTD. This thing is priced for disappointment. But that doesn't mean it's a buy — it means the bar is low and the burden of proof is on CRM.
PT range is WIDE: $175 (DA Davidson, Neutral) to $290 (KeyBanc, Overweight). Consensus cluster is $215-$250 among the Outperform/Buy camp:
Bull case: Valuation is already washed out. At sub-1x PEG, the market is pricing in zero acceleration, maybe worse. Agentforce $1B ARR is real, consumption metrics are inflecting, and management's Rule-of-50 framework for FY30 is a tangible target. If second-half acceleration materializes — driven by Agentforce, Data Cloud, and Slack pipelines — the stock re-rates meaningfully. Heads you win (multiple expansion), tails you don't lose much (already derated).
"With shares already deeply derated, we believe the risk/reward is asymmetric to the upside." — Freedom Broker
Bear case: Nobody believes the second-half story. CRPO in-line is a pattern now, not an anomaly. Guidance signals continued deceleration in the core app business. The ASR program is juicing EPS while masking underlying revenue quality. Agentforce monetization is still "under construction" — not a proven consumption model. You're paying 23x for a story that needs a miracle in 2H.
"Results were fine but guidance left a little to be desired. Investors are left having to trust management that second-half acceleration is around the corner." — Cantor Fitzgerald
New: Agentforce crossed $1B ARR. That's real. Consumption metrics (active workspace users, tokens, MCP usage) accelerated QoQ. Management introduced "Headless 360" as a new monetization vector.
Not new: The legacy business is decelerating. CRPO in-line (not ahead). The second-half acceleration thesis existed last quarter too. The ASR impact on FCF growth was already telegraphed.
Incremental: TD Cowen cut PT specifically citing the ASR program reducing FCF growth estimates by 5ppts. That's mechanical, not fundamental — but it matters for the multiple calculation.
"Salesforce opened FY27 with a clean headline beat that defends but does not advance the bull case." — Freedom Broker
That's the whole debate in one sentence. The quarter was good enough to keep existing holders from panic-selling. Not good enough to bring new buyers in.
CRM is a bellwether for the enterprise software thesis: AI monetization is real but lumpy, and the market is impatient.
If CRM can't sustain a rally on $1B Agentforce ARR and an 11.6% revenue beat, what does that say about the rest of enterprise software? It says the market is focused on acceleration, not just growth. Deceleration at a 12% grower gets punished. The read-through for ORCL, NOW, SAP: show me the acceleration or get derated.
The broader issue: consumption-based AI monetization is still under construction across the sector. CRM's experience — big headline numbers, unclear path to sustained reacceleration — is likely to repeat for peers. Watch the next round of enterprise prints for the same pattern: beat on AI metrics, miss on the core guide, stock goes nowhere.
The bull case on Marvell is stronger than the stock's post-print fade suggests. The market did the mature thing — sold the in-line quarter and modest Q2 guide, then sat there staring at a $16.5B fiscal 2028 revenue number that management basically drew a map to. Multiple firms piled on with PTs all the way up to $275 (Benchmark from $130 — that's not a tweak, that's a conviction upgrade). The pullback is a valuation reset, not a fundamental crack.
Benchmark went $130 to $275, maintaining Buy. TD Cowen to $200 (Hold), Cantor to $220 (Neutral), KeyBanc to $260, Deutsche/BofA/Rosenblatt all at $240. That's a PT cluster from $200-$275, with the bull case anchored in the fiscal 2028 framework — management explicitly walking through ramp timing on interconnect, switching, custom silicon, AECs, retimers, DCI, and scale-up optics. Not vague. Detailed.
Cantor flags an interesting tension: some investors are baking in >$12 EPS for CY2028. Cantor's at $10, consensus at $7.65. That gap matters.
Bull: The $16.5B fiscal 2028 revenue framework includes a fiscal 2029 custom silicon target that management views as mostly ahead of the model, not baked in. The interconnect franchise (optical, AECs, retimers) is the dollar growth driver this year, but the switching and scale-up optical programs are the multi-year compounders. PEG of 0.16 says the 65x trailing multiple is growth-justified.
Bear: Stock's up 208% in a year. Trading 29x CY2027 EPS while NVDA and AVGO sit at 14x and 17x respectively. The post-print fade tells you the market priced a lot of this before the print. And the "mostly in-line quarter" language from Benchmark is telling — the guide wasn't a blowout, it was a solid double. Not enough to re-rate from here without execution proof on the custom silicon timeline.
Benchmark's Cody Acree framed the central debate better than anyone:
"The central institutional debate is how much of Marvell's $16.5 billion fiscal 2028 revenue framework and fiscal 2029 custom-silicon target is already embedded in the share price, and how much value investors should assign to scale-up switching and newer optical programs that management still views as mostly in front of the model."
That's the whole setup in one sentence. The bulls think the framework is a floor. The bears think it's priced. The tape says the bears won the day — but $200 is still a 26% MoM gain. This is a digestion move, not a thesis break.
Cheap for a reason. Two neutrals and one bull on the print, but the bull (Citizens at $15) is the outlier. The rest of the street is anchored at $8 – exactly where UBS and D.A. Davidson landed. That's a wide spread ($8 to $15), which tells you the r/r is asymmetric but the path to the high end isn't clear.
### THE QUARTER AT A GLANCE
Solid beat on revenue ($205M vs $204M consensus, +9.5% y/y) and EPS ($0.10 vs $0.08). Non-GAAP operating margin popped to 11.5% from 8.8% – a nice step in the right direction. Gross margins held near 88.5% (unchanged, scalable software). AI products and seat expansions drove in-period dollar-based net retention to 97% (up from prior prints). Tech vertical returned to growth for the first time in 8 quarters – that's a real signal.
But the headline number hides the warts. Billings came in at $194M vs $204M consensus – a $10M miss. RPOs at $518M vs $526M. The company did raise FY27 organic revenue growth guidance to 8.0% constant currency from 7.8% – effectively just flowing through the Q1 beat. That leaves Q2-Q4 needing to deliver ~8.4% CC growth, only 40bps below Q1's level. Not a massive hurdle, but as UBS flags, “guidance leaves limited room for a potential deceleration” – exactly the kind of line that keeps me on the sidelines.
### CONSENSUS MESSAGE
Two neutral firms (UBS, D.A. Davidson) both at $8 PT, one buy (Citizens) at $15. The bulls are leaning on AI monetization (StackAI acquisition expands non-seat revenue potential) and the tech vertical re-acceleration. The bears cite macro budget pressures, the billings miss, and an already-low guidance band that doesn't provide a cushion.
Citizens (Market Outperform, $15 PT): "Asana's Q1 results highlight strong growth, particularly in its AI segment, which contributed to its earnings success."
UBS (Neutral, $8 PT): "Ongoing hiring uncertainty and broader pressures to applications IT budgets… the guidance leaves limited room for a potential deceleration."
### BULL vs BEAR
Bull case (Citizens): AI + tech vertical cycling a multi-year low → accelerating seats and expansions → margin leverage flows through → stock re-rates 2x+ from here. StackAI gives a new revenue vector beyond per-seat.
Bear case (UBS, D.A. Davidson): The billings miss is a leading indicator. Guidance is mechanical, not aspirational. Any macro wobble (hiring freeze, budget cuts) and the 8.4% needed in Q2-Q4 becomes a miss. The stock has been a value trap for a year – 63% decline. The $8 PT is effectively the floor; the $15 PT is a hope trade.
### WRAP
ASAN is a show-me story in a risk-off tape. The Q1 print was fine – not great, not horrid. The billings line is the real concern. Tech vertical coming back is a real catalyst, but it's early. I'd rather wait for a second consecutive beat with a stronger billings print before dipping in. The stock at $7.35 is pricing in zero growth, which is probably too harsh – but not enough to swing a PM who's already short cyclicals. Pass for now.
Verdict: Two catalysts, one thesis. META is layering two revenue vectors—checkout revamp via Stripe and enterprise AI monetization—on top of a core ad business that keeps chugging. The market hasn’t priced either option. Keep leaning in.
Citizens stays Market Outperform / $825 after META’s March partnership with Stripe’s Agentic Commerce Suite. The new Universal Checkout Platform lets buyers complete purchases in a few taps using saved credentials. Early days, but the key difference from the retired Shops experience: SUBSCRIPTION & LOYALTY CAPABILITIES built in. That matters.
"The system allows buyers to use saved credentials to complete purchases in a few taps." — Citizens
Bear case: merchants will balk because basket-building is clunky and they want direct consumer relationships. META wants the data and attribution. This is a long battle for merchant adoption.
Bull case: META is methodically expanding product tagging across content. If even a fraction of the 3B+ daily users click through, the revenue attachment rate dwarfs what Shops ever did. Plus, the agentic commerce angle (AI doing the checkout for you) is a 2027 story that starts now.
BofA reiterates Buy / $835 following META’s new Enterprise Solutions unit. The goal is repeatable deployment models for corporate AI adoption. CEO hinted at cloud computing if infra overbuild creates excess capacity. HUGE.
META’s P/E: 23x on 81.9% gross margins. That’s not a growth-at-any-price multiple. Enterprise AI sales are less macro-sensitive than ads—they provide a floor if ad revenue wobbles. BofA sees optionality: if META overbuilds capacity (a real risk), enterprise sales can soak up the slack, containing margin compression.
Risk: capacity overbuild could flood the market and depress AI pricing. But with the enterprise AI + cloud market >$1T by 2028, META doesn’t need a massive share to move the needle.
Bull — Two new monetization levers (checkout + enterprise AI) layered on a 26% revenue growth base. Optionality is real. 23x P/E leaves room for multiple expansion if either vector gains traction.
Bear — Merchant adoption lags, enterprise AI is a commodity race, and subscription fatigue (Meta One) could slow user growth. Also, the Vermont addiction lawsuit just got Supreme Court clearance—legal overhang remains.
Bottom line: META is a 3-vector story (ads, commerce, enterprise) trading like a single-product company. That’s the opportunity.
Verdict: MOD is pricing in perfection — and the market is right to do it. 103% YTD, 196% one-year, P/E 123x. The multiple says you're late. The thesis says you're early.
Two analysts raised PTs today — UBS to $355, DA Davidson to $330 — both citing the same catalyst: a $4B LONG-TERM AGREEMENT WITH A DATA CENTER CUSTOMER that extends through CY2029. That's not a deal, that's a franchise reset. DA Davidson also bumped FY27/28 EBITDA estimates and introduced initial FY29 numbers, citing "additional visibility" from the contract.
The bull case: This isn't a one-off chiller win. It's a structural re-rating of Modine from cyclical industrial to infrastructure-growth compounder. UBS analyst Neal Burk says consensus will have to revise higher over time. His math: Climate segment at 16x FY28 EBITDA while the whole company trades at 32.3x LTM — implying massive compression if growth is real. The $4B deal backstops that growth.
The bear case: You're paying 123x earnings for a company that still has component shortages in the current quarter. DA Davidson flags chiller volumes constrained in FQ1 as Modine qualifies new vendors. At 32x EV/EBITDA, one miss on the ramp and this stock gets cut in half. The setup is binary — execution or re-valuation.
Blockquote from the strongest source, UBS analyst Neal Burk:
"The combination of expected forecast revisions and what UBS views as a relatively low valuation signals potential upside to the current share price."
Low valuation? At 32x EBITDA? That tells you everything about where the data center frenzy has landed. Modine is no longer a manufacturing play — it's a capacity contract with a chiller attached. PMs who own it sleep well on the backlog. PMs who don't own it are hoping one of those newly qualified vendors trips.
NEUTRAL VERDICT WITH A SKEPTICAL TWIST. Two PT bumps to $12-13 (from $11-12) after a clean Q1 beat — but both firms stay on the sidelines. The bull case gets a little oxygen from the revenue surprise (+5.3% vs ests). The bear case? Still AI disruption overhang and a valuation that already bakes in the good news.
Q1 FY27: Revenue $418M (beat by $21M), EPS $0.15 (miss by a penny). ARR growth accelerated? Sure. But DA Davidson calls it "adequately priced." BofA uses a slightly higher multiple (2.8x vs 2.6x EV/CY27 rev) to reflect "more constructive execution view" — then slaps an Underperform on it anyway. That's the vibe.
"We believe growth is adequately priced in today as traction with new AI products remains early in providing more significant upside to revenue." — DA Davidson
That's the whole debate in one sentence. The beat buys you a PT bump, not a conviction upgrade.
PATH is a show-me story. The Q1 print was a step in the right direction, but neither house is ready to step in front of the AI-threat narrative. At $11.58, you're paying ~2.8x FY27 revenue for 10% top-line growth. Not cheap enough to get excited, not expensive enough to short into a beat. Neutral-to-cautious until we see AI product attach rates move the needle.
Verdict: Two analysts, two very different takes. Barclays goes full bull ($199 PT, Overweight) while UBS stays parked at Neutral with a $160 target. That spread tells you everything about the debate — valuation vs. trajectory.
### THE QUARTER AT A GLANCE
NTAP printed a clean beat: revenue $1.95B (2% above UBS’s $1.91B est), EPS $2.43 vs guides $2.21-2.31 and UBS’s $2.26. Gross margin hit 70.5% (70.0% est). All-flash revenue $1.22B — UP 18% YoY. Public cloud (ex-Spot) grew high-teens with margin expansion.
Management guiding product revenue up high-single digits in FY27. Organic growth — they specifically said minimal pull-in or pricing impact.
Barclays (Overweight, PT to $199 from $120): "Product margins exceeded expectations following a miss in the previous quarter." That's the key call-out. The Q beat isn't just a one-off — it's a reversal of the prior quarter's margin disappointment. They're buying the re-rate.
"NetApp delivered results that beat expectations on both top and bottom lines. The company’s margin performance was a key driver of the price target increase."
UBS (Neutral, PT to $160 from $113): They see the beat but flag the margin math ahead. Product gross margin expected to dip below 53% in Q1, total gross margin mid-69% in Q1, full-year ~69.2% — DOWN ~200bps YoY. The mix shift to higher-margin support and public cloud helps, but the product margin compression is real.
### BULL VS BEAR
BULL (Barclays crew): Strong organic demand, all-flash momentum intact, margins inflecting after the Q3 miss. At $199 (~28x forward earnings) you're paying for durable mid-teens growth with an expanding software/services mix. The beat proves execution is back.
BEAR (UBS camp): You're buying at 32x trailing earnings, 21x book, and product margins are rolling over. The ~200bps gross margin headwind in FY27 eats into EPS leverage. At $160 (~22x) it's a fair valuation for a storage vendor with slowing growth in the core. Not sure we can read too much into one quarter's beat given the guide implies a step-down.
Verdict: The beat-and-raise was real, but MaintainX is now the only story. Stock at $225 – 52-week lows – because the market is pricing in dilution risk and a 20x+ forward sales multiple on a deal that doesn't close until FY27. Core fundamentals are fine. The narrative pivot from "Design & Make" to "Design, Make & Operate" is a big bet. PMs need to decide if they trust the playbook repeat or see indigestion.
UBS notes the MaintainX move "addresses a known gap based on prior customer feedback" – the operations layer was missing. ADSK sees this as "potentially larger than Construction over time." If the Construction playbook repeats (and it has – double-digit growth for years), the bull case starts to get interesting.
Bull: Core business is accelerating (billings +13.5%, guide raised). MaintainX opens a greenfield TAM in facility management and MRO. ADSK already has the distribution, the brand, and the AI narrative. Margin dilution is temporary – the framework is in place for FY27-FY29. At $225, you're getting the core for cheap with a free call option on operations.
Bear: 20x forward sales for a growth asset that's 1/10th ADSK's size is aggressive. Integration risk is real – this is their biggest ever. The stock is already pricing in execution failure. And at 46x earnings, zero margin for error. If the TAM doesn't materialize as quickly as hoped, the multiple compresses further.
UBS nails the current tension:
"The quarter was encouraging and likely exceeded expectations, particularly given concerns about a potential guidance reduction."
That's the earnings story. The stock doesn't reflect it. The question is whether MaintainX is a bridge to a larger market or a drag on a perfectly fine narrative. PMs should watch the margin trajectory in the next two quarters for early clues.
DY is printing. The Q1 beat wasn't a weather story — it was a thesis confirmation. UBS and Cantor both ripped PTs higher post-print, and the market is repricing the name for a multi-year fiber cycle.
EPS of $4.42 vs $2.72 consensus (62.5% beat). Revenue of $1.965B vs $1.67B (17.4% beat). That's a massive top-line surprise for a services name. UBS raised FY27/28 EBITDA estimates to $1,096M / $1,271M from $992M / $1,126M — calling out Building Systems as ~2/3 of the EBITDA delta.
Cantor went to $654 from $436 (still Overweight). Stock at $535 as of the note, so there's room.
FTTH was the driver. Up 33% sequentially. Cantor points out management said weather made Q1 look like Q2/Q3, but demand had to be there to capture it. Not sure we can read too much into the "pull-forward" concern — management explicitly pushed back on that frame.
"Management said permitting and preparing local projects for construction remains the limiting factor."
That's the one line that keeps me honest. The bull case works as long as the bottleneck is a rate issue, not a volume cap. Cantor thinks it's the former.
Verdict: ServiceTitan still the best pure-play on trades digitization, but the stock’s been gutted (-29% in 6 months) and needs a growth re-acceleration narrative to re-rate. Earnings on June 4 are the catalyst — usage revenue is showing life, Max program gaining steam, but the market wants proof the 18.7% Q1 guide is conservative.
Two firms out this morning — Piper Sandler ($100, Overweight) and BMO ($92, Outperform) — both holding the line. Piper sees a “path to low-20% growth” driven by Max, Voice Agents, commercial, and roofing. Their key callout: HARDI revenue (a proxy for TTAN’s gross transaction value) inflected to +10.4% in March after a downtrend. That’s the first green shoot in a while. BMO echoes the usage revenue rebound theme, noting “upside potential” after Q4 headwinds. They think sequential growth improvement is attainable, but caution it needs to be “beyond seasonal and calendar factors” for the stock to re-rate.
“ServiceTitan’s exposure to the trades continues to make it one of the better stories in enterprise software.” — Piper Sandler
The bull case: You’ve got 24.5% trailing revenue growth, a $260B+ TAM in trades, and AI tools (Voice Agents, Max program) that could improve unit economics and retention. HARDI inflecting, CTO upgrade, path to profitability this year (analyst consensus). Multiple compression is the only reason TTAN trades at 5x forward EV/S vs. 8x+ for high-growth peers. If Q1 beats and guide ticks up, that gap closes fast.
The bear case: 18.7% guide is deceleration from 24.5% — that’s not a re-rate story. The stock’s down 29% for a reason: software multiples are under pressure, TTAN’s net dollar retention isn’t published (opaque), and the trades could face macro headwinds if housing slows. BTIG cut PT to $90 citing “software multiple compression” — same stock, different multiple environment.
Bottom line: Dozens of analysts are bullish (PTs $90-$125), but the stock is $65. That’s either the opportunity of the year or a value trap depending on whether June 4 shows a growth inflection vs. just seasonal noise. I’d wait for the print — usage revenue data is the tell.
Wolfe reiterates Outperform ($320 PT) on the logistics expansion thesis. They’re leaning into the May 4 launch of Amazon Supply Chain Services — specifically Freight and Global Logistics for non-Amazon sellers. This isn’t a near-term catalyst; it’s a multi-year TAM grab. Wolfe pegs the addressable market at $1.2T across freight, fulfillment, and shipping. Even low single-digit share gains would be additive to revenue and op income. Stock sits at $274, just shy of the $278.56 52-week high — not much room for error on execution.
Wolfe Research estimates the global addressable market for Amazon Supply Chain Services at approximately $1.2 trillion.
Wolfe is effectively betting AMZN becomes the default logistics layer for third-party sellers independent of marketplace presence (fills excess capacity, improves cost profile). The caveat: scaling takes years, and incumbents (FDX, UPS) won’t roll over. But the directional bet is right — AMZN’s infrastructure advantage is widening, not narrowing.
Verdict: Target goes to $1,750 from $600 — that's a 3x jump in the PT. The stock is already up 856% in a year and sitting at $970.56, but the thesis is structural, not cyclical.
Supply constriction is the story. Susquehanna sees tight memory through 2027, driven by KV Cache offloading killing the incentive to add wafer capacity. Blended ASPs are still climbing, margins are sticky, and the valuation re-rating is warranted. This isn't a Q4 spike — it's a multi-year setup.
MU HIT $1T MARKET CAP — first US memory maker to do so. The re-rating crowd is now playing catch-up. Multiple analysts have piled on: DA Davidson to $1,500, Mizuho to $1,150, Barclays to $1,175. All leaning on the same structural undersupply narrative. The consensus cluster is now $1,150–1,750, up from a $600 base.
"Supply is now expected to remain tight through 2027, sustaining elevated margins and thus warranting valuation re-rating." — Susquehanna
Risk? The stock's blown past every PT in the last 12 months. Sentiment is euphoric. But the rate of change on fundamentals is still accelerating — HBM demand isn't slowing, and nobody's adding capacity fast enough. That's a PM's kind of risk/reward.
BOLDEST CALL ON THE STREET LANDED. Susquehanna rips its PT to $3,250 from $2,000 — a 95% implied upside from here. The thesis is pure margin duration play: supply stays tight through 2027, KV Cache offloading kills wafer adds, and AI inference is the new demand vector nobody modeled last cycle.
"Supply is now expected to remain tight through 2027, sustaining elevated margins."
That's the money line. SNDK at $1,669.47 — up 4,153% in twelve months — and the bull case still has room. Gross margins at 56%, revenue +83% LTM. This isn't a peak-cycle story anymore; the memory pyramid is structurally different.
Three forces conspiring here, all positive:
Not just Susquehanna going bold. Mizuho's FY27 revenue estimate of $45.3B beats consensus by $1.7B; EPS of $184.95 vs $180.14 consensus. Barclays upgrades to Overweight with a $2,300 target, citing innovative contract strategies that lock in pricing. S&P upgraded to BB+ from BB — zero debt, net cash, data center revenue +191% YoY.
The cluster of PTs is widening but Susquehanna is the outlier by ~$1,000. Either they see something no one else does, or they're compensating for being late to the move. Given the return trajectory, I'd bet on the former.
Bottom line: this is a margin compounder in a supply-constrained cycle with a new demand driver (AI inference) that wasn't in the original thesis set. Don't fight the tape, but respect the volatility. SNDK prints its own weather.
Cantor reiterates Neutral at $300 — the stock at $315 trades above their 30x FCF vision for CY27 ($10/sh). The dinner with execs confirmed the data center pivot (now 12% of rev, TAM growing 65% in 2026) but they flag better r/r elsewhere. Pricing flat in Q2, potentially up H2 (July 1 letter already out). No changes to capex or deprec — they have the supply. The 75% one-year rally is pricing in a lot.
"The stock trades above 30 times our vision for free cash flow of $10 in calendar year 2027, with more than 50% of revenue coming from automotive and personal electronics. We see better relative upside elsewhere."
The bullish counterpoint (from Seaport, Mizuho, Stifel, BoA) is all about AI power leverage — stage 2 and 800V transition, power >50% of data center mix. BoA named it a top pick in AI power semis. But Cantor’s caution: the multiple is already there. Worth watching if data center re-rates the analog floor, but at 54x P/E today, the narrative needs to keep accelerating.
Verdict: The optics trade is still on fire, but at 360x earnings, MS is pointing out the ceiling — and they’re not wrong to stay Equalweight. Demand commentary from the Asia AI Summit was bullish across the optical space, but CIEN’s 599% one-year run is inviting profit-taking in a crowded trade. The fundamental story holds: peers (Lumentum, Cisco, Nokia, Coherent) all point to demand ahead of supply, and CIEN likely beats revenue by 3-5% with Q3 guidance $50-75M above Q2. But the margin expansion is a 2027 story — supply constraints and mix cap near-term gross margin upside.
“Demand ahead of supply and sustained optical and cloud strength.” — Morgan Stanley, citing peer results
The bull case is real: backlog swelled $2B to $7B, and analysts are still raising targets (TD Cowen to $675, Stifel to $585, BofA to $550). The bear case is just as real: Rothschild Redburn started Neutral at $416, pointing to valuation challenges. With earnings June 4 and options pricing a 13% move, this is a binary event in a name that’s already priced for perfection. Not a short — but the r/r gets worse from here unless guidance crushes AND margins surprise.
Cantor keeps AMAT as top pick, sees 28% upside to $575. The bull case here is simple: we are still in the early innings of the AI trade for wafer fab equipment. Cantor's management channel checks point to Agentic AI and greater CPU/DRAM density being 20% ACCRETIVE TO TOTAL WFE SPENDING — that’s a structural lift, not a cycle blip.
Three things jumping out from the note:
1. Capacity doubled to support $10B/quarter in semi revenue — that implies a WFE run-rate north of $180B. Not cheap, but they’ve got the order book visibility (eight quarters booked, customer discussions beyond 24 months). 2. EPIC co-innovation center is the real moat. Getting designed into customer roadmaps earlier via the materials/process integration play. 3. Applied Global Services is finally getting an AI boost — more revenue per tool, which is why long-term guidance got raised recently.
Multiple firms have chimed in with PTs in the $525-540 range (Mizuho, TD Cowen, Lynx). Cantor’s $575 sits at the top end — they’re leaning into the thesis that this isn’t a cap-ex peak, it’s a step-function shift.
"Applied Materials’ internal analysis indicates Agentic AI and greater CPU/DRAM are likely 20% accretive to overall wafer fabrication equipment spending."
That’s the number to watch. If you believe AI drives more compute and more memory per node, AMAT is the pick-and-shovel. The 179% return over the past year has baked in a lot, but Cantor thinks the next leg comes from the multi-year order book converting to revenue and margin expansion.
JEFFERIES BUMPS PANW TO $300, KEEPS THE BUY. BUT THE REAL STORY IS THE SHORT TERM SETUP. Stock already ripped 66% since April 10, sitting just 1% below its 52-week high ($261.41). You're 4 days from the print (June 2) and options are pricing an 8% MOVE either way. Jefferies sees shares "range-bound" near-term unless organic net new ARR surprises to the upside and services revenue accelerates from last quarter's 13% y/y growth.
The bull case rests on FY27 AI catalysts (observability/identity) and growing conviction they can hit $6.4B+ FCF by FY28. $300 PT = 37x FCF. Benchmark also raised to $270 (from $200), similarly expecting beats on NGS ARR, revenue, OI, and FCF. On the other side, UBS and Guggenheim are both Neutral — valuation just too rich at 40% YTD return.
"The firm sees shares as range-bound in the near term unless organic net new annual recurring revenue materially exceeds expectations and organic services revenue stabilizes or accelerates versus the fiscal second quarter's 13% year-over-year growth."
Biggest risk here is the stock already pricing a perfect print. If they deliver in-line and guide conservatively, that 66% run gives you a lot of room to give back. PMs should be watching whether services revenue (the dreaded 13% bogey) shows any life — that's the swing factor that separates a beat-and-raise from a sell-the-news.
Lynx Equity – after sitting on the sidelines for nearly a year while NVDA underperformed peers – is now leaning into the Computex keynote (Sunday night U.S.) as a near-term upside catalyst. Their read: Jensen will use the stage to address the stock's post-earnings drift, lay out the $20B revenue / $200B TAM for Agentic AI CPUs, detail Rubin ramp within the $1T GPU AI systems outlook, and clarify networking's long-term trajectory. The stock's PE of 32.8x with a PEG of 0.29 screams cheap relative to growth – but the bear case (ASIC share loss at hyperscalers) has been compressing the multiple. Lynx sees r/r skewed to the upside for the next few days.
"The CEO has an opportunity to make the case that the AI market is large enough for multiple solutions to coexist."
The "trillion yuan feast" dinner with Foxconn/TSMC/Delta/Compal/Pegatron signals supply chain confidence. Separately, Tigress bumped NVDA PT to $425 (Strong Buy), and Wolfe flagged 30% CPU market growth by 2028 on AI orchestration. Not sure we can read too much into those as incremental signals, but they reinforce the narrative that the bear case is priced in while the long-term TAM is still expanding. Watch for volatility around the keynote – a clean Rubin timeline and Agentic AI CPU detail could re-rate the stock back toward recent highs.
Verdict: RBC upgrades PT to $250 from $219, Outperform maintained. The call is about cloud migration + AI driving observability spend, with new product innovation as accelerant. Consensus is building — multiple firms raised targets post-Q1 beat (32% revenue growth, strongest in ~4 years).
RBC came out of a management meeting more confident, citing DDOG's positioning in increased observability spend, AI adoption, and new products. The PT boost reflects higher conviction and peer-multiple expansion — not just a fundamentals reset. Other analysts broadly echoed: Benchmark to $230, Rosenblatt to $220, DA Davidson Buy (reiterated). Only Truist stayed at $190 (Hold). FedRAMP High cert adds optionality for federal workloads, but that’s a long-tail tailwind — not the near-term driver.
"RBC Capital said Datadog remains a top growth idea following a meeting with management."
(Note: the Q1 beat was 4.8% above consensus revenue and 17.6% EPS upside — hard to argue with the trajectory, but the stock is already up 8%+ since earnings. At $200+ the r/r gets trickier vs. the $190 Hold crowd. Worth watching how PMs trade this into month-end.)
Key takeaway: CrowdStrike is running hot into June 3 earnings — Jefferies just swung the PT up 55% to $775, and Benchmark chimed in at $700. The stock's already at $671 (near 52-wk high), so the setup is binary: either guidance confirms ARR acceleration into 2H, or the multiple gets a haircut. Bulls own the narrative for now, but 24x CY27 revenue is not cheap.
The thesis in one breath: Expect ~$275M net new ARR in Q1, per Jefferies' positive checks and RSA takeaways. The real catalyst is whether Q2 and FY27 guidance signals re-acceleration — that's what justifies the premium. Moody's upgraded the credit rating to Baa2 on profit growth, reinforcing the structural story. Guggenheim remains sidelined on valuation, but that's a conviction call against momentum.
"CrowdStrike continues to be in pole position for the AI race." — Joseph Gallo, Jefferies (Buy, PT $775 from $500)
What to watch: Earnings on June 3. Options market is pricing volatility. If net new ARR guidance points to growth re-acceleration in 2H, $775 becomes the next floor. If not, the stock will re-rate lower before the AI thesis can bail it out.
Mizuho bumps PT to $180 (from $175), Outperform. The call is purely a function of lifting WFE estimates by 10% for 2026 and 20% for 2027 – driven by AI logic and memory fab buildout. Not a bold incremental take, but the thesis holds: Entegris’s asset-lite consumables model creates a high-margin tailwind off every incremental tool install.
“Entegris is among the best-positioned materials names in an extending wafer fabrication equipment upcycle…its asset-lite, consumables model translates every incremental tool installed into recurring and high-margin business.”
EPS bumped to $3.55/$4.65/$5.45 from $3.50/$4.55/$5.20. Stock up 92% in a year – valuation already reflects the upcycle extension. The delta here is whether WFE spending keeps ripping into 2028 (Mizuho now modeling $5.45 that year). Given the lead times in fabs, risk/reward still favors the name on an absolute basis, but multiple expansion off here is harder to justify. The retiree/swap at Materials Solutions division (Daniel Woodland out, Olivier Blachier in) is a non-event for the investment case.
IBM’s AI security narrative gets a fresh stamp from RBC. Matthew Swanson reiterated Outperform / $300 PT, leaning into Project Lightwell as the next logical extension of IBM’s open-source enterprise dominance. The pitch: >90% of Fortune 500 runs on open-source, frontier AI is accelerating vulnerability discovery, and IBM has both the incentive and expertise to protect that layer. Revenue growing ~10% with a PEG of 0.24 — cheap for the rate of change.
“Project Lightwell is consistent with IBM’s broader positioning as a leader in open-source enterprise software enablement.”
The broader backdrop helps. The $1B CHIPS Act grant for quantum, plus the $46M contract mod from DoD, reinforce the thesis that IBM is a slow-growth stable compounder with AI tailwinds. Stifel also reiterated Buy / $290 last week. Not a high-beta name, but the risk/reward is fine for positioning into an enterprise AI security cycle.
Northland picks a side on AUR — Outperform, $11 PT (55% upside from $7.07). Stock already up 84% YTD, so this is catching a rocket mid-flight, not a basement call. Thesis: AI models (specifically OpenAI’s o1) are accelerating the shift from digital to physical agentic AI, and autonomous trucking is the most tangible near-term use case. AUR deploying 200 drivers this year — that’s real, not lab demo.
"We see demand as almost unlimited with few competitors due to the difficult technology challenge involved in autonomous driving."
Revenue growth north of 400% expected this year. The Q1 EPS beat (-$0.10 vs -$0.11 consensus) helps the narrative that operating discipline isn’t completely MIA while scaling. Physical agentic AI as a 2026+ exponential theme — AUR is the purest play on that in public markets, even if execution risk remains high. Bear case: vertigo-inducing multiples on zero current revenue, and Tesla/Waymo lurking. But Northland is leaning into the r/r of the first-mover premium in a nearly uncontested niche.
Barclays stays Equalweight / $49 PT on VSAT — that’s a 43% downside from $86.69, which tells you everything about the gap between narrative and numbers. Stock up 835% in a year, now near its 52-week high. The thesis: revenue and EBITDA both missed consensus in FQ4, FCF came in $67M below Barclays’ bogey, and FY27 guidance is for flat EBITDA (back-end loaded, with a 2% drag from a business sale). Net debt still $4.8B. The EPS “beat” (-$0.02 vs -$0.43) is noise — that’s a rounding error on a $1.2B revenue base.
“Fair Value analysis suggests the stock may be overvalued at current levels.” — Barclays
Defense & Tech grew 12% in the quarter, Comms declined 2%. FY27 outlook: Defense mid-teens, Comms low single-digit. Maritime stabilization pushed to end-2027. Equatys D2D won’t hit until 2029. Capex guided to ~$1B, FCF ~$180M excl. Ligado. This is a long-duration story priced for perfection — the PT implies it’s trading like a satellite breakout, not a $49 stock.
Verdict: STOCK DOWN 34% IN A WEEK AFTER MISS AND WEAK GUIDE — CRAIG-HALLUM CUTS PT TO $42 FROM $48 BUT STAYS BUY. They see near-term headwinds (IC softness, memory constraints, geopolitics delaying tape-outs) masking a longer-term catalyst stack: Allen TX facility ramping qualification masks with revenue late FY26, plus a massive FPD quarter. The stock at $32 is pricing in a lot of pain, but the analyst math (16x FY27 EPS + $11 net cash) implies a double — you just need patience.
"Management noted early signs of tape-out recovery beginning in May."
Q2 (Apr) missed badly: EPS $0.42 vs $0.53 consensus, revenue $209.9M vs $216.7M (-0.5% YoY). Q3 guide midpoint $211M (vs street $218.5M) and EPS $0.39-0.45. That’s the near-term drag.
Allen, TX — entered qualification mask production. Revenue targeted late FY26. They’ll migrate mainstream work out of Boise to free up Boise capacity for higher ASP advanced nodes (meaningful contribution 2027+). That’s the primary bull case: a capacity-led step-up.
FPD business — strongest quarter on record at $62M (+13% YoY), driven by AMOLED in China and Korea reacceleration. Plus they installed the first-of-its-kind advanced FPD mask writer globally. That’s real, sticky demand.
$42 PT = 16x FY27 EPS ($1.98) + ~$11 net cash. At $32 that’s ~15% downside from PT, but 31% upside from spot. Not screaming cheap, but if the tape-out recovery and Allen ramp hit, multiple re-rates. Near-term risk: more Q3 misses.
Pipeline reg chatter is noise. BMO reiterates Market Perform / $279 PT on the Reddit-fueled name (stock up 1,462% in twelve months, now $275). The worry: third-party analysis says the Green Chile lateral feeding Project Jupiter doesn't qualify for FERC's blanket certificate program. BMO's take? No impact on 2026 guidance. The real story remains the 2.45 GW Oracle order and a guidance raise that has the Street pushing PTs to $235-295.
"We currently expect no impact to Bloom Energy’s 2026 guidance despite the pipeline concerns."
Don't let the headline spook you. The pipeline is a gas infrastructure issue, not a fuel cell demand issue. Jupiter isn't going anywhere – the growth rate (+130% YoY) and expanded Oracle partnership are the bogeys that matter. BMO's neutral stance is just risk management; the rate-of-change has the bulls in control.
RBC bumps PT to $9 but keeps Sector Perform — stabilization, not acceleration. PagerDuty delivered a clean print (EPS $0.32 vs $0.25e, revs $121M vs $119.5M) but the stock is still DOWN 54% over the last year at $7.44. The thesis here is margin expansion (85% gross margin, op margin beat) and a consumption model pivot absorbing the seat downsell, not top-line reacceleration. RBC likes the Operations Cloud early traction but isn't ready to upgrade the rating.
Bottom line: cheap on earnings, expensive on narrative. The 21% PT upside is a valuation floor, not a conviction long. Until the consumption model shows durable growth (not just profitability), this stays a show-me name for multi-managers. No blockquote needed — the call is too tepid.
Verdict: JPMorgan cuts CCOI to Neutral after the data center sale catalyst clears. Stock rallied 21% since the May 5 update, now at $19.82 — below the $22 PT, but the easy money is done. Focus shifts back to the core business and a balance sheet that still carries $2.66B in debt against a sub-$1B market cap.
“The data center sale catalyst is now behind the company, with focus shifting back to Cogent’s core operating trends and balance sheet repair.”
The $225M cash sale to I Squared Capital (closing June 12) chips away at leverage — net debt/EBITDA falls from 7.4x to 6.6x exiting 2026, per JPM. That’s progress, but still miles from the 4.0x target CEO Schaeffer flagged. Shares are down 57% over the past year. At 9.6x 2027 EV/EBITDA on the $22 PT, the r/r is neutral now — no hair-on-fire bear case, but no reason to chase either.
UBS is neutral at $130 but the POS narrative is the real signal here. Stock down 32.8% in six months — market hates the multiple compression story, but UBS sees a longer-term gem hiding in plain sight. POS business is ~12% of GMV today, on track to hit ~20% by 2035, contributing 300-500bps to GMV CAGR over 10 years. That's a non-trivial re-rating catalyst if it materializes. The Street is split: RBC out at $170, Cantor cut to $115 on margins, Piper and KeyBanc both Overweight. Thrive Capital just dropped $100M into the name, betting on AI as the next growth lever.
UBS's best line:
"The firm noted that Shopify’s Retail POS business comprises approximately 12% of gross merchandise volume today and could approach 20% by 2035. UBS stated this business represents an important part of the longer-term growth algorithm and a relatively underappreciated portion of the company’s terminal value."
Bottom line: POS is the structural story that doesn't get priced in — but margins and macro headwinds keep the stock pinned near $110. If POS delivery stays on track, the $130 target feels conservative. If margins keep bleeding, Cantor's $115 is the floor. PMs should watch GMV trends and the next quarterly print for margin direction.
Roth/MKM starts OUST at Buy / $75 PT — nearly 80% upside from the $42 close. Thesis is straightforward: Ouster is selling a full-stack sensing + perception platform, not just lidar hardware. They think the company can outgrow the broader sensor market as physical AI autonomy (mobility, industrial, defense, robotics) accelerates.
No balance sheet drama, but the path is long. Roth expects cash flow breakeven by end of 2027. Revenue runs $185M growing 57% — spectacular, but we’re still pre-profit. Stock’s already up 259% in a year, so some of this narrative is priced.
"Ouster is positioned to outgrow the broader sensor market with advanced lidar and stereoscopic camera offerings that enable an AI-compute and software-based perception platform."
Recent deals underpin the story: ARGUS for drone interdiction, Gecko for industrial inspection, Fujifilm for color + lidar fusion. Rosenblatt just bumped PT to $53 (from $40) after the ARGUS announcement. Bull case is a platform-shift story; bear case is valuation on unprofitable revenue. Right now the market is buying the physical AI tailwind.
H.C. WAINWRIGHT KICKS OFF COVERAGE WITH A BUY AND $14 TARGET — a ~80% implied upside from $7.79. The call leans heavily on Draganfly’s two-decade track record, full-stack positioning (hardware + AI software + services), and real-world validation in Ukraine. The firm is effectively saying this is not another speculative drone name — it's a field-tested defense/industrial play with multiple verticals.
“Draganfly has over two decades of drone innovation experience and offers full-stack end-to-end drone solutions that are integrated, compliant, and field-proven in Ukraine.”
Recent catalysts support the narrative — a DEVCOM contract for counter-drone systems, the Blitz payload platform exclusive deal, and the planned acquisition of Skip Dynamix for up to $7.5M. Needham already reiterated Buy / $12. This is a small-cap ($285M market cap) with 25% revenue growth and a widening moat in defense-adjacent drone ops. R/R looks favorable if you believe the Ukraine proof-point translates into more NATO contracts. Risk: cash burn, execution on the acquisition integration, and general small-cap volatility. But the signal here is that H.C. Wainwright sees enough bottom-up conviction to plant a flag at $14. Worth tracking for PMs with a drone/defense tilt.
Buy-the-dip case intact, but the composition of this beat gives shorts ammo. BRZE posted a revenue and cRPO beat, but the upside was all Professional Services — Subscription revenue was in line. Operating margin met guidance, but management didn’t raise FY margin guidance. Shares -28% YTD, now at $22.97 vs Cantor’s $38 PT (65% upside). The firm reiterates Overweight, calling it a buying opportunity for longer-term investors.
"Short-term noise is a buying opportunity for longer-term investors." — Cantor Fitzgerald
Other firms are staying constructive. DA Davidson, Stifel, and Citizens all maintain Buy-equivalent ratings (PTs $30–$35), citing market share gains, AI demand, and improving net retention. Mizuho trimmed its PT to $32 from $40 on valuation but kept Outperform. The fundamental thesis — core demand strong, AI tailwinds, 30% top-line growth — hasn’t broken, but the market wants margin acceleration and subscription-only prints, not a professional services fudge.
Bottom line: the narrative is bifurcated — bulls see a 65% upside setup from a washed-out price, bears focus on the beat quality and stagnant margin guide. Right now, rate of change on sentiment is negative (14 earnings revisions down), but the Street still rates it a Strong Buy. That tension creates an r/r that favors the dip if you’ve got a 12-month horizon.
Walmart deal is the narrative driver, not the Q1 beat. RBC maintains $20 PT (42% upside from $14.09) on the expanded partnership with Walmart Connect. Magnite’s SSP now enables Walmart-controlled audience decisioning across multiple DSPs — starting with VIZIO CTV inventory and closed-loop sales measurement.
"The firm assumes the partnership will take time to ramp but views it as an incremental opportunity and positive for sentiment around the company’s differentiation."
Yep — the timing is good (ad-tech sentiment improving, Magnite still cheap vs. fair value), but this is a 2027 story. Q1 EPS of $0.13 beat by $0.02, revenue $164.4M vs $159.2M consensus. Stock barely moved after hours. The Walmart hook gives PMs a CTV/commerce media angle to lean on, but don't confuse a positive headline for near-term acceleration. Incremental, not transformational — yet.
Verdict: Uber is a deep value trap or a massive opportunity — and the market can't decide. Stock at $70.92 (just 3% off its 52-week low) despite 19 upward EPS revisions, an S&P upgrade to BBB+, and the clearest AV monetization path in the sector. Citizens is leaning into the Waymo/Ojai production ramp as the catalyst to break the downtrend. The Delivery Hero bid ([reportedly rebuffed at €10-11.5B](https://www.investing.com/news/stock-market-news/uber-makes-bid-for-delivery-hero-93CH-1234567)) shows management is willing to deploy capital — but the market is treating that as a distraction, not a signal.
"We believe autonomous vehicles fundamentally expand the ride share market while Uber’s non-Waymo autonomous vehicle partners continue to move toward commercialization, with multiple partners expected to come online in the second half of 2026." — Citizens JMP (Market Outperform, $100 PT)
The macro setup is horrid — consumer weakness, regulatory overhang, AV capital intensity — but the rate of change on AV miles hitting Uber's network is accelerating. If the Ojai ramp is real, $70 is a gift. If not, the 52-week low isn't the floor. 3P r/r is skewed to the upside, but only if you believe the AV thesis actually prints this year.
Analyst day left the room cold, but Truist says buy the dip anyway. Raised PT to $215 from $180, CY2027 EPS up to $5.65 (38x multiple). The messaging on AI exposure, Penang/Syracuse/Eau Claire ramp — "less than expected." Yet the firm leans into the bigger story: cash flow deployment into differentiation and M&A, higher margins on the other side.
"The analyst day messaging was less than expected, but we view the modest disappointment as a buying opportunity – the company’s story continues to improve."
Needham and Stifel joined the chorus post-quarter (both Buy, PTs $208 and $205 respectively). The stock at $188.68, near the 52-week high of $200.68 (536% return last year — yes, that’s real). Q1 EPS beat ($0.75 vs $0.67e), rev $846M vs $787Me, and CEO guidance for FY2026 rev at $4B (+38% y/y). All pointing to AI-driven vertical integration.
One caveat: InvestingPro flags TTMI as overvalued at current levels. The 38x forward multiple on the Truist number is a high bar — but the rate of change story is undeniable. If you believe the AI complexity theme and the M&A pipeline, the current pullback from the high is your entry. If you don’t, that multiple is a trap. PMs can decide which side of that trade they want to be on.
Verdict: Cantor sees ~53% upside here, and they're not just extrapolating the 666% run — they're arguing the market is still missing the capacity story. PT to $99 from $77, Overweight maintained. The core thesis is straightforward: street models aren't pricing in the 670 MW of gross capacity hitting in 2027, let alone the 2028 pipeline. GPU pricing keeps stepping up, which puts upside pressure on IREN's year-end ARR target. Target multiple bumped to 7x CY28 EV/EBITDA from 6.5x.
"The market’s failure to account for approximately 670 megawatts of gross capacity that IREN is bringing online in 2027"
The recent $1.6B Dell deal for Blackwell systems and the $3B convertible note offering (1% coupon, 32.5% conversion premium) reinforce the thesis — they're funding the growth, not punting. The Awaken acquisition is small ball but shows operational ambition. BTIG also reiterated Buy at $80 (lower FY26 rev, higher FY27).
Bottom line: Cantor's PT is at the higher end of the Street ($41-$126 range) but the r/r argument holds — capacity value appreciating daily, and IREN still trades at a discount to where the 2027/2028 EBITDA build should land. Risk is execution and timing on those MW ramp-ups, but the delta between current price and forward capacity value is the play.
Jefferies is doubling down. Raised PT to $56 from $44 (Buy) after FPS reported a monster Q3 — orders UP 308% vs 100%+ for the coverage universe. This is pure electrification leverage through a custom niche business, and the backlog is stretching out. They’re lifting FY2028 sales/EBITDA estimates by 15%/16%.
"Forgent has high leverage to the electrification theme through its custom niche business... Further margin expansion remains possible."
The 64% YTD run to $52.62 already prices in a lot, but Jefferies sees FCF turning meaningfully positive in FY2027 and inorganic catalysts on the horizon. Caveat: the secondary offering (42.28M shares at $47, including 28.5M from Neos) hit after-hours — stock dropped 8.5% on the dilutive overhang. That’s the r/r for PMs: strong underlying momentum vs near-term supply.
Jefferies cuts PT to $9.80 from $11.70, keeps Buy. Stock at $7.79 — near 52-week low, trading at 4.95x P/E with a 7.3% dividend yield. The thesis hinges on ad spending divergence: internet apps and auto are strong, smartphones are soft. Q1 revenue in line, non-GAAP EPS beat in USD, but Q2 estimates trimmed because of that vertical mix shift.
World Cup advertising patterns are different this cycle thanks to time zone quirks — Jefferies thinks that's a wildcard, not a disaster.
"We expect Weibo to demonstrate focus on return on investment in spending."
R/R is asymmetric at these levels — 26% upside to the new PT, 5x earnings, and a yield that covers the downside risk if the ad recovery takes longer. Light coverage means no herd to fade.
TSM — CoWoS is the binding constraint, not node capacity. NVDA confirmed as TSMC's largest customer at >20% of revenue. Zero-sum packaging allocation means AMD and custom silicon are structurally capped. TSMC Arizona turning a profit (NT$16.9B vs full-year '24 loss of NT$14.2B) validates reshoring economics — changes the narrative from subsidy-dependent to operationally viable. Watch the June 4 shareholder meeting for pricing updates and CoWoS outlook. The competitive moat is intact.
ASML — Cleanroom constraint easing faster than feared. 2026 guidance raised to €36-40B, explicitly underpinned by improved pedestal visibility. Dassen's math: 80 Low NA EUV systems in 2027 deliver ~2x wafer-hour capacity of 2025. High NA EUV remains "not prime time" for DRAM today — this is a 2028+ roadmap story, not the current bottleneck. The raised guidance is execution scheduling, not demand allocation.
SFTBY — SoftBank committing €75B ($87B) to France AI infrastructure, 5GW total. Phase 1: 3.1GW by 2031 in Hauts-de-France with Schneider Electric as Dunkirk industrial partner. This is the first large-scale European AI DC signal. The 5-6 year deployment horizon is well outside the current grid bottleneck window — the constraint is power, not compute. Stock up 70% in 2026 on Arm and OpenAI stakes — the infrastructure buildout is not yet priced.
LSCC — Lattice Semiconductor CEO stating hyperscaler capex for 2027 exceeding $1 trillion and continuing to grow. This is a direct data point from a semiconductor management team validating the structural capex growth. Consistent with Goldman's compute >$1T by 2030 projection. The question is whether this accelerates or is already in the consensus view.
SBGSY — Schneider Electric is the industrial production partner for SoftBank's 3.1GW Dunkirk AI DC cluster in France. At $3.5M/MW AI DC anchor, 3.1GW implies ~$10.9B in facility infrastructure revenue — creating a large multi-year backlog for Schneider's DC power and cooling segment. This is a scale-mover for a name that's already been a beneficiary of the AI infrastructure buildout.
Hearing MF's channel check: "demand is ahead of buildout, you hear that universally from every CSP, hyperscaler, neocloud. Rental rates per GPU-hr stable or increasing across most sources." xAI has a 90-day optionality to reclaim capacity. This is the most direct demand durability signal in the feed — contradicts any narrative of demand softening.
Word is the $50K NVDA crash call circulating is dangerous narrative. No accounting data supports it. The actual risk is mechanical: crowded positioning meets a fundamental catalyst that doesn't materialize. GS TMT Most Short up 8 of 9 weeks (+25% YTD), hedge funds running out of IT buyers. The elongation threshold is being tested.
Channel checks suggest Goldman raised SK Hynix OP 4-24% and Samsung OP 5-34% through 2028. These are substantial revisions for a memory cycle extending beyond near-term restocking into sustained structural HBM demand. The 2027 and 2028 revisions are the key numbers — implies Goldman sees the cycle lasting 2+ years longer than consensus.
Hearing Vista Equity's Robert Smith gave a clear articulation of the context-not-compute efficiency gain: insurance claims evaluation drops from $8M → $3-4M → $200K as you layer general LLM (2x improvement), then proprietary systems with fraud claim context (15-20x further improvement). The mechanism is domain-specific context at the platform layer, not model capability. This is the second unlock bottleneck (orchestration) in action.
Word is the Amazon MCP gap is real — why isn't there a production-grade Amazon MCP connector for AI agents to execute transactions? If MCP becomes the agent-to-tool protocol, every missing connector is a workflow bottleneck that hasn't surfaced as a market event yet. Watch for this.
Hearing Meta may have the most incremental GW of training capacity coming online in 2026 among AI labs — including neocloud capacity, potentially ahead of the ~6GW OpenAI and Anthropic projections. The MSL team is reportedly producing competitive work. First public signal on Meta as a legitimate frontier model challenger.
Channel checks suggest Qualcomm opened Computex calling 2026 "the year of agents" with Snapdragon C at $300+ with Acer, HP, Lenovo onboard. The open question: what silicon backs the agent claim? Qualcomm is positioned for on-device AI inference if the agent narrative materializes, but AI PC adoption has been pushed to 2028.
Word is Jabil is gaining recognition as an AI infrastructure systems integrator combining liquid cooling, silicon photonics, and power distribution. Revenue growth strong, margins expanding — validates the mix shift narrative. Risk is customer concentration with a small number of hyperscalers.
Hearing optical and photonics stocks (LITE, AAOI) have gone parabolic and are now rolling over technically. Pullbacks towards 40-week moving averages may occur late summer/early autumn. No fundamental change in demand — just technical digestion after extended runs.
Channel checks suggest Dell's 33% surge on earnings puts pressure on HPE to deliver a strong print and guidance. Dell's first-mover advantage with NVL72 means HPE may lag in capturing the highest-value portion. The analyst community is still underestimating AI infrastructure demand, so HPE may also beat — but margin compression risk is higher given Dell's early lead.
Word is Berkshire's $8.5B Taylor Morrison acquisition in Greg Abel's first major deal is a non-tech signal. Capital deployment outside AI. Indicates Berkshire is finding value outside the AI mania rather than participating.
Hearing Murata and Nippon Chemicon are deliberately accepting thinner margins to block Chinese competitors from gaining commodity MLCC share. The upside is in MLCC equipment and materials — the upstream capex supporting Japanese manufacturers' capacity expansion, not the finished products.