Good morning.
Futures flat to slightly green after yesterday’s bloodbath — S&P -1.4%, Nasdaq -2.22%, but breadth was actually decent (advancers beat decliners 2,529 to 2,262). KOSPI crashed NINE PERCENT — fifth-largest single-day drop on record — triggered by a single Korean article about SK Hynix slowing memory expansion. Leveraged ETF rebalancing amplified the entire move. Not a thesis break. A tape break.
Micron reports tonight. That’s THE catalyst — positioning feels like “Nvidia earnings day” from 2023/24. Memory has become the single biggest cost driver in AI servers ( >50% of BOM by 2027 ), and the selloff yesterday created the most asymmetric setup in months. TSMC pushing 5-10% wafer price hikes across all advanced nodes — management is looking at memory’s pricing power and saying “us too.” That’s a direct tailwind for TSM and a pricing signal for the entire supply chain.
CXMT’s looming IPO is the other big story — projecting $55B revenue this year, potentially $80B+ next year, and they’re only pricing DRAM at a 5-10% discount to incumbents. That’s margin preservation, not dumping. Changes the memory landscape permanently.
Three themes frame today: (1) Memory structural demand vs. leverage-induced panic — we’re buying the dip in MU, WDC, and the Korean semicap names trading at single-digit multiples (Pemtron at 5x 2027 EPS). (2) TSMC’s pricing power confirms the foundry bottleneck is tightening — hyperscalers will accept the hike because there’s no alternative. (3) Rotation out of AI momentum into defensives is real, but it’s late-stage — the median stock was flat yesterday. We’re not chasing staples.
We’ll hit up MU, TSM, and NVDA first, then get to semiconductors and memory.
ACN IS GETTING ABSOLUTELY WRECKED. Down 54% over the past year, another 23% last week alone, now scraping 52-week lows around $118-125. This isn't just a guidance miss — it's a structural narrative break that has the street scrambling.
Three analyst actions this morning, and the collective thesis is grim. Truist cut to $150 (Hold), RBC to $175 (Outperform), Mizuho to $226 (Outperform). The $150-226 range tells you everything — nobody agrees on the floor, but everyone sees lower.
The common thread:
BEAR CASE: This isn't cyclical — it's structural. The Middle East conflict is a new variable with no clear end. AI isn't a tailwind yet, it's a margin headwind. Truist's call that "a larger portion of the guidance would be in play" due to macro uncertainty is the kind of language that keeps PMs up at night. When the largest IT services firm in the world is guiding down and watching bookings fall, you have to ask what the bottom looks like.
The strongest single read comes from RBC, who is still Outperform but slashed PT by $78:
"We view fiscal year 2027 as a recovery year for Accenture, with underlying AI demand remaining healthy despite near-term pressures on bookings and guidance."
That's a bet on the rate of change reversing. But the rate of change is currently negative, and momentum is the enemy of value.
Bottom line: ACN is pricing in a recession + AI disruption + geopolitical risk. The question is whether the stock at $120 (10x earnings, 5% yield, 17% FCF yield) is pricing in too much or just enough. The buyback is aggressive, the RSI is oversold, and 14 analysts have cut estimates. That's a crowded bear trade. But the narrative isn't turning yet — wait for bookings inflection before stepping in.
Verdict: Backblaze just got the stamp of legitimacy, and the market is repricing it in real time. The CoreWeave deal isn't just a revenue pop — it's a narrative shift from "second-tier storage vendor" to "AI infrastructure pick-and-shovel play." Two upgrades in one day, stock +40% to $11.47, now hanging near the 52-week high. This is a rate-of-change setup, not a value argument.
Both William Blair and Craig-Hallum upgraded BLZE on the same five-year, $335M multi-exabyte agreement with CoreWeave. That's a NINE-FIGURE COMMITMENT from one of the biggest neocloud players. Blair went to Market Perform (from Underperform), Craig-Hallum went to Buy (from Hold) with a $16 PT. The collective take: this validates Backblaze's HDD-based storage layer in the emerging AI infrastructure stack.
Blair's upgrade specifically addresses the two overhangs they flagged: AI defensibility and execution. The contract size and customer quality make those concerns look stale. They kept FY2026 estimates unchanged (ramp timing), but raised FY2027. Craig-Hallum is more aggressive — they expect top-line acceleration and meaningful upward revisions to both revenue and profit.
Key line from Craig-Hallum (the stronger source):
"Landmark $335M Deal With CoreWeave Changes The Game. Top Line Acceleration And Meaningful Upward Revisions To Revenue And Profits On The Way. Shares To Head Higher On EPS-Based Valuations. Raising Rating To BUY, Increasing Price Target To $16."
Bull: CoreWeave doesn't sign nine-figure deals on a whim — this signals Backblaze is now part of the AI infrastructure blueprint. Multi-exabyte scale with a sophisticated neocloud customer provides a referenceable, repeatable sales motion for the next 12-18 months. Combined with Q1 beat (EPS $0.04 vs -$0.004 consensus, revenue $38.7M vs $37.8M), the fundamental trajectory just inflected.
Bear: $335M over five years is ~$67M/year — real, but not transformative in Year 1. Blair kept FY2026 estimates flat. The stock has already +74% YTD, and at $11.47 it's pricing in a lot of optimism before any financial impact shows up. If the CoreWeave ramp disappoints or other neoclouds don't follow, the multiple re-rating unwinds quickly.
At the $16 PT from Craig-Hallum, that's roughly 40% upside from here. Blair's Market Perform rating suggests they think fair value is around here or a little higher — they're not chasing. The EPS-based valuation argument works if BLZE can sustain margins as revenue scales. Q1's margin profile looked strong (cash flow positive?), but the street needs to see the FY2027 numbers before the full story is written.
Key watch item: next earnings call guidance. If management raises the FY2026 outlook (even modestly), this stock rips higher. If they stay conservative, expect a drift lower as the "now what" pressure builds.
BLZE is no longer a story stock — it's a show-me stock with a much higher floor.
Verdict: The $367 stock is pricing in zero optionality. Two analyst notes today — one tepid Hold at $415, one aggressive Outperform at $550. That's a 45% spread. The stock doesn't care yet. But the rate of change in the bull case (Copilot usage billing + DeepSeek hosting optionality) is real. The bear case (Azure slowdown, Xbox bleed, capex overhang) is also real. MSFT at 3% above its 52-week low is a proxy for the entire AI narrative — and right now, the market is betting on disappointment.
Stock at $367.34. Down 24% YTD. Sits just above $356.28 low. Two distinct data points hit in the last 10 days:
Stifel (Brad Reback) – Hold, $415 PT Not bad, not good. Just... there. Reback acknowledges the power deal but treats it as a long-duration option with no near-term earnings impact. The hold rating at $415 (vs $367) implies +13% upside — acceptable but hardly a conviction call. The real message: wait for FID, wait for Azure reacceleration. Nothing to act on now.
JMP Securities – Market Outperform, $550 PT This is the interesting one. JMP is leaning hard into AI monetization via Copilot Cowork's usage-based model. The kicker: Axios reporting Microsoft is exploring a Microsoft-hosted version of DeepSeek V4 (or another open-source model) as a lower-cost alternative inside Copilot. That's a structural margin and adoption lever. If Microsoft can offer competitive inference cost without ceding data control, it solves the "too expensive to scale" critique.
"Any DeepSeek option would be optional for customers and hosted on Azure, keeping customer data within Microsoft’s cloud." — JMP Securities, paraphrasing Axios
JMP's $550 is 50% above spot. That's not a typo. It's a bet that the market is mispricing the AI platform's optionality.
Bull case: Copilot Cowork usage billing + potential DeepSeek cost optionality = faster enterprise adoption and better unit economics. $1.4B Air Force deal (Dell Federal) shows government demand is real. Capex heavy, yes, but the installed base is unmatched. At 24x trailing with a 25%+ FCF yield, MSFT is the cheapest mega-cap AI play in the group.
Bear case: Azure growth is decelerating (shareholder lawsuit alleges nondisclosure). Xbox margins down to 3% — layoffs incoming. $30B+ annual capex with no clear ROI timeline for the majority of AI spending. Project Kilby is 4+ years away. The stock is down 24% for a reason: the market sees a massive investment cycle with uncertain demand elasticity.
Two analysts, two worlds. Stifel's $415 is the floor — it's a "don't fight the tape" hold. JMP's $550 is the ceiling — it's a "buy the fear, sell the complacency" call. The truth is probably in the middle, but the stock is already pricing the worst case. PMs should watch Copilot Cowork billing data in the next quarter and DeepSeek hosting announcements. If either breaks positively, you're early. If not, the floor holds. r/r is skewed to the upside from here, but it's a waiting game on catalysts.
Delivery expectations are the only game in town this week. Both Baird and UBS are out with their Q2 number tomorrow (market open July 2). The range tells the story — and it's a wide one.
Q2 ESTIMATES RANGE FROM 392K TO 420K. Baird is the low man at 392.9K (below 401.1K FactSet consensus). UBS sits at 405K (up from their prior 380K). Goldman apparently went 420K. Buyside whispers are 400-420K per UBS. This is a binary print setup — miss low and the $1.44T market cap gets heavy.
Baird (Outperform, $522 PT): Bullish on the name but underwhelmed by the number. They see Q3 bouncing to 461.5K vs 454.1K consensus, Q4 at 469.3K. Full-year 1.68M vs 1.66M consensus. They're leaning on international registration data and third-party trackers. Also floated the SpaceX merger chatter — classic tie-your-bull-case-to-Elon's-other-company move.
UBS (Neutral, $364 PT): More conservative structurally. 405K delivery est, 13.4 GWh storage (up 53% QoQ). But here's the rub — STOCK AT $388 IS ABOVE THEIR $364 TARGET. That's a yellow flag for PMs running relative value books. They see buyside expectations at the lower end of the range.
Bull case: Baird's $522 target implies 36% upside. Combine FSD regulatory momentum (Finland approval ahead of EU decision, Netherlands already greenlit) with storage compounding and the SpaceX narrative — this isn't just a car company anymore, it's an energy + autonomy platform. Baird's thesis hinges on the margin story as deliveries re-accelerate into H2.
Bear case: $388 on 373x P/E (per UBS) requires every single thing to go right. UBS is effectively saying "you're already pricing in perfection." One miss on deliveries and the multiple contraction will be violent. Plus NHTSA just opened a fatal crash investigation in Texas — regulatory overhang that the bulls are ignoring.
"We acknowledge the possibility of upward bias if Tesla closes the quarter strong, but our base case remains below consensus and buyside expectations." — UBS on Q2 delivery risk
Jefferies reinstates at Hold, PT $453 — the "priced in" thesis. The 20% run to near ATH ($438, 3% off $451 high) already bakes in recent PJM and FERC tailwinds. They roll forward to 2029, add ~$2B excess cash flow, but trim upside from unannounced gas data center deals. The incremental positive from PJM's backstop procurement shrinkage and FERC's interruptible load update? Fully reflected.
The real rub: additionality and affordability kill the easy trade. Jefferies sees FERC opening a window for premium data center deals on existing assets. But material new generation is required to meet affordability commitments. TLN's "hybrid" model (existing + new gen) partially addresses that — but dampens upside. They're modeling ~1.5 GW storage NPV and ~10% probability on data center deals. Hard to underwrite in a world where regulators care about unplugging grandmas.
"FERC opened a window for premium data center deals on existing assets, though Jefferies believes material new generation will be required to address affordability commitments."
Rest is noise: Q1 EPS $1.33, rev $1.13B, $47M/yr debt refi savings. New GC. None of that moves the needle. Stock's fully valued on current visibility — next leg needs either a de-risked new-build deal or a step-change in probability. Not there yet.
Verdict: FERC just handed BE a structural tailwind, but the stock already baked it in at $342. UBS reiterated Buy with a $322 PT (now a stale bogey) after the Commission agreed to fast-track grid connections for large data centers – a direct needle-mover for Bloom’s on-site fuel cell pitch. The thesis: shorter interconnection timelines remove a key bottleneck for AI-driven demand, and Bloom’s solid-oxide platform stands to benefit as hyperscalers seek speed-to-power.
“The key goal is to shorten the timeline and remove bottlenecks in grid interconnection for very large loads.” — UBS’s Manav Gupta
But the setup is messy. Bernstein just initiated at Market Perform ($276) and BMO has a couple of takes – Outperform at $279 but flagged a project pause with Tallgrass Energy AND a separate Market Perform on FERC compliance issues for a pipeline. That’s three overlapping signals under one roof. Net: the regulatory tailwind is real, but BE is at 1,412% Y/Y returns. The easy money is in the past. Rate of change on the narrative is positive, but the valuation is already pricing a lot of perfection. PMs should watch for FERC implementation speed and any project-level hiccups (Green Chile pipe, Tallgrass pause) – those are the real swing factors from here.
Deep value trap or just early? TD Cowen is holding the line with a Buy and $34 PT (143% upside) on a stock that's DOWN 69% IN THE PAST YEAR and shed another 14% just last week. Their call: the Waves growth narrative is intact, just a timing issue as AI inference workloads ramp. The market doesn't believe it.
The bull case rests on two planks. First, the data center asset sales ($225M from 10 facilities, with 14 more to go) generate cash and simplify the story. Second, AI inference is a real tailwind for Cogent's network — it's not if, but when. The sum-of-the-parts at 9.5x assumes 6-8% long-term growth.
The bear case is obvious: the market sees a shrinking business with a management team selling assets to fund operations. JPMorgan downgraded to Neutral post-deal announcement (not exactly a vote of confidence). The Q1 revenue miss ($239.2M vs $241.3M est) doesn't help.
"Lackluster Waves growth is already reflected in the recent stock pullback. We view Cogent Waves traction as a matter of timing rather than uncertainty."
At $15.19, you're betting the analyst is right about timing. If he's wrong, there's a long way down from here.
MP is a strategic national asset, not a victim. China just put MP on its export control list alongside defense contractors. BofA says shrug — little to no impact. The stock is up 61% in the last year, trading at $61.14, and BofA keeps $85 PT. DA Davidson also reiterated Buy at $82. Thesis is straightforward: MP is the only vertically integrated rare earth magnet manufacturer outside China. They've already been de-risking supply chain away from China. This list move just reinforces the narrative.
"MP Materials as a strategic U.S. national asset" — BofA's Lawson Winder frames the export control listing as a positive signal, not a threat.
The Q1 beat (EPS $0.03 vs est -$0.01, rev $90.65M vs $73.57M) backs up the operational trajectory. After-hours dip was noise. PMs should read this as a defense play tied to rare earths — not a China exposure nightmare. The export control list is a badge of honor here.
TD Cowen doubles down. Raised PT to $70 from $55 (Buy) — stock already up 124% YTD, trades at $58.80. Thesis: customer destocking is done, and the market is about to re-rate ALGM on secular growth in automotive sensing and AI data center power. They call it a top SMID pick with favorable r/r.
"We expect the market to recognize Allegro’s secular growth opportunities in automotive and AI data centers… profitable mid-teens percentage growth over the long term."
The 23% LTM revenue growth helps. But at this valuation (nearly 3x the YTD low), PMs should ask: how much of the "destocking over" narrative is already priced in? The 124% move suggests a lot. Still, TD Cowen sees more room — and they’ve got a decent track record on analog names. Watch the next print for proof that de-stocking is truly behind them, not just a hope.
The ChatGPT tailwind is real. DA Davidson reiterates Buy/$24 (44% upside from $16.57, near 52wk low of $15.57). The key number: 2,000+ brands now advertising on ChatGPT via Criteo’s platform, doubling from 1,000 in early May. Conversion rates beat other referral channels, CTRs are 2-3x higher. This is a rate-of-change story in a stock priced for near-zero growth (7.97x P/E, 21% FCF yield).
"Criteo now has over 2,000 brands advertising on ChatGPT through its platform... conversion rates continue to outperform other referral channels."
The caveat: the stock is cheap for a reason. Multiple firms (Benchmark, DA Davidson itself, Bernstein) slashed PTs post-Q1 on a reduced FY2026 outlook tied to geopolitical/ad spending headwinds. The bull case is a binary bet that ChatGPT ad dollars scale faster than the market discounts. The 7.97x P/E bakes in a lot of pessimism – if the OpenAI partnership keeps doubling, that multiple snaps. If macro drags the core business, the floor is $15.
RBC bumps PT to $80 (from $72) on sector multiple expansion, stays Sector Perform. Management meetings offered no near-term color, but the call was about the structural story: the worst in Mobile RF content is behind us, 6G and Physical AI are the longer-term hooks, and Broad Markets (WiFi, DC, Auto) is recovering. The QRVO deal timeline tightened to late 2026 from early 2027 — no real regulatory visibility though.
"The worst appears behind in Mobile from a content standpoint, with management highlighting longer-term RF content opportunities in both iOS and Android markets driven by 6G, Physical AI, and a potentially consolidated supplier base."
The action: OPTIONS ACTIVITY SPIKED TO 32,697 CONTRACTS — HIGHLY CALL-SKEWED — HIGHEST SINCE FEB '25. A bet that the deal closes and re-rating continues. But the stock already +22.6% YTD, trading at $74.94 vs the new $80 target. Not much upside from here without a catalyst. The broader chip selloff (Feb 24) hit SWKS too — risk-off tape is the near-term lid. Limited r/r until we get regulatory clarity or quarterly proof that Broad Markets recovery is accelerating.
RBC re-ups Outperform, $300 PT on OpenAI cyber hookup. The real signal here isn't the PT — it's IBM playing the AI enterprise security adjacency game, not the LLM arms race. They're plugging frontier models into their consulting workflow via a read-only, governed architecture. That's smart positioning for a legacy tech giant that can't out-spend hyperscalers but can out-trust them in regulated environments.
"We view the announcement favorably as IBM continues to invest in security by focusing on adjacencies to its core competencies." — RBC's Matthew Swanson
The new app security service builds on Project Lightwell (backed by $5B from IBM + Red Hat). Clients start with targeted evaluations, scale to continuous monitoring. IBM Consulting Advantage is the delivery vehicle. This is incremental, not transformative — but it's rate-of-change positive in a narrative where IBM needs to show it's doing something with AI beyond Watson-era PTSD. Dividend streak (30 years) remains the floor.
Truist raising PT to $205 but staying Hold says it all — the optical/solar narrative is well understood, the question is how much is left. UBS ($223, Buy) and Wolfe ($230, Buy) are more constructive, leaning on the NVIDIA partnership and $35B 2030 sales target. Consensus is already pricing 27% EPS CAGR through 2030 (Truist's own estimate). Stock at 46x 2027 P/E after a 291% run — rich even for a compounder.
The Optical segment is the AI infrastructure story, Solar adds a clean tailwind. Combined they're 50%+ of revenue and expected to drive ~17% CAGR 2026-2030. Management's Springboard extension to $35B by 2030 is a big round number, but the path relies on operating leverage from those two segments. The NVIDIA partnership (three new US facilities, 3,000 jobs) is tangible.
"The Hold rating reflects a lack of upside to consensus estimates and a tempered view on valuation."
That's the bear case in one line. Bull case says multiples compress as earnings catch up, but at 36x 2028 P/E you're still paying for perfection. Rates of change matter here — the next catalyst is Q2 2026 print (Truist raised estimates ahead of it). Watch the incremental.
Verdict: The Fin deal is a smart bolt-on for AI agent capability, but the stock is a value trap narrative until organic growth re-accelerates. $150 is pricing in a lot of pain.
Salesforce shelled out $3.6B in cash for Fin (formerly Intercom) – an AI-native customer support platform with its own proprietary model (Apex). This is a tactical play to plug Agentforce into the SMB/commercial segment faster, bypassing the long-tail integration slog. The stock is down 46% from highs, trading at a P/E of 17.3x with a 77.6% gross margin. That’s cheap for a reason: revenue deceleration and zero near-term guidance revision.
Citizens reiterated the $315 target, which is essentially 2x the current price. Others are circling: UBS neutral, Jefferies and Canaccord stay bullish on the AI narrative hook. Monness Crespi Hardt just upgraded to Buy, calling the valuation washed out. One clear consensus: the market wants to see Agentforce revenue materialize before it pays up.
“Fin’s packaged offerings and proprietary models will complement Agentforce with additional fast-to-value deployment options for service organizations, particularly SMB and commercial customers that need to launch quickly and integrate with existing systems.”
The deal closes in FQ4 FY27. No change to FY27 guidance or buyback. This is a long-lead catalyst – not a near-term EPS pop. The r/r is asymmetric if you think management can execute, but the tape doesn’t trust a turnaround story yet. Watch for any Agentforce attach rate disclosures on the next print.
Verdict: Morgan Stanley is late to the party but right on the narrative. PT to $130 from $90 (still OW) on China memory expansion driving revenue visibility and ECP adoption scaling. Stock already up 318% in the past year — this isn't a discovery call, it's a positioning extension.
MS’s Charlie Chan is betting the market still underestimates ACMR’s broader WFE and advanced-packaging pivot beyond wet cleaning. They’re factoring in incremental non-mainland backend opportunities too. The improving operating leverage as new tool categories scale justifies the higher target.
Q1 was a beat ($0.34 vs $0.28 consensus) on $231.3M revenue — nothing new, but it confirms demand momentum hasn't peaked. The $6.8B market cap at 75x trailing P/E is stretched for a China semi equipment name, but the rate of change in new product adoption (ECP, packaging) is the offset.
"In our view, the market still underestimates ACMR’s transition from a China wet-cleaning leader to a broader WFE and advanced-packaging equipment platform."
That's the bull case in one line. Risk: the 2.9M share offering at $52 (bogey relative to current $98) is a small overhang, but Tekne Capital taking it down suggests informed money sees more upside. Keep it on the watchlist — not initiate new longs here, but own if you can stomach the multiple.
Verdict: TD Cowen just went $73 PT, calls FPS best SMID cap idea — and they’re not alone. The data center “picks and shovels” thesis is tightening. Record Q1 2026 US leasing at 9.4GW, lead times elongating, and FPS sitting on 85% expected revenue growth this year. Both Jefferies ($56 PT) and TD Cowen ($73) are Buy-rated, though the 119% YTD run and a $47 secondary (32.8M shares) adds noise.
“Forgent Power Solutions is well positioned to capture NT upside.” — TD Cowen
The bull case: backlog extending, market share gains, and a sector that’s not slowing down. The caveat: the stock is trading near a $66 52-week high, and the secondary overhang (Neos + corporate selling) limits near-term torque. Still, rate of change on orders trumps balance sheet noise here. PMs should watch the next leasing print and any capex commentary from hyperscalers. FPS is the small-cap leverage play.
BofA raises PT to $195 (still 12% below the $222 print) but keeps Underperform. The sell-side is bifurcated: Bernstein upgraded to Outperform at $210 this week, Cantor sits Neutral at $200. The catalyst is Wednesday's Investor Day — QCOM is expected to lay out a $2-5B near-term TAM in data center by FY27/28, across agentic server CPU, decode AI accelerators, and edge/physical AI. BofA is rolling forward estimates to CY28, hence the higher target, but they're not buying the narrative yet.
The bull case (Bernstein): AI features in smartphones are the real driver — think on-device inference driving chip content per phone. QCOM owns the Android flagship slot. Data center is optionality, not the thesis.
The bear case (BofA): QCOM is "re-entering a fast-growing but hyper-competitive AI market after multiple prior data center entry attempts." Competition list is a horror show: NVDA, AVGO, AMD, MRVL, Cerebras, ARM, plus custom chips from AMZN and GOOGL. BofA's fundamental call hasn't changed — they just updated the modeling.
"Qualcomm is re-entering a fast-growing but hyper-competitive AI market after multiple prior data center entry attempts." — BofA
Also in the mix: Tenstorrent acquisition chatter ($8-10B price tag) and a smart eyewear partnership with Inspecs. M&A headlines give momentum traders a narrative, but the core debate is whether QCOM can actually execute in DC this time. Wednesday's deck is the swing factor.
Piper Sandler upgrades to Overweight (PT to $550 from $450) — the catalyst is Intel recovering faster than expected. The call is pure positioning shift: if Intel’s foundry actually works (Apple deal, Google TPU win), SNPS’s IP business gets a multi-year tailwind after a painful reset nine months ago. Not buying the EDA story here — buying the Intel recovery derivative.
“Nine months after a difficult reset to IP numbers, Intel’s 18A-P node now presents a very different demand environment.”
Q2 print was clean (rev $2.28B, EPS $3.35, OP margin 39.5%) and guidance raised. Stifel, KeyBanc, Rosenblatt all raised PTs to $575-600 after the quarter, but Piper’s upgrade is the first explicit bull case tied to Intel’s renewed capex cycle. Consensus already pricing 24% upside to current $464 — feels like momentum PMs will chase if Intel’s 220% QTD rally holds. r/r tilted long, but this is a high-beta play on Trump-era domestic chip policy execution, not a fundamentals grind.
Verdict: Show-me story trading at 52-week lows. UBS is neutral with a $32 PT (43% upside from $22.42), but the narrative is all about H2 improvement that markets don't believe yet. Broadband erosion is real, parks just got softer, and the stock has zero momentum. The bull case rests on political ad ramp, NBA lapping, and new pricing/packaging in connectivity - but that's all back-half weighted. Q2 estimates just got trimmed (revenue +1.6% vs prior +2.0%, EBITDA -6.6% vs -6.3% on Versant-adjusted basis). Not a crisis, but enough to keep the stock pinned.
UBS sees "gradual improvement in the second half as new pricing and packaging at connectivity and the NBA deal is lapped while political advertising ramps." That's the entire thesis in a nutshell. The problem: connectivity EBITDA declines are accelerating in Q2, parks is now a headwind (Orlando/Japan attendance), and the UK theme park investment (£5B, opens 2031) is a long-duration capex cycle that adds zero near-term visibility.
"UBS updated second-quarter estimates to reflect softer parks performance and now looks for total company revenue to grow 1.6% and EBITDA to decline 6.6% in the second quarter adjusted for Versant, compared to prior estimates of 2.0% revenue growth and 6.3% EBITDA decline."
Rosenblatt cut to $24 (Neutral) this week - that's the only other analyst action noted. So you've got a $24-$32 PT range with the stock at $22. Broadband churn and competitive losses are the bear's favorite punch line. Wolfe's subscriber work just reinforced that telcos are eating CMCSA's lunch on fixed wireless. No one's pounding the table. The stock needs a catalyst - either Q2 print surprises positively (unlikely based on revisions) or broadband stabilizes faster than models imply. Right now, it's a waiting game. Cheap on P/E, but cheap for a reason.
KeyBanc is getting comfortable with the expense picture. They reiterated Overweight and $300 PT, raising EPS estimates for FY28-30 (now above consensus for FY29/30). The thesis: OpEx growth moderation is sufficient to offset gross margin pressure from AI hyperscaler infrastructure spend. Gross margins are 65.82%, PEG ratio 0.86 — attractive r/r if cost clarity holds.
"Operating expense growth moderation appears sufficient to offset gross margin pressure."
The broader analyst consensus is still constructive — Mizuho ($320), BMO ($220), TD Cowen ($300) all out with positive notes post the Q4 print (IaaS revenue +90% YoY). No one is ringing the alarm on the AI capex drain. The bet here is that the expense trajectory is now de-risked, and the multiple (PEG < 1) has room to re-rate if growth sustains. Not a screaming buy, but a credible risk/reward if you believe the margin narrative sticks.
Ad stack upgrades are printing. B.Riley reiterates Buy / $250, pointing to Cannes Lions disclosures showing advertiser ROAS above industry averages in key verticals. Revenue surging 71% LTM to $2.47B with a 91% gross margin — the monetization flywheel is real. User growth also tracking strong: DA Davidson flagged 20.3% Q2 DAU YoY (beat consensus 18%), and Citizens noted app installs up 129% YoY with lower acquisition costs. BofA remains Neutral / $190, arguing the AI-training data narrative is priced in. Near-term risk/reward skews attractive at 15.9x FY27 EV/EBITDA.
“Advertiser return on ad spend is above industry averages in key categories” — B.Riley
That 50% of U.S. shoppers say they verify AI recommendations on Reddit? That’s a structural moat for both ad spend and data licensing. Meta launching a Reddit clone (“Forum”) is noise — not a near-term threat to the niche community flywheel.
Cantor goes long on solid-state — initiate Overweight at $18. That’s ~29% upside from $14, and the call is all about breaking the energy density ceiling. They’re framing FAC as the pure-play gravimetric energy density bet for drones, defense, mobility — the stuff that actually needs lighter, denser power. National security tailwind to onshore battery supply chains is the macro hook.
“Factorial is one of the few solid-state developers that has placed automotive-format cells into real vehicles with global automakers.”
Key differentiator: OEM validation (Stellantis Dodge Charger Daytona road test already live). Capital-light model, three-to-five year growth cycle projection. Stock is volatile — InvestingPro flags high vol but liquid assets > short-term obligations. Not a name you ignore if you want exposure to the next gen battery narrative.
TD Cowen is sticking to its guns. Reiterated Buy and $38 PT on a stock that's now trading at $19.79 — roughly HALF its 52-week high. That's a near-100% implied return from a house that picked PINS as their Best Smidcap Idea for 2026. The thesis isn't new, but it's getting sharper.
The core story is Performance+ ramping adoption as the lower-funnel growth catalyst. Pair that with a revamped go-to-market strategy and better measurement/optimization for advertisers, and the monetization story starts to look less like hope and more like a trajectory. Off-platform bets via tvScientific are a bonus kicker.
"The ramping adoption of Performance+ as a key factor expected to drive lower funnel advertising revenue."
Supporting data from the Q1 print helps. Revenue of ~$1.008B (+18% YoY) cleared consensus of $965M. UBS raised their PT to $30, Guggenheim stuck at $24 with a Buy — all pointing to the same narrative: ten consecutive quarters of double-digit user growth, AI-driven ad improvements, and a U.S. business that's firing. Benchmark's $33 target reinforces the cluster.
Risk? The stock has been cut in half for a reason. Growth is decelerating. Competition for ad dollars is fierce (Meta, TikTok). But at 19.79 with a 38 PT and a bull case that's actually gaining operational evidence, the r/r is asymmetric for those willing to look past the tape noise.
Don't overthink the executive departures. Jefferies is sticking with Buy and $445 PT on GOOGL despite the recent weakness. They frame the pullback as a three-part story: executive exits (Noam Shazeer to OpenAI being the latest), Mag7 rotation ahead of frontier-lab listings, and multiple normalization. The talent bleeding is real — but they argue Alphabet's AI heritage and bench depth mean it can absorb it.
"The talent movement dynamic will continue due to scarcity in AI, but Alphabet's massive distribution network, accelerating Cloud growth, and vertically integrated TPU advantage justify the rating."
Options market agrees — 928K contracts traded yesterday with call volume more than 2x puts. That's not retail noise. Meanwhile, the $75M A24 investment is a $ sign Google is serious about AI content partnerships.
Bear case: Multiple compressing from premium, Shazeer is a high-profile defection, and Mag7 selling could get uglier before frontier-lab IPOs clear. Bull case: PEG of 0.6 (17.5% rev growth, 20x NTM EBITDA) is cheap for a Cloud-accelerating AI play with a deep bench. Jefferies' $445 is 20x EBITDA — not heroic. Citizens went $515. Range from $340-$515 tells you the street is split on timing, not thesis.
The exits have headline risk, but the underlying rate of change (Cloud +28%?, TPU ramp, search share stable) hasn't budged. PMs should watch the rotation bids — when fronti-er-lab floats, Mag7 could re-rate higher as the AI narrative gets a liquidity event. GOOGL gets the bid first.
UBS doubling down on the AI narrative, just not the pass-through kind. They reiterate Buy, $370 PT. Databricks disclosed at its investor briefing that pass-through AI model/compute resale is accelerating revenue to ~80%. Snowflake told UBS its own pass-through AI revs are immaterial. UBS calls that a net positive — means the core consumption engine is driving mid-high 30% growth without relying on low-margin pass-through. The AI lift is still early-stage.
Blockquote from UBS:
"Snowflake’s multiples of 14x and 59x on CY26 estimated revenues and free cash flow are defensible given mid-high 30% growth."
Other buyside nods: Truist raised to $300, Benchmark to $290 after Snowflake Summit. Both cited strong AI momentum and raised FY27 product revenue guidance. Databricks delaying its IPO (citing unfavorable market conditions) removes a near-term overhang — no competing public AI data story to steal attention. Snowflake at $232 / ~$81B market cap with 31% LTM revenue growth and a $370 PT implies 60% upside. PMs should watch the next product cycle — AI consumption is coming, just not through the resale channel yet.
Hearing the Korean circuit breaker was triggered by a single article suggesting SK Hynix is slowing HBM4 expansion. Leveraged 3x ETF rebalancing amplified the move — dealers had to sell into a falling market. The actual Hynix check suggests HBM production is accelerating, not slowing. Suspect the article was mistranslated or misinterpreted.
Word is the CXMT IPO valuation debate is heating up. Some desks pitching $55B revenue this year, $100B+ by 2027, with only 5-10% discount to incumbents. The bull case: China wants margin and scale, not dumping. The bear case: that's still 17% global DRAM capacity addition that eventually crushes pricing.
Channel checks suggest TSMC's 5-10% price hike is being accepted by hyperscalers without pushback. One ODM contact said "they have no choice — there's nowhere else to go for 3nm." Confirms inelastic demand for advanced compute.
Hearing Blackstone's $30B Japan data center investment is specifically targeting decommissioned nuclear plant sites for grid-scale power. This bypasses the years-long Western interconnection queue. Japan is emerging as a dark horse in AI infrastructure.
Word is Alibaba's T-Head silicon entity filing to increase registered capital 3x+ and restructuring under PingtouGe is a precursor to an IPO. They're ramping PPU (Processing-in-Memory) and data center ASIC efforts aggressively.
Channel checks suggest Vertiv's liquid cooling orders are transitioning from "evaluation" to "required spec" for hyperscalers. One DC operator said "if you're not spec'ing liquid cooling for new builds, you're building obsolete infrastructure."
Hearing the optical module pricing rumor (NVDA forcing $920 cap) was aggressively short-seller driven. NVDA IR denial was unambiguous — annual framework orders executing at normal commercial prices. Look for networking supply chain names to recover.
Word is some hedge funds are building short positions in META on the Arena distraction. Thesis: Zuck has no capital discipline, capex is a black hole, and the latest app bets are FOMO moves. The counter: AI-driven ad revenue acceleration is real and Llama open-source models give Meta a platform advantage.
Channel checks suggest enterprise server renaissance is real — traditional IT departments are refreshing infrastructure to handle AI inference workloads. Legacy enterprise hardware demand is turning positive for the first time in years.
Hearing JNTC (Korean glass substrate play) is trading at ~9x 2027 EPS despite being the key supplier for Intel's glass substrate program. Distribution-only opportunity if the CXMT IPO draws attention back to Korean semicap names.
Word is the Dell AI server margin compression story is still in play — one ODM said "the GPU is 70% of BOM cost and Dell has zero pricing power on that component." Bull case relies on Services and Storage attach rates improving.