Good morning.
IBM -25% ERASED $69B IN MARKET VALUE. Biggest single-day drop since the dot-com bust. The preannounce is a structural signal — customers shifted spend from mainframe and related software to AI initiatives late in the quarter. This is not a one-off. SaaS selloff is ripping through ACN, CTSH, GLOB, but we think the punishment is overdone for quality names like NOW. Meanwhile, SOXX +2-3% — semis rallying on the read-through that AI infra spend is real, just reallocating. GS AT ALL-TIME HIGH, +3.2%, global banking & markets rev +53% y/y — the financing engine is humming. MU +8%, SKHY AT ALL-TIME HIGHS as Rubin memory BOM triples to ~25% and cooling inflation supports the bull case. ORCL -58% from highs, credit spreads blown out 136bp — complete market rejection, but contrarian interest piqued sub-$100.
Asia mixed overnight. DeepSeek pre-IPO targeting $71B valuation, raising another RMB 10B+ — Chinese AI funding machine spinning up. Power grid short 6.8GW on DC demand — infrastructure bottleneck becoming the binding constraint.
Three themes framing the tape today:
First, IBM IS A SPEEDBUMP, NOT A CRATER. The SaaS indiscriminate selloff creates opportunity. The rate of change matters more than the level — AI budget share is expanding, it's just coming from legacy. Goldman notes momentum response muted on both legs, suggesting risk appetite still finding footing. We'd be buyers of the SaaS hole on quality (NOW, WDAY).
Second, MEMORY IS HAVING A MOMENT. SK Hynix ATH, MU ramping, SNDK getting pulled along. The structural bull case: Rubin server memory BOM tripling, cooling inflation, MATCH Act tailwind. Don't fade this — it's a positioning shift from DRAM as a commodity to AI memory as a bottleneck.
Third, THE GREAT FINANCING DIVERGENCE. GS printing record IB revs, JPM seeing $700B+ AI capex. But credit markets rejecting neocloud buildout — default swaps widening. Lower rates help the financing trade, but the disconnect between equity buyers (hardware) and credit sellers (powered shells) is real. Watch for an air pocket if power constraints tighten.
Will hit up IBM, MU, and GS first, then get to semis and the SaaS dislocation.
SPCX at $138-139, down 9.4% in the past week and scraping its 52-week low. That’s absurd for a company that prints 106% revenue CAGR and where the bull case is “monopoly on orbital access + Starlink = $3.3 TRILLION by 2040.” But the market is pricing in Starship execution risk. Flight 13 is Thursday. If it works, this thing re-rates hard.
PTs span $190 (Stifel) to $800 (Raymond James), with a cluster at $230-300 (Evercore at $230, Bernstein at $239, Morgan Stanley at $300). That’s a 65% to 475% upside from current levels. Not a single Sell or Hold in the group. The bull case is unanimous on the long-term story; the debate is timing and Starship derisking.
"an extraordinary company on a real path to reshaping the future of humanity... a vertically integrated near-monopoly on orbital access through reusable, low-cost launch technology." — Evercore ISI (Kutgun Maral)
"The July 16 date implies a move from 221 days between Flight 11 and Flight 12 to a projected 55 days between Flight 12 and Flight 13. That represents a roughly four-times compression in the flight-to-flight cycle." — Raymond James (on Flight 13 as the critical path to the investment thesis)
SPCX is the bellwether for the space thematic. Every successful Starship flight de-risks the entire ecosystem — Starlink, orbital compute, even terrestrial AI infrastructure. If Flight 13 works, expect read-through to:
Verdict: Preannouncement was a shitshow – revenue miss, software weakness, guide cut incoming. But the street is split on whether this is a buying opportunity or a value trap. The bull case rests on Red Hat acceleration and quantum optionality; the bear case says the multiple is delusional for mid-single-digit growth. Stock's at $220 – near 52wk lows. This one's going to trade on the full report next week.
Range is $191 (HSBC) to $350 (Oppenheimer) – that's a 45% spread. Consensus is... not. BofA maintained Buy but cut to $280. Evercore kept Outperform at $310. Oppenheimer raised to $350. HSBC downgraded to Reduce, PT $191. Net net: the bulls are leaning on Red Hat, quantum, and AI bookings to justify the premium; the bears see a miss on the only thing that matters (software growth) and a stock that's pricing in a recovery that isn't here yet.
Bull (Oppenheimer, Evercore): The miss is a timing issue – clients are consuming hardware now, software later. Red Hat accelerating, AI bookings (Watsonx >$12.5B) are real. Quantum valued at $65/sh by Susquehanna. ELA renewals showing 20-30% uplift, normalizing. At 0.27 PEG, the stock isn't pricing in the reacceleration. Oppenheimer's checks show no customer attrition and steady demand for Red Hat, WebSphere, DB2, HashiCorp. They expect the guide to go UP.
"Demand for RedHat, Websphere, DB2, and HashiCorp products remains steady. Non-mainframe infrastructure continues to perform strongly." – Oppenheimer
Bear (HSBC, BofA cautious): The software miss is structural, not timing. HSBC flags the 22x CY27 P/E vs sector median 16.9x – you're paying a 30% premium for sub-sector median EPS growth (10.7% CAGR vs 19.2%). BofA notes it's unclear when the capex spending issues affecting software will resolve. Infrastructure declining mid-single digits. The entire "hybrid cloud / AI" narrative requires software to be the growth engine – it's not delivering. If free cash flow misses the $1B YoY target, the bull case breaks.
"IBM is trading at 22.0x CY27 estimated non-GAAP P/E, compared to a sector median of 16.9x. EPS CAGR of 10.7% over CY26-28, below the sector median of 19.2%." – HSBC
"IBM reported preliminary second-quarter revenue of $17.2 billion, missing the Street estimate of $17.9 billion... software and infrastructure divisions missed estimates significantly." – BofA
"HSBC downgraded IBM to Reduce from Hold... cited stretched valuation... new PT of $191 implies 33.6% downside from current levels." – HSBC
"The company attributed the shortfall to weaker-than-expected mainframe performance across infrastructure z systems and software Transaction Processing." – Evercore ISI
This is a "budget rotation" signal, not a demand-collapse signal. Enterprise IT wallets are shifting away from legacy mainframe/transaction processing toward AI infrastructure (servers, memory, storage) and cybersecurity. IBM's miss is a macro read on where the money is going – not that it's drying up.
Verdict: Apple is a two-front war. The OpenAI lawsuit is noise for the stock – it threatens OpenAI's hardware ambitions, not Apple's. The real battle is KeyBanc’s bearish iPhone thesis vs. Monness/Evercore’s product-cycle optimism. Net net: consensus is split, but the stock is near highs, and the market is pricing in the bull case. The next catalyst is July 30 earnings (Cook’s last call). Watch Services growth and iPhone build rates – those determine the next 10% move.
Wide range of PTs: $250 (KeyBanc) to $365 (Evercore). Consensus sits around $320-335. The divergence is about iPhone unit elasticity and Services deceleration – not about the OpenAI dustup.
The OpenAI lawsuit (filed Friday) is the fresh catalyst. Evercore is right to call it a potential overhang for OpenAI’s consumer hardware, not Apple. The key allegation: former VP Tang Tan brought Apple parts, codenames, and directed current Apple employees to bring batteries, logic boards, and back glass to OpenAI interviews. OpenAI now employs 400+ ex-Apple people. If Apple gets injunctive relief or forces discovery, it could delay OpenAI’s hardware launch (reported for 2027). Zero read-through to Apple fundamentals – but it keeps the narrative of Apple as the platform gatekeeper alive.
KeyBanc’s downgrade is incremental bearish evidence. Their proprietary spending data shows monthly indexed spending down -2% MoM vs a 3-year avg of +9%. That’s a sharp deceleration. They also note slowing iPhone builds with price increases – the classic squeeze on unit volumes.
Monness’ upgrade to Buy (from Hold?) is the flip side: they see Q3 revenue of $109.4B, above street, and note sequentially Q3 decline of only -2% vs -9% historical average. That’s less bad seasonality.
Bull case (steelman): Apple’s installed base is sticky, Services is a hidden growth engine (still 15%+ YoY per Monness). The Broadcom custom silicon deal ($30B+) locks in chip supply and margin. Sell-side models don’t capture the replacement cycle from AI-on-device features (iPhone 18 gen). At 35x trailing EPS, it’s expensive but not silly when you consider the annuity-like Services revenue. Monness: “We expect Apple to meet Q3 estimates; the comps get easier in H2.”
“The legal action creates a potential overhang for OpenAI’s consumer hardware ambitions, particularly if Apple pursues injunctive relief or forces discovery around OpenAI’s product development process.” — Evercore ISI
Bear case (steelman): iPhone unit growth is done. U.S. carrier subsidies are being cut – that’s a structural headwind. KeyBanc’s data shows spending deceleration. If Services growth falls to 7%, the bull case premium multiple collapses. KeyBanc: “Slowing unit growth will pressure user base expansion, impacting Services revenue.” The stock trades at a 2-standard deviation premium to the S&P 500 and Nasdaq – that’s a crowded long. And the Broadcom deal? Great for margins but doesn’t drive revenue.
“We view consensus iPhone growth expectations of 8% in fiscal 2027 as too aggressive, particularly as U.S. carriers pull back on device subsidies.” — KeyBanc
Apple is a proxy for premium consumer hardware + AI tailwinds. The OpenAI lawsuit is a reminder that Apple is the talent magnet and IP fortress in hardware – read-through to GOOGL (Android ecosystem) and MSFT (OpenAI partner) as potential collateral targets. But the immediate signal is about iPhone demand – check QCOM and SWKS for component orders. If KeyBanc is right about builds, the semi supply chain gets hit. If Monness is right, the iPhone cycle is still accelerating.
Bottom line: Positioning is split – open interest on AAPL options is heavy at $320-330 strikes. The earnings print on July 30 is the binary. Services growth and iPhone revenue guide will decide if the $250 target or the $365 target is closer. Right now, the stock is betting on the bull.
VERDICT: AI tailwind intact, but stock priced for perfection. TD Cowen bumps PT to $175 (from $135) — a 30% hike — but sticks at Hold. That's the crux: everyone loves the story, no one willing to underwrite more multiple expansion from here.
TD Cowen was the solo mover yesterday (July 13) raising to $175, but they're not alone in the bull camp. The broader analyst cluster has coalesced around $175-185 — Barclays $180 (Overweight), BofA $185 (Buy), Evercore $180 (Outperform). The collective thesis: AI infrastructure demand is a multi-year supercycle for interconnect, and APH is the purest play on that theme. Revenue up 54% LTM to $25.9B. The Q1 IT Datacom print crushed expectations (+81% organic vs Barclays' 60% estimate). Q2 guidance of $8.15B and EPS $1.15 — $0.10 above consensus — sets the table for another beat July 29.
TD Cowen specifically noted the narrative shift: "Concerns about copper have largely been replaced with debate on upside from CPU versus GPU ratios." That's the new frontier — not whether AI demand is real, but which compute architecture wins and how that flows to connector content.
"Shares are positioned favorably heading into second-quarter results, with Street estimates likely to move meaningfully higher following the report."
That's the bullish hook. But they kept the Hold because the stock already prices a lot of that optimism in.
Bull case: APH is the toll road for AI data center builds — fiber (post-CommScope acquisition), copper, power connectors. Revenue growth is accelerating, margins are sticky, and every hyperscaler capex cycle adds incremental content per rack. The Q2 guidance raise signals backlog visibility into 2H. If GPU ratios shift toward CPU-heavy inference workloads, APH's breadth of interconnect products means it captures value regardless.
Bear case: At 57% return over the past year, APH trades at ~35x forward earnings (rough math on $5.00+ EPS). That's a lot of AI optimism already in the price. Copper concerns may be "replaced" but tariffs, supply chain shifts in China/Mexico, and potential hyperscaler capex normalization in 2027 are real risks. The €1.1B senior notes offering adds leverage — manageable, but a signal they're funding growth through debt, not just cash flow.
Verdict: The narrative is starting to bifurcate. Evercore cuts the PT ($250 from $260) while Goldman holds firm ($242, Buy). Both see the same crosscurrents — Agentforce momentum and the Fin deal offsetting real pain in Tableau, Commerce, Marketing. The question for PMs: does the reacceleration story in 2H have legs, or is this a 12-24 month slog where headwinds keep winning? The tape says the latter (down 35% YTD), but the rate of change on agentic adoption is worth watching.
Evercore’s checks are “net positive” on Agentforce development but flag DATA READINESS and ROI concerns, especially in channel-influenced business. They trimmed Q3 revenue on Tableau weakness. Goldman’s Borges hosted Salesforce IR in London — takeaway: organic acceleration expected in Q3, driven by headless monetization (consumption pricing, premium bundling, agentic licenses — more at Dreamforce in Sept). Slack is getting a lift from new LLM-based features. Both agree Tableau, Commerce, and Marketing are headwinds that will persist for 12-24 months.
“Data readiness and return-on-investment concerns remain obstacles to broader adoption of the company’s products, particularly in channel-influenced business.” — Evercore ISI
Goldman’s key call: Salesforce is being realistic about the timeline. The Fin acquisition ($3.6B for Intercom) gives them one of “just three to four leading assets in Service” — the integration with Agentforce is the swing factor. If they can harmonize the tech stack and accelerate down-market adoption, this becomes a different story.
BUY. The Blackstone-led consortium deal is exactly the kind of catalyst the BTM (behind-the-meter) theme needed. WMB monetizes a 49% stake in five power projects for $5.34B – $4.4B of that is growth capex funding, plus ~$900M incremental cash to WMB. Both UBS ($91 PT) and Goldman ($82 PT) reiterate Buy. Stock at $76, up 26% YTD, still has room if the narrative holds.
WMB retains 51% ownership and operational control. Blackstone, Apollo, and insurance vehicles get non-controlling equity. Key numbers:
UBS frames it as a balance sheet unlock – WMB gets $900M upfront, plus capex relief, without losing commercial control. The 53-year dividend streak stays intact.
Bull: This is a template for how midstream can finance data center load without equity dilution. WMB keeps control, gets cheap capital, and has a buyback option that effectively gives them a free call on the projects’ upside. The 5.7x initial multiple looks low, but the buyout path flips the r/r.
Bear: You’re selling a 49% interest in your highest-growth projects at a sub-6x multiple. If these projects deliver, you just gave away half the economics. The buyout option helps, but execution risk on repurchasing – and the opportunity cost of not retaining 100% – is real.
Bottom line: WMB used a smart structure to de-risk the balance sheet while keeping the BTM strategy moving. The market wants more proof on the EBITDA ramp – Q2 call should provide it. We like the set-up.
Verdict: Show-me story at a 5-year low multiple. Stock sitting at $73.83, just above the 52-week low of $70.86, down 21% YTD. Engagement fears are the narrative du jour – Rosenblatt's data points to caution, and the market has dumped the name 41.5% over the past year. But the valuation is getting stupid: EV/EBITDA approaching mid-teens (low end of 5-year range), PEG ratio of 0.5. Earnings Thursday is the pivot point. Either management proves the engagement dip is noise, or this thing gets washed out further.
Bulls (BofA): Management has navigated skepticism before – 2022 was far uglier (-50% peak-to-trough). They pivoted to paid sharing and ad tier, reaccelerated growth. BofA sees "long runway for subscriber and ad growth" and a world-class brand. $125 target implies 70% upside from here. They're leaning into the "innovator" narrative and the live/content opportunity (World Cup bids, quality metric hitting highs).
Bears (Rosenblatt / the market): User engagement is the real bogey, not just sub growth. Rosenblatt says the data they track "suggests reasons for caution." Multiple firms flagged engagement concerns – Morgan Stanley cut to $90, KeyBanc to $92. The AI disruption to content creation and competitive M&A (Warner Bros stuff) add structural overhang. At $73, the stock is pricing in a lot of bad news – but it could get worse before it gets better.
BofA: "The management team has consistently demonstrated an ability to adapt to changing market conditions, execute effectively and create long-term shareholder value."
That's the bull case in a nutshell. They've earned the benefit of the doubt once before. The question is whether this time is different (engagement decay is structural, not cyclical) or whether the market is overreacting to a period of normalizing growth after the post-pandemic pull-forward.
Two camps. BofA holds the bull flag at $125 Buy. Rosenblatt sits neutral at $95. In the background, Evercore is Outperform $115, while Morgan Stanley and KeyBanc both cut to $90-92 (still Overweight). Citizens is Market Perform. So the range is $90-$125 – wide dispersion for a name this liquid. Consensus is not clean. The bulls think the multiple is a gift; the bears think engagement is the next sub-prime.
Key number to watch Thursday: that "internal quality metric" management has been teasing. Rosenblatt expects discussion. If it's strong, the narrative flips. If it's weak, the stock tests $70.
This is a high-r/r, binary event trade into the print. The stock is pricing a worst-case scenario. If engagement stabilizes or the quality metric surprises, this thing rips. If it's horrid, the $70 floor breaks and you get another leg down. Not a position to size heavy, but worth a small tactical long if you buy the "management has a track record of adaptation" thesis. I'd wait for the print to confirm before adding scale.
Verdict: Stock is pricing in a re-rate that analysts aren't buying yet. QLYS at $152 trades north of every single price target on the Street – the highest being TD Cowen's new $145 (from $90, still a Hold). The rally has momentum (up ~13% in a year), but the underlying narrative is a tug-of-war between FedRAMP / AI tailwinds and a deeply competitive vulnerability management market.
The bull case leans on FedRAMP High Authorization (sponsored by the DEA) and AI partnerships (Anthropic Claude Mythos, OpenAI GPT-5.4-Cyber). JPMorgan upgraded to Neutral from Underweight, lifting PT to $139, citing "growth in the vulnerability management category." Management itself highlighted increased customer interest from those AI tie-ups. Gross margins are 83%, balance sheet clean (cash > debt). Q2 CCB guidance is +9% y/y, in line, and EPS ests at $1.91 vs consensus $1.79 on lower opex growth.
The bear case is competitive and structural. TD Cowen:
"We remain cautious about the level of competitive rivalry in the vulnerability management space. The emergence of new AI tools adds uncertainty about the potential size of the vulnerability management total addressable market."
That's the key tension – AI could be a feature, not a moat. Meanwhile, Scotiabank cut its PT to $100 (from $135) after 8% current billings growth in Q1, a deceleration signal. Even RBC's $90 PT (raised from $85) is a Sector Perform – no conviction.
Bottom line: QLYS is pricing in a beat-and-raise cycle, but the analyst consensus is still stuck at Hold / Sector Perform. The stock needs to deliver not just in-line billings but actual acceleration to justify the premium. Risky r/r from here unless you believe AI partnerships dramatically expand TAM rather than just compress vendor margins.
Roth/MKM initiates with a Buy and a $125 PT — 54% upside from $81.35. But let's be real: this thing is already UP 131% IN SIX MONTHS. They're not early, but they're first with a formal stamp. The thesis: edge AI's biggest bottleneck isn't compute — it's power. Ambiq's ultra-low-power semiconductor tech solves the problem of running complex models on battery-constrained devices (think wearables, hearables, IoT sensors). That's the wedge.
"The technology addresses power consumption challenges in environments where battery availability is most constrained."
Recent noise: closed a $179.4M public offering at $78 (2.3M shares, fully exercised greenshoe). Northland also bumped PT to $86 (Outperform) citing new design wins in wearables. So you've got two firms now — both bullish, albeit with wide PT dispersion ($86 vs $125). The difference? Roth/MKM is pricing in the edge AI secular story; Northland is more near-term on execution.
Risk worth highlighting: the stock has already repriced aggressively. Post-offering dilution is absorbed, but the float just got bigger. If growth doesn't accelerate from here, $125 feels aspirational. But the narrative is clean — and in this market, clean narratives get premium multiples. Worth a look as a high-beta winner in a crowded "AI everywhere" trade.
Cantor is doubling down. Reiterated Overweight and $450 PT on a stock that’s already up 111% in the last year (+43% YTD). They’re not just chasing momentum – they see accelerating EPS power: $10 this year (vs $9.39 consensus) and $20 by 2028 on a 25% semi CAGR + 30% data center CAGR. The take: management sandbagged Q1 guidance, and both segments have material upside.
“Management set conservative guidance last quarter, with upside likely in both business segments.”
This isn’t just a semi-equipment story anymore. The ADH series DC-DC converters (98.2% peak efficiency) are live for AI data centers, and the $1.15B convertible note raise gives balance sheet flexibility to retire the 2028 converts. Stifel also reiterated Buy at $385, but Cantor is the bull flag – $450 implies 46% upside from here. Risk? The 0% coupon notes imply dilution if the stock stays strong, but the optionality on 2028 EPS is the real driver.
Cantor stays Neutral at $175 – orders are unprecedented, but the market already knew. Stock down 33% from highs (group avg -21%), now at 30x 2027 EPS of $4.50-5.00. That’s not cheap enough to ignore the competitive noise from Onto and KLA picking up in 2H.
Order momentum is real – $105M in recent wins, HBM and OSAT-driven – but Cantor flags that most of this strength was already baked into prior commentary. Meanwhile peers are guiding >50% advanced packaging growth, narrowing the relative edge.
“Increasing competitive risks into the second half from Onto Innovation and KLA Corporation” – Cantor, on why they stay sidelined.
For context: Evercore and Jefferies both raised PTs to $200; Stifel downgraded to Hold $185 on valuation. Divergent views, but the common thread is that CAMT’s booking take is strong enough to support the story through 2027 – just not enough to escape the multiple compression the group is feeling.
EUV is the bottleneck and ASML is the only supplier. Multiple shops boosting PTs today – RBC to $2,000 (from $1,700), Bernstein to $2,623, BofA to $2,268, Morgan Stanley to €1,660. Collective thesis: GenAI capex, DRAM tightness, and leading-edge foundry competition all driving EUV demand. RBC now models 2027 shipments closer to 90 low-NA units vs prior >80. Customers at record profitability => pricing power (unit + mix tailwinds). Near-term upside is modest given long lead times, but medium-term visibility is clearly improving.
"EUV is emerging as a bottleneck and supply tightness is not likely to ease for the next two to three years." – RBC Capital
Terafab (Musk's project) adds optionality on an already strong base. Stock up 115% in the past year, sitting at $1,726 – still has room if the EUV shipment ramp materializes as modeled.
Split-adjusted targets hitting the tape. Rosenblatt cuts PT to $206 (from $825 pre-split) — pure arithmetic, no thesis change. Still Buy. Rest of the Street mostly matching the math: Stifel to $220, Barclays $169, Morgan Stanley $172. UBS actually raised to $235 on AI growth narrative after management meetings. Stock +58% YoY, trading ~$188. Low drama. High conviction.
"The price target adjustment reflects only the arithmetic of the split with no change to its rating, dollar-based estimates, valuation methodology, or investment thesis."
UBS is the outlier — that $235 is a vote of confidence in the AI security pipeline (Falcon + data protection). Everyone else just adjusted the share count. The takeaway? No one’s walking away from the thesis. CRWD remains a core holding for the SaaS security platform bet. The split jitters are over.
Verdict: This is a $10B+ market cap stock that’s up 207% in a year and still gets upgrades. Cantor, Oppenheimer, Morgan Stanley, Deutsche, Stifel — all bullish. The thesis is a one-two punch: Dragonfly G5 ($1B TAM) and Iris G2 ($750M+ TAM) share gains in advanced node AND advanced packaging. Cantor sees scenario EPS power of $12 in 2027 – $15 in 2028. At 141x trailing, multiple is horrid, but the rate of change on earnings matters more here. This is a product-cycle compounder, not a value trap. The Rigaku deal close and convertible interest income add EPS upside that most models haven’t fully baked.
“Cantor Fitzgerald highlighted increasing confidence in share gain opportunities across multiple products, including Dragonfly G5 in front-end applications with a $1 billion total addressable market and Iris G2 in critical films with a $750 million-plus total addressable market.” Scenario EPS upside to $12/$15 in 2027/2028.
Not a single analyst voice dissenting. Oppenheimer bumped to $450, Morgan Stanley started at $371, DB at $350. The only question is whether the Street can price the 2027-2028 ramp without sliding into discount rate panic. For now, the flow supports it. Keep it on the radar for any pullback — rate of change here is better than most semi equipment names.
KEYBANC GOES BOLD: $155 PT FROM $110. That’s a 50%+ upside to current $103. The call is built on agentic AI driving server CPU demand (25-30% unit growth this year, >50% next) plus foundry momentum that is actually showing up in numbers.
The headline grabber is 18A yields hitting >85% from 65% last quarter — that’s a real rate of change. KeyBanc also cites design wins with Apple, AMD, Nvidia, Marvell, MSFT, Micron, OpenAI. Second major EMIB-T win with AWS Trainium 3 (alongside Google TPU). They’re modeling $10.6B foundry rev and $22B EMIB-T by 2030.
"Intel has secured its second major design win on EMIB-T with AWS Trainium 3, in addition to Google's TPU Humu Fish."
Pricing power is back: client CPU price increases of 6-15% starting Q3 2026. They’re not just selling more — they’re capturing more per unit. Stifel is more cautious (Hold, $120) but KeyBanc is swinging for the fences on a 20x 2030 EPS of $7.58.
The bull case is a foundry narrative that is no longer theoretical. The risk: INTC still isn’t profitable on a TTM basis, but analysts see that flipping this year. If 18A ramp stays on trajectory and EMIB-T becomes a real revenue line, the re-rating has legs.
BULLISH SIGNAL FROM SOVEREIGNTY PLAY. Cantor sticks to Overweight / $95 PT after Rubrik committed $500M to UK expansion, naming London EMEA HQ. This pairs local infrastructure with regional leadership — sovereignty preferences are real, pulling budget toward vendors with onshore offerings. Revenue up 46% LTM to $1.42B, gross margins 81%, and 2,000 EMEA customers with nearly half using 3+ products. New logos include Harbour Energy, Manchester City Council, Scottish Government.
“Sovereignty preferences are pulling budget toward vendors with local infrastructure.” — Cantor Fitzgerald
Other analysts are also leaning in: BMO to $98, Baird to $110 (still Bullish Fresh Pick), Truist at $90. The collective read: RBRK is gaining traction in a secular growth vertical (backup/resilience) where EU regulatory tailwinds drive stickier, higher-value contracts. Not priced for perfection at ~$90, but the r/r skews positive if the narrative accelerates.
One watch: still unprofitable on GAAP basis, and the Fair Value gap noted in the source (~0.38x relative). But the rate of change in ARR (32% sub-ARR growth, $103M net new ARR vs $85.5M expected) is the metric that matters for a growth book.
UBS lifts PT to $115 from $100, but keeps it at Neutral. Demand stable in Q2 checks — no material improvement, no deterioration. The AI conversation isn't getting any better with customers/partners. They see a modest guidance beat (consistent with recent quarters), but not enough to re-rate a 62x P/E stock that's down 27% YTD.
"The firm expects only modest upside to ServiceNow’s guidance, consistent with recent quarters."
The contrast with the rest of the Street is stark. Evercore stays Outperform with a $150 PT, Bernstein Outperform, Truist raised to $130 citing strong enterprise AI adoption (Anthropic’s Claude Code specifically). UBS is the outlier — they're not hearing what Truist is hearing on AI. That divergence matters heading into the July 22 print.
Bottom line for PMs: UBS is right to be cautious on the re-rate story, but wrong to ignore the comps tailwind in H2. The stock is pricing in zero AI benefit. If Q2 is clean and guidance is ho-hum, NOW bounces but doesn't break out. The real r/r is binary: either AI adoption is real (Truist camp) and the stock doubles, or it's flattish (UBS camp) and you're paying 62x for 15% growth. I lean toward the former but need to see Q2 cRPO acceleration first. Light position, wait for the print.
KeyBanc bumps PT to $330 from $310, Overweight maintained. The call is a net positive on demand but honest about a Vera Rubin delay — thermal heat lid issues and HBM4 qualification slippage mean a more muted Rubin ramp. KeyBanc sees minimal earnings risk because Nvidia just ships more B300s to fill the gap. The real signal is CoWoS supply getting revised up 69% to 1.1M interposers in 2027 — that's a demand data point, not a supply chain note.
"The ramp of Vera Rubin appears slightly delayed, with the Rubin ramp more muted due to thermal heat lid issues and SK Hynix qualification delays on HBM4 memory... KeyBanc sees minimal risk to estimates as it anticipates Nvidia will ship more B300 GPUs in place of R200 chips."
Other brokers still in the bull camp (MS $288, Cowen $275 Top Pick, Stifel flagging supply constraints not demand). But the noise overlay is messy — Meta delays AI agents, Burry shorting AI infra stocks. None of that changes the NVDA tape today: demand is real, supply is the bottleneck, and the stock is already pricing the $5T+ narrative. The Vera Rubin hiccup is a 2027 story, not a Q3 miss.
Stephens upgrades to Overweight, PT to $400. The call is straightforward: WCC pulled back ~10% from summer highs, and the data center thesis is getting stronger by the day. The firm expects roughly HALF of total sales to come from data centers and utilities by 2027 — that's a massive mix shift that makes the mid-to-high single-digit revenue growth target look increasingly achievable.
"roughly half of Wesco’s total sales is expected to come from data centers and utilities by 2027"
Recent 11% revenue growth supports the story. DA Davidson also initiated with Buy and a $440 PT last week, so there's a chorus forming. The $136M Newark Engineering acquisition (data center cooling, Southeast Asia) is small but strategically aligned — $60M of revenue in 2025. WCC trades at 23.4x P/E, which isn't cheap, but the growth trajectory (36% YTD, 68% over the past year) keeps the momentum crowd engaged. The pullback created a better entry for PMs who missed the first leg.
KeyBanc throws in the towel on the QRVO merger thesis. Downgraded to Sector Weight from Overweight. The original bet was that Skyworks buying Qorvo would unlock $500M+ in cost synergies and strengthen their Apple socket position. That logic is dead.
Smartphone market is getting torched. Cuts across Apple, Samsung, and Chinese OEMs. Memory shortages — severe, persistent — are the culprit. KeyBanc sees memory makers slashing smartphone allocation by 20% next year to feed data center demand. This is a structural headwind for a company where handsets are >60% of revenue. They expect the smartphone market to contract again in 2027.
"It will be difficult for the stock to work in a contracting market despite potentially compelling risk-reward and the stock trading at 8 times consolidated earnings per share inclusive of the merger." — John Vinh, KeyBanc
Stock sits at $58.24, down 18% YoY. 8x consolidated EPS (pro forma for the merger). Minimal catalysts over the next 12 months. The one bright spot: 4.9% dividend yield, 12 consecutive years of hikes. That’s a hold-your-nose floor, not a thesis.
Other noise: RBC Capital raised its PT to $80 (Sector Perform) — basically saying the business isn’t imploding. Options activity spiked to 32,697 contracts (highest since Feb 2025), mostly calls. Could be positioning for a bounce or hedging the merger spread. Hard to read that as conviction given the fundamental headwinds.
Verdict: Stay away until smartphone demand shows a pulse. The merger math might work eventually, but memory-driven contraction is a multi-year overhang. 8x EPS and a 5% yield are traps if earnings keep declining.
Verdict: Still a show-me story but Citizens plant a flag at $15. Stock at $7.69 after a 44% YTD drubbing (Russell 3000 +10%). One of the cheapies left in software, but cheap for a reason – negative FCF, decel risk, and insider selling spooking the tape. Citizens met with CEO Rogers, CFO Megji, and IR Leung at HQ and came back reaffirming Market Outperform / $15 target (93% upside from here). That’s a loud outlier vs. UBS at $8 Neutral.
The thesis: gross margins of 88.5%, more cash than debt, and five upward EPS revisions in the rearview. FedRAMP Moderate authorization for Asana Gov opens a federal vertical door. But the market isn't buying it with the stock hugging $7.5 – the same range it traded in before the FedRAMP news. Rate of change on growth is the core tension: 9% CC revenue growth in the latest quarter is not bad, but it’s not accelerating, and the company isn’t profitable on a GAAP basis.
"Met with CEO Dan Rogers, CFO Aziz Megji, and Head of Investor Relations Eva Leung. Maintained Market Outperform and $15 price target."
The insider selling from Rogers and Megji is the ugly asterisk. Hard to get PMs excited about a deep value software name when management is cashing out at these levels. That said, if they convert the FedRAMP pipeline into actual traction, this could be the turnaround setup in a sector that hates everything below $10B market cap right now. Keep on watchlist – not a position today unless you have the patience for a multi-quarter re-rate.
BUY the transformation story, not the multiple. BELFB got three fresh initiations — BofA (Buy, $330), Citi (Buy, $325), JPM (Overweight, $370) — all converging on the same thesis: the company isn't the cyclical connector maker you remember. It's a structural compounder riding aerospace/defense, space, and data center power.
BofA's PT based on 28x CY27 EV/FCF, which telegraphs confidence that margin expansion is durable, not a one-time comp effect. The $330 implies ~25% upside from $263, but JPM's $370 suggests the bull case has further to run if the AI data infrastructure optionality crystallizes.
"Bel Fuse has transformed into a higher-quality electronic components supplier with stronger profitability... growing exposure to more resilient and structurally attractive markets supports a valuation premium to its historical range."
Key points:
BofA cuts target to SEK77 (from SEK88), stays Underperform. Q2 revenue miss + Q3 guide below street. EBITA actually beat, but the headline EPS was a disaster — $0.126 vs $1.19 consensus (yes, a ten-bagger miss — smells like a one-off item or weird FX translation, but the market hung the stock accordingly). The bigger structural issue: component cost inflation (memory pricing) hitting Networks margins, and the RAN market remains dead flat.
"Rising component costs driven by elevated memory pricing present a challenge for Ericsson as the radio access network market remains flat."
BofA shaved FY26-28 revenue by 0-4% and EBITDA by 1-14%. The new target uses 6.0x FY27 EV/EBITDA — well below the current 8.6x. That multiple gap implies either they see material downside or the rest of the street is pricing in a recovery that isn't coming. Light coverage today, but this is a show-me story on cost pass-through and RAN inflection. Not there yet.
Barclays goes overweight on SKHY with a $330 PT – finally some institutional coverage on the ADR. Thesis: HBM pricing power plus US front-end manufacturing can close the valuation gap with Micron. Target multiple is 8x (discount to MU), but they see near-term gross margin upside on top of an already-impressive 68% LTM GP margins.
"Front-end manufacturing in the United States could help narrow the valuation gap between the two companies."
Barclays’ 2027 revenue sits MATERIALLY ABOVE CONSENSUS – driven by HBM pricing stickiness and SKHY’s dominant position. Reminder: this is a 85% revenue grower over the last twelve months. The ADR listing itself was a $149 offer on 177.9M ADRs (now trading +6.8% premarket after a volatile Korea session). Not sure we can read too much into that price action given the Philly Semi index was -4.8% the day prior – could be a catch-up trade or a positioning squeeze on the first real institutional stamp. Worth watching if the multiple re-rates toward Micron’s as the US narrative builds.
Clear Street initiates Buy, $20 PT (54% upside from $12.99). They see OSS as a differentiated defense AI compute play — ruggedized infrastructure for P-8 Poseidon, expanding into Army vehicle programs and commercial aerospace. Revenue trajectory is the hook: from ~$32M in FY25 to nearly $67M by FY28. Adj. EBITDA goes from a loss to ~$7.4M over the same window.
"One Stop Systems is a differentiated supplier of rugged AI compute infrastructure for defense applications."
The stock's already up 156% in the past year and 81% YTD, so a lot of good news is priced in. Recent Q1 2026 EPS beat ($0.01 vs -$0.05 consensus) and an $8.4M defense contract that could scale to $44M over four years support the narrative. Russell 2000 inclusion next week adds index demand. Platform Fair Value flags the stock as overvalued at current levels, but Dobson is betting on the growth inflection and margin expansion. Risk/reward works if you believe the $67M revenue bogey is real — not yet in the price.
Cantor Fitzgerald just re-upped its Overweight and $600 PT on MKSI, naming it the top SMID cap pick ahead of the July 29 Q2 print. The thesis is simple: this is the best way to play the WFE equipment cycle without buying the system makers. MKSI's leverage as a subsystem supplier to wafer fab equipment, combined with tight advanced substrate markets fueled by AI demand, should drive fundamental outperformance vs. the OEMs. The stock is already up 238% in the past year, but at 27x NTM consensus vs. semi cap equipment average of 39x, there’s still room — and rapid deleveraging adds an EPS tailwind that could unlock a broader investor base.
“Meaningful upside to guidance and consensus estimates into CY27 and CY28, with EPS potentially reaching ~$20 and $26 compared to consensus of $15.22 and $17.32.”
Don't sleep on the China expansion either — MKSI is doubling its Guangzhou facility by late 2027, a $25M bet that the onshoring demand is real. Mizuho raised to $415, BMO initiated at $453 with Outperform, but Cantor is the outlier at $600 (25x discounted CY28 EPS). The Q1 beat (EPS $2.30 vs $2.04 est) and 2.86% revenue surprise gave the tape a fresh bid. The 2028 story is the catalyst, not this quarter's print.
Hold rating, PT bumped a dollar to $94. Jefferies isn’t buying the rebound. They see NEE’s premium compression as structural — stock now trades at just a 5% premium to peers on 2029 pro-forma P/E, down from a much wider spread before the Dominion deal was announced. That’s the core take: the market is pricing in execution risk and regulatory friction.
Data center growth is still the bull case, but Jefferies frames it as a “standalone” story now — not a merger-driven one. They expect large data center deals in 2026, at least in Florida (safe harbor). The proxy materials show strong standalone growth and Southern Company’s bid on Dominion as a wildcard. Key upcoming catalysts: Virginia and South Carolina merger applications.
“Regulatory approval data points since the announcement have been mixed but net constructive. Standalone execution on data center campuses remains the focus for the company.”
Other voices: BofA stays Neutral at $93 (citing higher interest expense and nuclear outage costs). UBS and Bernstein are more bullish at $105 and $107 respectively, hammering the dual business model. The spread between $94 (Jefferies) and $107 (Bernstein) tells you everything about the r/r — it’s a binary on regulatory outcomes and data center pipeline visibility.
Bottom line: NEE is cheap on a PEG of 0.47 and has a 56-year dividend streak, but the premium compression suggests the market is waiting for proof points, not promises. The next few months (merger filings, 2026 data center announcements) will determine whether the stock re-rates or stays stuck. Neutral with a narrow path to upside.
CANTOR STAYS OVERWEIGHT BUT THE H2 SETUP IS GETTING SQUEEZED. They keep the $600 target, love the long-term metrology intensity story (leading-edge leverage, advanced packaging share gains), but the near-term math is getting ugly. Stock down 20% from highs at $477, and Cantor flags the H2 ramp required for NVMI to outgrow WFE is a real bogey — last Q they guided to mid-teens WFE growth vs peers closer to 25%. That gap closes or the stock stays dead money.
"Concern about the second-half ramp required for the company to outgrow wafer fabrication equipment in 2026."
They're still projecting EPS upside to $15 in '27 and $18 in '28 (consensus $12.85/$14.56), so the long thesis is intact. Low-30x forward multiple isn't "overly demanding" per Cantor, but PMs are going to discount that out-year math if the H2 cadence doesn't show up in Q3 guidance. BofA raised target to $612 last week, so the Street is supportive — but this stock needs proof of life on the WFE outgrowth narrative, not just a cheap tape.
Cantor reiterates Overweight, $175 PT — stock at $117, up 218% in the past year but 27% off highs. The call is a low-conviction reiteration; they expect a modest beat-and-raise next week but see near-term upside capped by manufacturing inefficiencies ahead of the Farmer's Branch pilot line (Q4 2026, rapid ramp in 2027). The real thesis is 2027+ — HBM/DDR demand, share gains at Nvidia/AMD/ASICs, agentic AI CPU demand, and CPO opportunity. Cantor views sub-$120 as an attractive entry point, but warns the stock could trade sideways until Farmer's Branch ramps.
"Near-term capacity constraints will limit upside, and the firm believes the stock could see a trading range until Farmer's Branch ramps."
Bottom line: FORM is a long-term compounder, but for the next 6-9 months PMs should expect a chop. The 2030 target model has meaningful upside, but the P&L needs to deliver on manufacturing efficiency first. Not a catalyst-heavy name for the next few prints.
Canton stays Neutral at $35 despite an incrementally positive midquarter update. The consumption metric drama is real: first-quarter paid CRR revised down from ~$20M to ~$15M after adjusting for Flex and stripping out one-time credits. But it rebounded to >$20M by June 30 — so the trajectory intact, just noisy. New logos up 30% YoY and first orders in Q2 to date are +100% YoY. Management is leaning hard on agentic capabilities and platform innovation to drive that.
"Cantor Fitzgerald views the update as incrementally positive on new logo trajectory and AI attach, though the Consumption Run Rate remains a small and still evolving metric with limited financial materiality today."
Stock at $32.42, below the $35 PT and InvestingPro fair value of $34.41. Not broken, but not a catalyst bomb either. Neutral makes sense until Flex shows real scale.
Barclays sees a clean read-across from IBM’s miss: security spend is sticky, and firewall budgets are accelerating. IBM’s CFO admitted clients pulled forward capex into servers, storage, and memory to beat price hikes — and Barclays’ own checks confirm the same behavior in firewalls. Fortinet is a direct beneficiary. First-quarter firewall product growth already accelerating, pricing rising to reflect input costs. The broader backdrop? Mythos-driven threat environment is supporting security spending even as other IT budgets get squeezed.
"IBM disclosed that clients shifted quarterly capital expenditures toward servers, storage, and memory purchases to secure supply-constrained infrastructure ahead of expected price increases. Barclays said its checks suggest this behavior also applies to firewalls, benefiting vendors including Fortinet."
The counterpoint? IBM also said large deals failed to close because of client distraction from rapidly-evolving cybersecurity concerns. That’s noise — field reports point to faster sales cycles in security. Barclays is right to overweight the tailwind. FTNT prints next week. Watch for another quarter of product revenue acceleration.
BARCLAYS REVS UP THE TARIFF THESIS. PT to $279 from $213, keeps Overweight. Stock sits at $221 – room to run if the August tariff decision cuts their way. The driver: a long-term ASP bump to $0.40/watt, up from prior assumptions. Earnings nine days away, but the real catalyst is policy, not the print.
Barclays’ read: investors are “increasingly focused on what a new tariff regime could mean for U.S. module pricing, First Solar’s earnings power, and the competitive landscape.” That’s the whole ballgame. Bernstein still UW at $217, worried about Section 232 and tax credit extensions. Deutsche Bank flipped to Buy at $272, citing 2027 EV/EBITDA valuation. Two camps, one stock. The block trade of 672K shares at a 3.9% discount last week tells you conviction is mixed.
“A decision on tariffs is now likely in August.”
That line is the only macro that matters here. Everything else – ASPs, earnings, the Deutsche upgrade – is just positioning for that binary event.
AEHR — The real signal isn't the Q4 beat — it’s RECORD BOOKINGS OF $60.7M AND FY27 GUIDANCE OF $130-150M, UP 160-200% YoY. Wafer-level burn-in is now the fastest-growing test segment, adopted by a lead AI production customer. Hyperscale follow-on orders and a second device with 2x power per package confirm this isn’t one-and-done. The only concern is dilution from $97.4M in stock issuance, but the revenue ramp more than compensates if execution holds.
AVGO — Morgan Stanley fired back hard on the TPU share debate, calling the MediaTek displacement bear case premature. MS sees AVGO’s share at ~80% and growing; MTK doubles its business but the pie expands. That’s the critical signal — the main overhang is being dismantled in real time. Optical switching and NPO/CPO transitions add further pricing power. Risk/reward still favors the bull case.
MU — Memory content per AI server triples to ~25% of rack BOM on NVIDIA’s Rubin platform. HBM sold out through 2026, DDR5 contract prices up 40-50% QoQ, and SK Hynix ADRs surging 23-27% on day one — the memory upcycle is accelerating, not peaking. Softer CPI print adds macro tailwind. Earnings power is dramatically underappreciated.
SNDK — Lumped in with MU and LITE as a key beneficiary of the data center buildout and potential CPI cooling. NAND demand from AI storage and rising SSD content per server support the thesis. The memory/storage trade is driven by the same structural forces — AI absorbing wafer capacity and driving pricing power. Valid, but less immediate than HBM names.
GOOGL — Optical switching fabric is an underappreciated moat — a decade of co-design from TPU pod to datacenter spine creates a barrier merchant silicon can’t replicate. Google is pitching TPUs as cheaper and more reliable than NVIDIA GPUs to cloud providers, and securing 1.6GW of solar + storage for AI power. Gemini Managed Agents open a new monetization pathway. Vertically integrated advantage is real.
AMD — MI300X price target raised, demand outlook strengthening. AMD obtained incremental wafers at TSMC N4 from MediaTek cuts — supply relief for server CPU growth target of +15-20% this year. Some analysts see AMD on track to hit $1000. The NVIDIA CUDA lock-in remains the biggest barrier, but ROCm is improving.
ORCL — The market has rejected Oracle’s compute buildout completely. Stock down 58% from ATH, erasing $500B. Credit spreads widened 136bp vs benchmark’s 40bp. Management is doubling down on capex with no pivot. Contrarian interest at sub-triple digits, but the pain trade is lower until management shows discipline. Wall of future issuance to come.
GS — Q2 net revenue $20.34B (+39% YoY), EQUITIES S&T REVENUE $7.42B (+72% YoY) — ALL-TIME RECORD. ROE 23.5% vs estimate 16.4%. Goldman described the AI buildout as “relative early innings.” Lead-left bookrunner on SpaceX IPO and Alphabet’s equity raise. The central intermediary for AI infrastructure financing. Lower rates only help deal flow.
JPM — LARGEST QUARTERLY PROFIT IN US BANKING HISTORY: $21.2B. NII guidance raised to $105.5B. But CEO Dimon explicitly stated AI benefits accrue to customers, not shareholders. ~1,000 AI use cases identified, with only ~50 truly important — some areas have seen 30-40% headcount reductions, but most employees were offered other positions. Near-term earnings driver is AI capex financing, not internal AI gains.
ACN — Down 9% on IBM read-through — market pricing in AI capex cannibalizing traditional IT services budgets. ACN has significant legacy transformation and mainframe modernization exposure. The IBM miss raises questions about enterprise deferral of large-scale IT services engagements. More sentiment read-across than fundamental deterioration for now.
TSEM — Tower Semiconductor (backed by Japanese govt) expanding 300mm silicon photonics, SiGe, advanced packaging. Targeting $1.2B net profit and $3.6B revenue in 2028. At ~$28B market cap, that’s ~23.3x 2028 forward P/E. The market is beginning to understand that CPO delays benefit pluggable module incumbents, and Tower is a key enabler of the near-term NPO transition.
ATI — The public-market chokepoint for Starship-era superalloy demand. A full Starship stack carries ~39 Raptor engines, each consuming 800kg-1t of nickel superalloy — 25-30x the content of a Falcon 9. VIM melting capacity running flat out. ATI trades at 31.9x trailing EV/EBITDA vs peers at 43.3x and 38.4x — a steep discount despite exposure to five structural demand drivers.
CBRS — Cerebras’ investment thesis hinges on single-user throughput advantage for agentic AI workloads. Massive on-chip SRAM eliminates need for giant batches, enabling low-latency inference. On pure aggregate tokens-per-dollar for large batch workloads, NVIDIA still dominates. But for real-time, single-user tasks (cybersecurity, autonomous agents), Cerebras is differentiated. The market may be mispricing this niche.
SMTC — Initiated at Buy with $180 target (25x FY28E P/E). Transitioning from analog chip company to AI interconnect beneficiary. FY27 AI revenue expected >$350M, FY28E estimated at $1.22B, driven by 1.6T optics. Google TPU expansion a catalyst for CopperEdge ACC. NPO adoption is the next major growth driver post-2027.
LITE — Referenced alongside NVDA, MU, SNDK as a data center buildout beneficiary. Optical components directly tied to AI networking demand. But Rosenblatt noted optical stocks have been weak due to China InP capacity concerns and CPO scaling delays. Near-term challenging, but the CPO delay actually benefits pluggable incumbents like LITE for longer.
META — Framed as an ads company building a token factory. AI-powered ad tools and potential API-based monetization through Llama open new revenue streams. Cut ~8,000 jobs, reassigned 7,000 to AI. The shift from pre-sales to post-sales talent signals a focus on adoption and usage. Efficiency and AI reinvestment paying off operationally.