Good morning.
Futures flat to slightly red. NDX -0.3% as AI capex anxiety lingers ahead of MSFT and META prints. SKHY -4% AH on the sales miss — but HBM4 shipments started. BE +8% post-print, closing in on that $25B market cap. KLAC, STX, GLW all green in after-hours.
Asia mixed. Nikkei -0.4%, Hang Seng flat. Yen stable. FOMC tonight — market pricing a slim chance of a hike (that's the scary tail). Corporate credit impulse fears hitting a wall, but spot GPU rental pricing is 2x contracted rates. That math doesn't work forever.
Three themes framing the day:
1. Power is the new choke point. BE’s quarter was a paradigm shift — $1.065B revenue, 2x guided growth, 34% gross margin, $175M FCF. Customers cancelling turbines and reciprocating engines for fuel cells. The project finance shelf went from $5B to $25B. Power availability is now a direct AI revenue multiplier (ORCL/NBIS confirming: 1 GW = $12B-$24B annual revenue). Positive read-through to VRT, ETN, and the 800 VDC equipment cycle. Negative for CAT, CMI on structural share loss in onsite power.
2. Optical is real — scale-out only so far. GLW printed +32% in optical, +77% segment net income, and multi-year agreements with META (up to $6B), AMZN, NVDA. Verdict: demand so strong that “if we could make more, we could sell more.” But fiber count is neutral per GPU in Rubin (200G lanes). Scale-up (photonics) is the next leg. Positive for ANET, AVGO, MRVL on switching and SerDes content. Also for CIEN and fiber names.
3. Storage cycle tightening. STX saying mass capacity is demand-constrained, price-led, with contracted visibility through 2029. That’s a strong pricing environment for HDD and memory. SKHY missed Q2 but HBM4 is now shipping to ~10 customers under LTAs. MU and WDC are the direct beneficiaries. Bear in mind: Kimi K3 open source at 2.88T parameters shifts inference to cloud/API — negative for on-prem GPU servers (DELL), positive for hyperscaler storage demand.
We’ll hit up BE, KLAC, and STX first, then get to optics and the memory/storage group.
Cadence ripped the band-aid off with a Q2 beat - then dropped the biggest guide raise on a dollar basis in company history. Stock is up in after-hours, and the chatter is shifting from "is this priced in?" to "how much more is there?".
Revenue $1.584B (+24% Y/Y). Adjusted EPS $2.11 vs $2.05 consensus. Operating margin 45.5% vs 45.0% expected. This is not a beat on some one-time item - it's operational leverage playing out across a business that's already printing 86% gross margins.
Price targets span $349 (Piper, Neutral) to $450 (Benchmark, Buy). Consensus sits around $420-425 on the bullish side. Five firms all out with notes - Stifel ($432), KeyBanc ($425), Rosenblatt ($420), Benchmark ($450), Piper ($349). The bulls are loud. Piper is the lone neutral holding the line.
Broad-based strength across every product line:
The guide raise: revenue growth outlook lifted to 19% from prior ~17%. That's the best growth since FY22. The raise is front-loaded into Q3 with Q4 looking conservative - KeyBanc flags this as potential upside opportunity in H2.
BULL CASE (KeyBanc, Benchmark, Stifel, Rosenblatt): This is a compounding machine accelerating. 86% gross margins, 45%+ operating margins, and revenue growth re-accelerating to 19%. AI chip design activity is expanding TAM - not just for EDA but for Cadence's Agentic tools and AI-optimization software. Multi-year Intel Foundry agreements provide visibility. Benchmarks Mobley puts it simply: "investors have an opportunity to buy shares with a next-twelve-months P/E near the low end of the historical range against accelerating revenue and EPS." At current levels (~$339), this is the cheapest Cadence has been relative to its own growth trajectory in years.
"Strength in Core EDA, IP & Hardware Fuel FY26E Outlook Lift to 19% Growth" - Piper Sandler
"AI is a significant medium-term tailwind for Cadence's core EDA tools business" - Rosenblatt
BEAR CASE (Piper Sandler, cautious voice): Piper maintains Neutral at $349. Their logic: this is a great company at a full price. The 19% guide raise is impressive, but H2 organic recurring growth outlook implies "meaningful deceleration" per KeyBanc (who sees this as upside, not a negative). At 79x earnings, the multiple bakes in a lot of perfection. Piper notes the guide beat was only $10M vs the midpoint - the raise is real, but the base quarter wasn't a blowout. The Kimi 3 sell-off (China AI concern) that Mobley referenced shows how vulnerable the multiple is to rotation.
Incremental (the new data):
Peers: Direct read-through to SNPS (Synopsys) and ANSS (Ansys). If Cadence is seeing 40% IP growth on AI/HPC demand, Synopsys likely has similar dynamics in their IP portfolio. The non-semiconductor improvement Rosenblatt cites also bodes well for Ansys' industrial simulation exposure.
Thematic: EDA is becoming an AI-enabler, not just a design tool. Cadence's Agentic tools and AI-optimization software are expanding TAM beyond traditional chip design. The Intel Foundry multi-year deal signals foundry ecosystem buildout is sticky, long-duration revenue. This is the same narrative driving KLAC, AMAT, LRCX - but with a software margin profile that crushes hardware.
Risk: Multiple compression is the only real bear case. At 79x earnings, any growth disappointment (or AI narrative shift) will hit this stock hard. But right now, the rate of change is your friend.
THE MARKET IS OVERCOMPLICATING THIS ONE. Four notes, same conclusion: AWS is accelerating materially, the stock is cheap, and the only debate is whether you trust the 2H26/2027 ramp or get spooked by capex. Verdict: long.
Street is all over the place on PTs — $305 (UBS) to $360 (BMO), with $315-320 as the consensus cluster. That's a 15% range on a name with a $2.5T market cap. Tells you the models are diverging on one thing: AWS slope.
Everyone agrees on Q2. Revenue of ~$196B (17% YoY) is table stakes. Prime Day moved into Q2 this year — Evercore estimates $5-10B incremental revenue from the event, using 2021 as a proxy. Q3 gets messy because of timing (Evercore 7% below Street on Q3 rev). But this is calendar noise, not signal.
BULL CASE: AWS is the story, and it's not priced in.
Cloud demand is surging — Mizuho now models 33.5% AWS growth in Q2 (vs 31% Street), BMO's channel checks show "robust revenue acceleration" constrained only by capacity. The kicker: UBS projecting 36% AWS growth in 2H26 and 48% in 2027 — nearly 20 points above Street. Think about that divergence. If UBS is even close to right, the stock is absurdly cheap. > "At 14x our updated 2027 EPS, we believe there is significant room for the stock to outperform. Amazon should not trade at a discount to the market multiple." — UBS
BEAR CASE: Capex is the anchor, and the revenue shift is real.
UBS straight-up moved $3B of AWS revenue from 2026 into early 2027 due to OpenAI's Trainium deployment getting pushed. That's a real headwind to near-term comps. Mizuho cut PT to $320 on "valuation" — the stock trades at 23x their 2027 GAAP P/E vs a 34x three-year average. When an outperform-rated analyst is trimming because multiple compression is the base case, that's not nothing. > "The shares are very attractive given healthy e-commerce trends and accelerating AWS growth with an improving competitive position." — Mizuho (still Outperform, just not as punchy on price)
Material: The OpenAI/Trainium $3B shift is new. Not sure we can read too much into one customer's deployment timing, but it explains why UBS cut PT by 8% while keeping Buy. That's a modeling change, not a thesis change.
Material: Bedrock now offers OpenAI models (Mizuho). This is the real story no one is talking about enough. AWS's competitive position is improving, not eroding. For the hyperscaler narrative, that matters a lot.
Noise: Leo satellite project is a 2028 story. Workforce cuts in AGI group are ongoing optimization. Neither moves the 2026 numbers.
The actual debate: BMO ($360 PT) is the most aggressive on retail/advertising — citing Olympics, World Cup, elections, and Prime Day as "cyclically strong 2026." Evercore is the most cautious on near-term estimates but still Outperform. UBS is most explicit on the 2027 EPS divergence (49% above consensus on OI).
This is a Bellevue trade. If UBS is right on 48% AWS growth next year, MSFT Azure and GOOGL GCP comps should also be accelerating. But the spread between AMZN's multiple and its growth rate is wider than MSFT's. The AWS re-acceleration thesis is the strongest it's been since the post-pandemic cloud digestion ended.
The DeepSeek scare from earlier this year? Dead. Nobody's talking about it. Capacity constraints are real and bullish for capex. The only question is whether PMs have the stomach for the Q3 noise to own the 2027 ramp.
I do. Long AMZN.
Verdict: The AI observability train hasn’t left the station — but the conductor is already pricing in the express ticket.
14 analyst moves in the last 48 hours. PT range now $280-$330, with the Street converging on $300 as the median. Consensus revenue estimates for 2Q are 30.5% growth — but everyone with a Buy thinks DDOG will print mid-30s. The only holdout? Jefferies, who downgraded to Hold at $280 solely on valuation. That’s the bull case in a nutshell: the fundamentals are accelerating, but the stock is already up 85% YTD with a P/E of 670x.
PT cluster moved to $280-$330 from prior $225-$270 range. 13 firms raised targets (Citi $300, MS $300, TD Cowen $300, Guggenheim $300, JMP $311, BTIG $289, Benchmark $330 — the high-water mark). Consensus view: DDOG beats guidance, guides up. The bull case is built on three pillars:
Bull: "We see runway for stock to grind higher as growth accelerants strengthen." – Citi. 34-36% revenue growth is achievable this year. Core enterprise is accelerating off Q1’s mid-20s pace. GenAI spend is the structural driver — not a one-off. And DDOG is grabbing share from legacy vendors. The FCF margin is >20%, Rule of 50+ is real. Benchmark says the $330 target is "the highest on Wall Street" because they think DDOG is becoming a comprehensive AI ops platform.
Bear: "Position as AI leader already reflected in stock performance." – Jefferies. At 16x calendar 2027 sales, you’re paying for perfection. The premium margins and growth are real — but the stock is pricing multiple years of acceleration. If net new business growth disappoints even a little (e.g., if the AI ramp slows from LLM providers adjusting capex), the multiple compression would be violent. The Jefferies downgrade is the canary: they like the story, hate the price.
Already known: AI tailwinds, competitive wins vs Splunk, Q1 mid-20s core growth, healthy gross margins (~80%). The "DDOG is an AI play" narrative has been the dominant driver all year.
New / incremental:
"While bar is higher heading into 2Q, we think momentum for Datadog is sustaining with line of sight to mid-30% growth vs consensus' 30.5%. Fundamentals remain strong, as core enterprise improving off Q1’s mid-20s pace, with AI-natives ramping from LLM provider and 2 research labs." – Morgan Stanley
This is the cleanest "AI infrastructure spend is real" signal we’ve seen in months. If DDOG prints 34%+ revenue growth on August 6, it validates the thesis for the entire observability / infrastructure monitoring space. Dynatrace (DT) should get a lift — BTIG reiterated Buy on both. Cisco (CSCO) is the loser here: Splunk and AppDynamics are bleeding share. Snowflake and MongoDB are thematic cousins (data infrastructure + AI workloads) but DDOG is the purest proxy for AI-native consumption today.
The key risk for PMs: multiple expansion is already priced into the August 6 print. The real test is guidance. If DDOG guides Q3 above 32% (MS expects 32%, consensus at 27% high-end), this thing rips. If they merely beat and hold — expect a sell-the-news reaction. The bar is high. The setup is binary.
THE DEBATE ISN'T OVER. FFIV is up 60% YTD and the Street is still fighting it out. Bulls see an underappreciated AI beneficiary riding a hardware refresh cycle with no competition. Bears say that refresh is a one-time sugar high – and the stock at 33x earnings with only 7% software growth is priced for perfection. PT range is WIDE: $300 (BofA) to $508 (RBC). Consensus is a polite fiction – there is none.
Bull: This isn't a one-cycle pocket. The hardware refresh is structural because enterprises are ripping and replacing aging gear for cloud-native architectures. F5 has no hardware competitor in the space (Piper). AI demand is real – not just showrooming – with customers budgeting specifically for AI security in a platform approach. The direct monetization plus halo effect (RBC) means revenue quality is higher than the topline suggests. At 33x on a 20%+ earnings grower with expanding FCF margins, multiple expansion is justified. Stock is still ~6% off its 52-week high – room to run.
Bear: The hardware refresh is a known, finite catalyst. Once the iSeries base is migrated (end-of-service early FY27), Systems growth goes negative. Software is the long-term story and it's only 7% growth – not showing the kind of acceleration that supports a 33x multiple. The AI customer growth (50% QoQ) sounds huge, but Needham notes it's "early stages" – no revenue visibility. BofA points to "abnormal comps" in FY26 that flatter the numbers. PEG ratio of 2.07 on a stock that's already up 60% YTD is not a bargain.
"F5 Networks remains an underappreciated AI beneficiary with potential for multiple expansion." — RBC Capital
"The refresh cycle remains consistent and F5 lacks a hardware competitor." — Piper Sandler
"We are not confident that AI tailwinds and Software can drive outperformance [after the hardware refresh]." — Needham
"Growth upside should moderate going forward as F5 works through the remaining refresh opportunity." — BofA Securities
FFIV is a microcosm of the "AI infrastructure – real or cyclical?" debate. Hardware-driven AI beneficiaries (think servers, networking gear, ADCs) are all facing the same question: is this demand structural or a one-time upgrade cycle? Peers like A10 Networks (ATEN) and Radware (RDWR) are seeing similar refresh tailwinds. If you believe the bear case on F5, you're implicitly shorting the whole ADC hardware cycle. If you buy the bull case, you're betting on a multi-year platformization that spills into software (security, app delivery, edge). For PMs: watch the software attach rate trajectory. If F5 can show that software growth accelerates after the iSeries wave crests, the bear thesis breaks. If not, 33x is a trap.
Bottom line: This stock is a conviction call. The bulls have the upside on their side (RBC $508, Piper $461). The bears have math on theirs ($300 implies >25% downside). The earnings report was undeniably good – but the question is whether the market is pricing in the peak of the cycle or the start of a new one. We'll know more in two quarters.
THE TAKE: MSFT prints F4Q26 Wednesday and the setup is ugly—stock down 19% YTD, -30% from highs, trading at 22.9x P/E (low relative to growth). But the analyst community is circling the wagons. UBS cut to $480 (from $510) on Azure capacity constraints and token optimization fears. Everyone else is holding Buy/Overweight at $502–$646. The debate is simple: is Azure AI growth peaking, or is this a classic "buy the dip" moment before the AI monetization cycle really inflects?
We’ve got a cluster of Buy ratings but wide PT dispersion ($480–$646). UBS’s Karl Kierstead is the outlier on the downside, having talked to a dozen customers/partners. His concern: capacity constraints on external customers + early token optimization could cap Azure upside. He also flags that the market hasn’t rewarded higher capex signals (see: Oracle, Google). Compare that to Cantor ($502), Benchmark ($525), Piper ($540), Citizens ($550), and Bernstein ($646)—all betting the AI narrative is intact and the stock is simply mispriced.
"Capacity constraints on external customers and early token optimization efforts could limit potential for material Azure guidance upside." — UBS (Kierstead)
Piper Sandler is leaning into the security angle—MSFT launched MAI-Cyber-1-Flash (96% on CyberGym benchmark, 50% lower cost vs. current MDASH) and Project Perception. They see MSFT as the largest security vendor and a key player in agentic security. That’s a medium-term catalyst, not an earnings driver this week.
Citizens points to MSFT’s letter on open-weight AI models as a strategic move to shape regulation. Not a near-term P&L item but signals management is playing offense on policy.
BULL: You’re buying a 22.9x P/E, 18% revenue grower with Azure demand "healthy" (UBS’s own survey says AWS and Azure both look fine). GitHub Copilot pricing change is a modest growth kicker. AI security products are high-margin and differentiate the stack. The selloff is emotional—not fundamental. Bernstein’s $646 PT implies 65% upside.
BEAR: Token optimization (models doing more with less) is eating cloud revenue per unit. Capex is going up, margins are compressing, and the market is punishing anyone who spends big without immediate cloud revenue acceleration. UBS explicitly calls out that "the market has not rewarded stocks that signal higher capex or fail to post upside to cloud revenue growth." MSFT’s Azure guidance Wednesday needs to be a beat AND a raise—otherwise this stock gets coiled lower.
BOTTOM LINE: This is a binary event. If Azure AI revenue surprises to the upside (or at least inline with whisper), you get a relief rally into a low multiple. If they guide Azure deceleration or capex spikes without commensurate revenue, the $480 UBS target starts to look optimistic. I’d want to see the print before adding—but the risk/reward at $392 with a $480–$646 street range is skewed long if you believe the AI capex cycle has legs.
Verdict: The AT&T deal is the catalyst the bulls needed. Three analyst firms all landed on Buy today, and the common thread isn't a product roadmap or a 2032 gate-model dream — it's a real enterprise use case producing real results. The stock is at $18.79. These PTs cluster around $30-45. The gap between price and narrative is narrowing, and that's the trade.
Three separate analyst actions today, all tied directly to the AT&T expansion. Benchmark initiated at $30. Stifel reiterated at $35. Rosenblatt held its $43 PT. The common thread: AT&T is using D-Wave's annealing optimization to cut network processing time from an hour to UNDER 15 SECONDS.
That's not theoretical. That's a Fortune 50 telecom deploying quantum in production. Stifel called it explicitly — this is a "validation point for D-Wave's annealing technology and an industry traction proof point." Not a lab experiment. Not a research grant. A live network workload.
The agreement includes an exploratory path into outage detection, technician routing, network build planning, and traffic management. AT&T is also evaluating D-Wave's forthcoming gate-model systems for quantum security and communications. This is a deepening relationship, not a one-off pilot.
Bull case: AT&T is the blueprint. A blue-chip telecom deploying annealing quantum on real network problems creates a reference architecture that other Fortune 500s will copy. The path to gate-model (2026 simulator launch, 2032 roadmap) adds a call option on the next leg. PT range of $30-45 implies 60-140% upside.
Bear case: No financial terms disclosed on the AT&T deal. Revenue impact is guesswork. The stock is still 34% off its 52-week high of $46.75. Gate-model quantum is years away and capital-intensive. Analyst consensus doesn't expect profitability this year. This is still a narrative stock that needs to convert excitement into dollars.
"The announcement serves as a proof point for D-Wave's Advantage optimization solution." — Rosenblatt
Stifel went further: this is "practical, near-term commercial application" from annealing systems. Not a ten-year vision. Right now.
Q2 results are due August 6 pre-market. That's the next real catalyst. We'll be looking for any AT&T revenue contribution, customer pipeline expansion, and comments on gate-model simulator pre-orders.
The listing transfer from NYSE to Nasdaq (effective today) is noise. Don't get cute with arb spreads. The AT&T deal is the signal. The PMs who are long are betting on more of these enterprise deployments before year-end.
Verdict: Bookings momentum is the story. Three analysts (Cantor, Needham, Oppenheimer) all stepped forward after the FQ1 beat-and-raise. Consensus is building around organic re-rating potential. Not a meme. Not a macro fluke. This is a business inflecting on execution.
FQ1 print: $0.49 on $87.7M revenue — beating consensus by $0.10 and ~$1.7M. Management raised full-year rev guidance and bumped subscription growth outlook to 32% from 30%.
BUT the headline isn't the quarter. It's the backlog and bookings trajectory.
Needham called it out: "Bookings represented the best six-month stretch in company history." The key stat: THREE seven-figure ecosystem wins in Q1 alone — a Las Vegas Strip casino, an Arizona casino resort that ripped out a two-decade incumbent, and a nine-property Australian resort. That's real competitive displacement, not just wallet share expansion.
PMS revenue surpassed POS for the first time. That's structural. The Marriott PMS rollout is underway and "on plan" per management. Cantor noted the Marriott conversion to subscription is directly driving the guide raise.
PTs went to $118 (Oppenheimer), $130 (Needham), and Cantor holds at $140 with Overweight. That's a $118-140 range on a stock at ~$106. Cluster is constructive, not stretched.
Needham's line — "continued upside to estimates and re-rating potential" — matters more than the specific PT. They're arguing the rate of change justifies a multiple expansion, not just earnings catch-up.
Oppenheimer was slightly more measured: flagged a decline in Q1 deal volume and questioned whether the high-teens POS growth guide is realistic given competition. Not a torpedo. Just a real debate point for PMs managing downside.
"The strong start to fiscal year 2027, with first-quarter records for sales, revenue, and profitability that beat expectations on both the top and bottom line." — Needham
Bull case: Record bookings, POS-to-PMS mix shift, Marriott tailwind, and AI product development (central reservation system, revenue management) create a multi-year growth algorithm. Organic re-rating is plausible.
Bear case: Premium valuation (P/E ~77x) fully prices in the acceleration. POS competitive intensity could cap growth. Deal volume dipped in Q1 — not sure we can read too much into that, but it's a data point. If the bookings pipeline starts to roll over, the multiple compresses fast.
Bottom line: This is a show-me story trading above the market's average conviction level. The bookings data makes a near-term case to hold or add. For the PM thinking about position sizing — AGYS is a conviction name, not a hedge.
Verdict: Two Street voices double down after Q2 — Baird and Needham both reiterate with $20-21 targets (75-84% implied upside). The AI power architecture thesis is alive and accruing. But the P&L is horrid. Stock gets the narrative trade, not the earnings trade yet.
Baird (Outperform, $20) and Needham (Buy, $21) both held ratings after the Q2 print. Their collective view: the HVDC 800V transition in AI data centers is the real deal. Baird called the first wave of new power architectures “stronger than expected.” Needham cited a record book-to-bill and backlog extending into 2027. That’s not just momentum — that’s a structural demand signal.
The stock sits at $11.41, so these are bold upside calls. But they’re based on a multi-year inflection, not next quarter.
AI infrastructure is now 1/3 of Q4’26 revs — up from a prior target. That’s a huge mix shift for a company that was basically all consumer/EV GaN a year ago.
New product line expands TAM by ~$1B by 2030 — the 1.2 kV SiC JFET. And the Magnachip license deal for Gen 4/5 Trench-Assisted Planar gets them process technology without building it internally. Smart capital allocation for a company losing $0.95/share.
Needham highlighted the revenue inflection in AI Infrastructure as the core reason to stay long. The three other “inflection points” in the 800V HVDC transition are all on schedule — second wave (SiC/GaN in data centers) starts contributing mid-2027 per Baird.
“The first wave of new power architectures in AI data centers is underway and stronger than expected, driven by heterogeneous AI compute architectures.” — Baird
Bull case: This is the GaN/SiC pure-play story finally hitting commercial scale. $10.5M Q2 revs — up 22% QoQ, beat consensus — and backlog pushing into 2027. The 800V architecture shift is inevitable, Navitas is the best positioned small-cap vehicle for it. Multi-bagger potential if they execute.
Bear case: The non-GAAP loss of $0.95/share vs. consensus of -$0.04 is a face-plant. Revenue of $10.5M is still tiny. Multiple analysts flag “not profitable this year.” The thesis depends on a technology transition that’s still nascent — any delay in hyperscaler adoption kills the stock. At 30x forward revenue? No thanks.
NVTS is a pure narrative compounder right now. Q2 revenue beat keeps the dream alive; the loss shatters any near-term fundamental anchor. The $20-21 PT cluster implies the Street expects the narrative to win. PMs who want exposure to AI power architecture without buying the incumbents should own a small position. But don’t look at the P&L.
Verdict: Stifel just ripped the PT to $500 from $445 — and that might still be conservative if the AI networking thesis plays out. Q2 was a beat-and-raise masterclass. The narrative is shifting from “supply-constrained” to “capacity being lined up against 52-week lead times.” This is a structural re-rate story, not a quarterly pop.
CLS crushed Q2: REVENUE $4.7B vs guide of $4.4B, EPS $2.54 vs $2.30 — both above the high end. Adjusted operating margin hit a record 8.2% (guided 8.0%). Not a one-off. Growth was broad-based:
Stifel’s note focuses on the 2027 outlook — management sees CCS revenue growth accelerating to ~85% again, pointing to a $33.8B revenue floor. That’s the hook.
“Strong demand for 400G and 800G products, ten active 1.6T programs and next-generation AI compute ramps… two multi-billion dollar 2027 opportunities including OpenAI/Broadcom Jalapeno and AMD Helios for rack-scale and scale-up solutions.”
This is the crux. CLS isn’t just a contract manufacturer anymore — it’s the plumbing for AI cluster interconnect. The 1.6T ramp is the next catalyst. Stifel explicitly calls out supply as the bottleneck, but capacity is in place or planned and component lead times exceeding 52 weeks are being offset with binding customer forecasts. Material risk is “increasingly mitigated.”
Gross margin stays mid-11% (current 11.95%). Operating margin expansion will come from HPS mix, site productivity, and ATS recovery — not GM leverage. That’s a clean story.
BULL: CLS is becoming the go-to partner for hyperscaler AI networking. The 2027 revenue floor is $33.8B, implying ~65% CAGR from the new guide. Margins are still expanding even at scale. The 1.6T cycle is just starting — and CLS has ten active programs. This is a multiyear compounder trading at ~44x forward EPS — cheap for 85% revenue growth.
BEAR: Supply constraints are real. If component lead times stretch further or demand snaps, CLS gets caught with fixed cost overhang. The stock has already returned 84% in a year — how much of 2027 is priced in? Also, the customer concentration risk (Broadcom, AMD) is non-trivial. One program delay could crater the narrative.
Stifel’s move to $500 signals they see the 2027 story as derisked. The Q2 beat wasn’t a surprise — the magnitude of the guide raise and the specificity on 1.6T programs was. CLS is the pick-and-shovel play in AI networking that most PMs still think of as an EMS box. That perception is changing fast. Buy the pullbacks, but don’t fight the tape here — the momentum is too clean.
Verdict: Cleanest way to play data center load growth in regulated utilities — Q2 proved execution, but the revenue miss keeps the bears alive. Two firms raised PTs (Bernstein to $108, BMO to $96) on the back of a $1.15 EPS beat (+9.5% y/y) and monster large-load pipeline at FPL. The 3.6GW of renewables & storage origination in a single quarter? That’s not just noise — that’s a backlog growing to 35.1GW from 33GW. The bull case is intact. The bear case: revenue whiffed by ~$660mm and the Dominion merger still has regulatory risk.
EPS of $1.15 topped consensus by $0.02 (Bernstein) or $0.03 (BMO). Revenue printed $7.53B vs $8.19B expected — a miss that gets shrugged off in a regulated utility where earnings quality > top-line timing. Management kept FY26 guidance at the high end of $3.92-$4.02 and reaffirmed 8%+ standalone long-term growth through 2035 (proxy math implies ~9% CAGR). That’s the kind of visibility PMs want.
Florida Power & Light: 21GW of large-load interest, 12GW in advanced discussions, 8GW in development by 2032 (up from 6GW). Management expects at least one large-load transaction at FPL by year-end. That’s the data center anchor. Bernstein’s take:
“NextEra’s FPL pipeline is the single strongest data center load signal in regulated utilities. The 8GW by 2032 is conservative — we think it could double if hyperscaler demand continues at current pace.”
BMO noted the 3.6GW origination beat “highlights continued demand for solar and storage solutions and continued execution.”
Bernstein Outperform, PT to $108 from $107 — effectively a rounding error on PT but the message matters: nothing in Q2 derailed the thesis. BMO Outperform, PT to $96 from $94 — their sum-of-the-parts mark-to-market creeps higher. Mizuho (not a mover today) stayed Neutral at $95. The PT range of $95-$108 reflects the bull case vs. the reality check on execution. Bernstein is clearly the more aggressive, leaning into the data center narrative.
Bull: Regulated data center load is the next 10-year demand cycle for utilities. NEE has the best positioned regulated utility (FPL) in the fastest growing state (Florida) and the largest renewables development arm (Resources). 8%+ organic growth is essentially guaranteed; the pipeline suggests upside. PEG of 0.37 is absurdly cheap for that.
Bear: Revenue miss raises questions about non-regulated earnings quality. The Dominion merger (closing 2H27) could be a distraction or regulatory headache. If AI / data center demand cools, NEE’s premium multiple (20x P/E) gets unwound fast. Mizuho’s $95 says the stock is fairly valued here.
SpaceX sitting at $107.45 – that’s below KeyBanc’s last call and just a hair above its 52-week low. DOWN 8% IN A WEEK. The market is pricing in a massive risk premium on timing. Starship Flight 13 was near-perfect, but “near” matters when you’re burning $1.5T in market cap expectations.
The collective analyst take: Both KeyBanc (Sector Weight) and Bernstein (Outperform / $239 PT) agree Starship progress is real. KeyBanc flags the real tension – SpaceX stopped taking Falcon 9 bookings after 2028, and Starship isn’t yet at rapid cadence. Bernstein notes the V3 booster still hasn’t been recovered (engines failed to relight on Flight 13). That’s the chasm.
Bull case (Bernstein, Raymond James): Starship V3 design upgrades validated. Flight 13 deployed 20 actual V3 Starlinks (non-operational, but the deploy mechanism works). Full reusability is now an engineering problem, not a physics problem. Raymond James PT of $800 implies 7x from here if execution hits.
Bear case (the price action): Valuation is entirely forward-looking – 108% revenue growth projected, but profitability just turning this year. KeyBanc basically says “we don’t know when these growth vectors monetize.” If Starship cadence slips, multiple compression is vicious from these levels.
“Valuation increasingly appears contingent on the timing of future milestone achievements and commercialization progress.” – KeyBanc
That’s the whole debate in one sentence. The next 12 months are about turning flight counts into payload revenue. Earnings due in SEVEN DAYS. Expect fireworks.
Verdict: Still the AI server infrastructure trade, but the concentration risk is getting harder to ignore. Evercore ISI reiterated Outperform / $500 PT (stock +242% YTD, now $275B market cap). The bull case rests on Dell's ability to capture an outsized share of the largest AI buildouts — the firm estimates three customers (CoreWeave, xAI, IREN) will drive ~70% of Dell's FY'27 AI server revenue guidance of $60B. That's a lot of eggs in three baskets.
Evercore ISI notes that Dell’s ability to rapidly bring next-generation systems online remains a competitive advantage.
The bear case is not from Evercore but from the broader tape: GF Securities just downgraded to Hold on valuation after the massive rally, citing limited upside. Supply constraints are also expected to tighten in 2027. Still, the rate of change in AI infrastructure spend is accelerating — Dell's PowerEdge XE8812 with Nvidia Vera Rubin NVL4 is another proof point. PMs should watch for any signs of order slippage from the big three; if that holds, the $500 PT stays in play.
Hold the phone – literally. META slides 7.8% to $603 ahead of a must-watch July 29 print. Benchmark reiterates Hold, and the message is clear: show me the money on AI infra, or the premium PEG (2.88x on 25% ROIC) cracks.
The crux: compute capacity targets for 2026-27 are aspirational, not reality-based. Benchmark ran low/base/high capex models – they're not buying the gigawatt glide path. Five questions for management prepared; we'd add "what's the real ROIC timeline?".
"Aspiration and reality are increasingly diverging, particularly regarding compute capacity."
Needham also reiterated Hold, citing value destruction risk across multiple projects. BlackRock marketing $12.3B in bonds for a Meta data center doesn't help the "spend is out of control" narrative. New product features (Muse Spark 1.1, Facebook Verified, Seller app) are nice, but they don't move the needle on the capital allocation debate.
Verdict: META is a prove-it name. Either they deliver concrete AI monetization data or the multiple compresses further. We'd rather wait for the print than step in front of it.
$167, 35% off the high, and one lone bull ($300 PT) saying the handset fears are overdone. Market has priced Qualcomm for a commodity chip story — Benchmark disagrees, sees data center upside nobody’s capturing.
Earnings July 29. Consensus $9.68B rev / $2.23 EPS — Benchmark inline on both. The nuance: they’re CAUTIOUS ON HANDSETS (the growth engine everyone worries about) but see UPSIDE IN AUTO + IOT + QTL. That’s a 3-legged stool that keeps the thesis alive even if phone units suck.
Most PMs I talk to view QCOM as a low-growth dividend stock (2.2% yield, 18.4x P/E) with an earnings ceiling. Benchmark’s $300 implies ~80% upside. That’s a bet on the data center pivot — Qualcomm is actively telling investors about it. The stock doesn’t move until they prove it.
Cantor cut to $200 (Neutral) — the other side of the trade. They think Qualcomm beats June but only guides slightly above consensus. Not exactly a barn-burner setup. Meanwhile, QCOM just told customers it’s raising prices double-digits starting September — a margin offset, but demand risk if they push too hard.
Options market pricing an 8.2% swing on July 29. That’s low vol given the 35% drawdown — tells you positioning is already neutral/underweight. Any print that shows non-handset momentum accelerating gets squeezed.
Benchmark is the only firm on the street with a $300 PT. They’re leaning into the RATE OF CHANGE narrative: data center is starting from zero, auto design wins stack, IoT is recovering. Handsets are the side show. If QTL licensing revenue holds up, that’s high-margin gravy.
Bear case: handset volumes keep deteriorating, price hike kills orders, and data center remains a story (read: no revenue). Cantor’s $200 is the floor in that world — you’re basically buying a 12x P/E no-growth business.
My take: the $167 price already bakes in a lot of pessimism. Earnings risk is to the upside if they show any data center traction. Not a conviction long, but the r/r is interesting for a late-summer contrarian bet.
VERDICT: ROSENBLATT THROWS $900 PT ON WDC — HDD SUPPLY CRUNCH IS THE NEW NARRATIVE, AND THEY’RE BUYING THE MULTIPLE EXPANSION. That’s a 2x jump in target from their prior $500. The thesis is brutally simple: nearline HDD is sold out, LTAs are resetting at $20-30/TB vs current $14-15, and incremental gross margins are approaching 80%. FY28 EPS estimates have surged ~170% YTD. WDC trades at 27x trailing — but on a forward pricing framework they see $65-70 eps power, implying high-single-digit P/E. That’s the hook.
"Current spot pricing trades at a 30-40% premium, with forward long-term agreements resetting toward $20-30 per terabyte versus current levels of $14-15 per terabyte."
Lot of faith in structural undersupply persisting into FY28. Stock’s already up 189% YTD (+624% over one year), so nothing is a secret. But Rosenblatt’s 25x on that forward earnings is aggressive unless demand stays above supply for another 2+ years. Worth watching if Seagate ($1300 PT) gets dragged along.
Verdict: Intel is the turnaround story that’s actually working — but the stock has already repriced the good news. Q2 crushed: $16.1B revenue (+25% y/y), $0.42 EPS (double consensus), gross margins 41.8% vs 39.2% expected. Client $8.9B (beat by $900M on ASPs), DCG $6.3B (beat by $800M, supply still tight). The analyst crowd is throwing PTs in a $100–$125 range (Baird $125, Stifel $110 on valuation, TD Cowen $115, Bernstein $110). The bull case is straightforward: Intel is executing under Lip-Bu Tan, supply is constrained, and narrative momentum is real. The bear case: at ~8x EV/sales on 2027-28 numbers, the multiple is pricing in perfection. Stifel trimmed the target on valuation concerns — that’s the tension.
Bernstein’s take: "Intel is pulling through with both the market and narrative increasingly supportive."
Not sure we can read too much into a single 346% move over the past year — that’s already a lot of good news. The next leg needs gross margin improvement to actually sustain, not just beat low bars.
Verdict: Mobley thinks the Q2 beat + inline guide is "good enough for the bulls" given the stock is DOWN 40% FROM THE 52-WK HIGH and trailing the SOX YTD. Reiterated Buy, $165 PT. That's 29x his new NTM EPS – near the 52-week low multiple.
The quarter was clean: Q2 REVENUE A RECORD $207.4M (first time above $200M), EPS $0.77 vs $0.72 consensus. Product revenue hit a new high of $99.2M, driven by DDR5. Q3 guide sees product revs up 14% sequentially at the midpoint.
The bear case? Stock has been dead money – down ~9% just in the past week. But Mobley's point is that this setup already reflects pessimism. The results prove the memory interface / silicon IP cycle is intact.
"Considering RMBS shares have underperformed the SOX YTD, and considering shares are down more than 40% from the 52-wk high, RMBS’ results/guide should be good enough for the bulls."
KeyBanc stays Underweight at $250, and the thesis is simple: this is a unit growth story masquerading as a services story, and the math doesn’t work at 37x FY27 EPS. The rotation out of semis into AAPL as a “safe” name is exactly the kind of positioning that gets punished when the iPhone 18 cycle disappoints. Baird bumped their PT to $330 and Evercore is at $365, but those are backward-looking tailwinds (M6 refresh, leasing program) – KeyBanc is looking at the rate of change in upgrade rates and go-to-market shifts that haven’t hit consensus yet.
“KeyBanc noted that a recent sell-off in the semiconductor sector has driven investors toward Apple as a perceived safe haven… the firm warned that rising iPhone prices could lead to falling unit demand, which would slow user growth and likely impact Services revenue.”
The 40.7x trailing multiple leaves zero room for error. If the FQ3/FQ4 print is “adequate” (their word) but units are flat-to-down, the rerating comes fast. PMs should watch the mobile network operator upgrade data – that’s the canary.
BofA reiterating Buy at $180 ahead of Wednesday's print. Stock up 137% in a year — P/E at 48.7x, so not exactly hiding in the weeds. But the thesis remains intact: cloud/power growth, improving mix, and CPI segment margin expansion >100bps as programs ramp. Q1 model $7.5B rev / $0.91 EPS (in line). Q2 model $7.77B / $0.99.
"BofA expects favorable mix and the recovery of prior CPI investments to support operating margin expansion of more than 100 basis points in the segment."
Freedom Capital also upgraded to Buy on pullback dynamics, and the JetCool liquid-cooled Dell server launch gives the narrative a nice AI-infrastructure bow. Big run, but if they deliver the margin story, multiple compression risk is manageable. Earnings Wednesday – watch what they say about H2 acceleration.
Verdict: In-line quarter priced in, but the September guide is the real swing factor. Benchmark sees Q3 revenue of $925M ($1.03) – barely a blip vs Street. Mobile down 3-4% QoQ, Broad Markets flat. Ho-hum. The real question is what Skyworks prints for September and whether the QRVO acquisition thesis (cost synergies, Apple leverage) still holds after KeyBanc’s downgrade to Sector Weight. RBC lifted PT to $80 but kept Sector Perform – no fireworks. Options market bracing for 9.1% move on print.
Bull case: SWKS at 25.5x P/E looks cheap if the QRVO cost synergies finally materialize. Call options surged to 32,697 contracts – someone’s positioning for upside. PCIM Europe demo in AI data center / EV / industrial power applications could open a new growth vector beyond phones.
Bear case: Apple concentration + smartphone unit softness = structural headwind. KeyBanc pulling the ripcord on the QRVO thesis is a big deal – if the Street stops giving credit for that deal, multiple compression follows. September EPS $0.05 below consensus is a brick. Not sure we can read much into that options surge – could be a hedge.
Hold, don’t chase. The print itself is noise. Watch the September guide and listen for any QRVO synergy commentary on the call. If management can't convince the Street the deal was worth it, $80 (RBC’s PT) becomes the ceiling, not the floor.
BTIG stays Buy at $38, but the market is pricing in execution risk — not the AMD transformation. Stock down 12% on the announce to $19.91 (off a $22.75 close yesterday), despite a deal that delivers roughly $14B in contracted revenue over 15 years across 530MW of IT load. AMD directly takes 380MW on a triple net lease (BTIG sees 10-11% unlevered returns) and a neocloud takes 150MW on a modified gross (7-8% returns). AMD also gets warrants for 30M shares at $23.47 and reservation rights on another 1.9GW through 2028 — a huge call option on CORZ’s buildout.
CORZ now sits on >1GW of leased hosting capacity, with ~440MW already billing (up nearly 200MW since Q1). The bull case: this is a blue-chip anchor tenant that de-risks the entire pipeline and validates the pivot from Bitcoin mining to HPC colo. The bear case: capital needs are still massive, timelines are long (15-year build), and the stock is already up 50% in the past year — the market may be asking what’s next? after this one mega-deal. BTIG thinks the r/r is fine at $19.91 against a $38 target, but the street wants to see the cash flow math pencil out before re-rating.
Benchmark slaps a Buy on Rigetti with a $25 PT (64% upside from $15.24). The call isn't just stock-picking — it's a sector re-initiation, complete with a deep-dive primer on quantum compute. Thesis: tech legitimacy building, fault tolerance needed for universal Quantum Advantage, but this is NOT a winner-takes-all market. Benchmark thinks the right move is a portfolio approach across names. (Consensus PT range: $20-$40, but the stock's already up 8% this week on the CHIPS Act LOI.)
"Prior to the broad achievement of Quantum Advantage, the best strategy for investors may be to take a portfolio approach to investing in the sector."
Beyond the initiation: Rigetti just inked a $100M CHIPS Act LOI (DOE gets a pro-rata equity stake), Q1 FY26 beat on multiple system sales, launched Cepheus 1-108Q on AWS and Azure. Cantor ($30), Craig-Hallum (Buy), Needham ($31) all reaffirmed. Revenue still microscopic ($10M TTM on a $4.7B market cap) — pure optionality on quantum's timeline.
Benchmark drops a buy with a $60 PT (stock at $35.27, down 28% in six months). The analyst is framing this as a quantum computing sector call — not a winner-takes-all bet. Mobley's thesis: the tech is gaining legitimacy on fault tolerance progress, and IonQ has real commercial traction (335% revenue growth LTM, cash-rich balance sheet). But the key line is the portfolio approach — they're basically saying don't go all-in on one name.
“In the early days of quantum computing, prior to the broad achievement of Quantum Advantage, the best strategy for investors may be to take a portfolio approach to investing in the sector.”
The $60 PT is near the low end of the $44.78-$100 analyst range — not screaming conviction, but a clean floor. Recent Q1 revenue print ($64.7M, beat) and the Clavis XG Multiplex launch (quantum-safe encryption on existing fiber) add near-term narrative juice. Still a show-me story until Quantum Advantage is proven at scale.
Dead money until May 2027. TD Cowen throws in the towel, downgrading INTU to Hold from Buy and slashing the PT to $304 from $504 — right where the stock sits ($303.91). The call is brutally honest: no catalyst path until the FQ3 '27 print next May, when management gives a preliminary view of tax season. Until then, you're staring at a horrid FQ4 '26 print (August), a September investor day that Levine expects to yield zero needle-moving announcements, and a PEG ratio of 0.53 that the market is happy to ignore with AI risk and competitive intensity hanging over the name.
The key thesis: TD Cowen expects F'27 revenue guidance to come in below both consensus and the 10%-plus long-term target when INTU reports next month. They also think management will lower long-term growth targets for Global Business Solutions and TurboTax. This isn't just a cycle problem — it's a structural narrative problem. AI disruption, IRS competition, and a pricing shift that's spooking the Street. The only bull case left is valuation, but the market is refusing to catch that falling knife.
"Near-term catalyst path is skewed more negative than positive, limiting potential recovery in shares until at least the company’s third-quarter fiscal 2027 print in May 2027."
Don't confuse the 0.53 PEG with a buy signal. Multiple firms have piled on recently — Morgan Stanley to Equalweight, Stifel to Hold, Freedom Broker to Hold — all echoing similar themes. BofA is the lonely bull at $400 on cash flow estimates. Mizuho still Outperform but cut to $500. The consensus is clear: this stock needs to show it can grow again before anyone steps in. Sustained outflows from the AI trade are the primary risk to even a neutral view. Stay on the sidelines.
Upgrade to Overweight (PT $50) vs $19.88 close. KeyBanc sees Firefly as a pure-play lunar infrastructure bet. The story: 71% revenue growth (to $185M), three new NASA contract awards totaling $219M in the last few weeks (MoonFall $75M, CLPS $144M, SkyFall $13M), and a clear path to three more Alpha launches in 2026 plus Eclipse first flight in 2027. The 67% drawdown from $73.80 high is the entry point – if they execute on launch cadence and lunar lander missions.
The missile tracking angle is a near-term kicker:
"KeyBanc analyst Michael Leshock noted that SciTec remains positioned across missile warning and tracking initiatives. The firm expects heightened missile tracking demand amid the Iran war to provide near-term tailwinds."
Beyond lunar, the acquisition of Space-ng (AI vision navigation) and the Sweden launch pad milestone (orbital launch from Esrange target 2028) broaden the TAM. But the bull case hinges on Blue Ghost and Elytra driving the next leg. Risks: still pre-revenue on the high-margin lunar services, capex for Eclipse development, and the usual SPAC hangover overhang (FLY went public via SPAC). At 10x forward revenue (assuming $324M this year), it's not cheap for a company still burning cash – but if lunar awards keep accelerating, the r/r flips.
THE POST-EARNINGS SELLOFF IS A GIMME FOR ANYONE WHO WASN'T ALREADY LONG. Guggenheim keeps Buy / $175 (implied 38% upside) — they see the drop as a healthy correction after a 66% six-month run that lacked fresh catalysts. Q1 delivered a clean beat (EPS $1.42 vs $1.16e), underlying demand is "strong," and the entire FY27 NNARR guide of $190M (~19% growth) is intact.
"The post-earnings sell-off as a healthy correction following intra-quarter stock appreciation that lacked fundamental catalysts."
The real story: 81% gross margins, SaaS NRR above 120%, one-third of new business from security, and a fresh agentic resilience narrative with Microsoft Azure as a second-half catalyst. Selloff is positioning, not thesis — I'd be leaning into the dip here.
Needham stays constructive on the restructuring, but the market isn't buying it yet. Shares at $4.98 against a $10 PT tell you everything about the skepticism. CEO's open letter lays out a three-group product consolidation to funnel resources toward AI-monetizable products. Modest margin benefit expected. Guidance reiterated — that's actually meaningful given Upland's history of only "roughly in-line" prints (Needham's words).
"Needham views the guidance reiteration favorably given the company’s history of roughly in-line results."
The reverse split (1:10, effective June 17) is a mechanical fix for Nasdaq compliance, not a signal. Low conviction name, but the r/r is interesting if you believe the restructuring actually focuses capital allocation. Light catalyst path near term — need to see Q2 print.
Trimming the tree, but not everyone sees dead wood. Truist cuts PT to $87 (Hold) on lower services revenue and higher interest expense. Bernstein and Citizens also trimmed — to $96 (Outperform) and $120 (Market Outperform) respectively — but they see the Q2 beat and fiber divestiture as clearing the decks for a tower leasing recovery. Stock at $76.93, yielding 5.73%. The bull/bear tug-of-war is about rate of change in free cash flow, not just AFFO.
Truist lowered 2026-2028 estimates 1-3% below consensus, citing the services headwind and elevated rates. Meanwhile, the company delivered on strategy: ground lease buyouts, cost cutting, and a 73.7% gross margin. New edge compute initiatives get a nod as potential future return drivers.
The divergence is the story. Truist sees the rate burden and services drag as structural near-term. Bernstein/Citizens think the leasing cycle is gaining traction after the fiber exit. The $87-$120 PT range is a 13% to 56% upside band — tells you the market's pricing in a lot of uncertainty on the recovery slope. Dividend yield anchors the downside, but PMs should watch lease renewal rates and interest expense trajectory more than headline AFFO.
WEDBUSH STAYS AT OUTPERFORM / $109 INTO Q2. Not a massive call but the setup is solid. Stock at ~$101 gives you ~8% upside to PT plus a 12% FCF yield — that’s good r/r for a name with a clean narrative.
Thesis is simple: margin expansion continues, Domains chugs along, and Airo is starting to show in Applications & Commerce bookings. Last quarter’s early Airo AI Builder success + expected step-up in marketing spend means the growth-vs-margins balance is the live wire. Biggest watchpoint: bookings growth slowed to 9% (negative 1% in core), and management says it converges with revenue over H2. PMs will want to see that trend line flatten — or at least not worsen — Thursday after close.
“After bookings growth slowed to 9% with negative 1% in core business, management expects the metric to converge with revenue growth over the remainder of the year.”
Wedbush also flags the upcoming investor event (likely new midterm targets) and ongoing buyback as catalysts. Vibe coding tailwind is real but early — Wedbush explicitly calls out “positioning as vibe coding grows” as a key debate. Bear case: UBS started at Neutral with a $100 PT, citing valuation after the run. Options flow showed a spike in puts, though single print noise is hard to read. No one’s pounding the table, but the FCF yield is hard to ignore.
BMO doubles down on the top pick call despite the merger overhang. Stock at $35.81, trading 14.5x their 2027 EBITDA — cheap for a business that just printed a beat-and-raise quarter (Q2 EPS $0.47 vs $0.43 consensus, revenue $977.9M vs $859.1M). The pressure comes from the SOLS merger announcement and the broader semis/AI sentiment spillover into electronic chemicals, but BMO sees fundamental strength in electronics and semis end-markets. At ~20x EV/EBITDA and 60x P/E (nosebleed multiples, yes) but 14.5x forward EBITDA, the r/r is skewed to the upside if the thesis plays out.
"We see multiple ways for Element Solutions shareholders to benefit from current levels."
BMO’s $50 target implies ~40% upside. The Q2 print shows the operating momentum is real — the question is whether the market will look past the M&A noise. Not a name we'd fade here.
Wells Fargo starts Overweight with $18 PT (27% upside from $14.13, near 52-week low of $13.32). Thesis: post-BLD close, execution is the game now. Roofing looks closer to a fundamental trough, but the Kodiak LBM side is still in the penalty box. Stock down 39% in six months – that's a lot of pain priced in.
Blockquote from Wells: “Investor skepticism based on one-way feedback remains elevated, which reduces some risk by setting a lower bar for performance.”
That’s the key line. Low expectations mean beats are easier. BLD is a high-quality asset, QXO inherits a solid integration platform – but the macro backdrop (rates, housing) hasn’t been kind. KeyBanc cut PT from $32 to $28; Benchmark still at $50. Wide dispersion tells you uncertainty is high, but the setup is asymmetric. If they execute, the re-rate could be violent.
BMO STICKING TO THEIR GUNS AT $100 — STOCK AT $48. Outperform reiterated. The call hinges on engagement inflecting into summer and the trough valuation argument. Peak concurrent users up W-O-W for 6 of last 8 weeks. Age verification headwinds from Q1 are "appropriately reset" in estimates. The firm sees scalability in new games like +1 Speed Keyboard Escape.
"Shares trade near trough levels at approximately 14 times the firm’s fiscal 2027 adjusted EBITDA estimate."
That's the whole pitch. 14x forward EBITDA for 53% revenue growth this year — your classic "growth at a deep value discount" setup, but the stock is DOWN 60% over the last year for a reason. Execution risk on the age verification rollout and the Kids/Select account structure is real. Barclays at $60 (Equalweight) and Raymond James at Market Perform suggest the Street isn't ready to step in front of that story until bookings actually trough in Q3.
Earnings on July 30. Options market pricing a 14% swing. That's the make-or-break moment.
Clear Street throws the flag at $161 — 72% upside from here, and they’re not messing around. ATEX already up 329% YTD (pulled back 11% last week), but the firm sees this as just the opening act. The thesis: utilities are scrambling to replace dying analog networks, AI data centers and EV charging are crushing the grid, and Anterix holds the only nationwide licensed 10 MHz band for private LTE. That’s a $3B pipeline.
“Utilities need to replace obsolete analog networks, which drives a $3 billion pipeline. Anterix holds nationwide licensed spectrum to power the upgrade.”
1. Analog obsolescence – forced upgrade cycle, not optional. 2. New demand vectors – AI data centers, EV electrification, wildfire hardening. Every utility is spending. 3. Satellite optionality – active trials with Lynk Global for device-to-device and satellite backhaul. Adds a second TAM beyond utilities.
Recent FQ4’26 earnings beat was a catalyst: EPS $0.98 vs. consensus -$0.54 loss. Revenue $1.96M vs. $1.47M estimate. Spectrum monetization and new product traction. At $93.76, r/r is asymmetric if the pipeline converts. Risk? Execution on utility adoption timelines — but the rate of change in grid modernization is accelerating, not decelerating.
Needham sticks to Buy + $600 target after Verizon’s Google dark fiber win. Stock’s already up 316% in the past year — $377 now, so the PT implies ~60% upside. The $1B+ Verizon/Google project should feed CIEN >$100M in sales across FY27-28. That’s real revenue, but the bigger narrative shift is Verizon entering a market Zayo/Lumen dominated. This opens a new channel for CIEN’s long-haul kit — rack modems, line systems, multi-rail amps (the latter >35% of equipment revenue via hut upgrades).
“Verizon’s entry into the dark fiber market marks a major strategic shift.”
Rate of change is what matters here. The deal isn’t a surprise (Verizon signaled it), but the follow-on pipeline matters — Verizon expects more similar deals. Separately, Stifel flagged that AI infra companies (including CIEN) remain SUPPLY-CONSTRAINED, not demand-limited, heading into June/July prints. That’s a bullish setup: beats-and-raises are the path of least resistance. However, the $2.875B zero-coupon convertible note closing is a potential overhang — dilution optionality, though zero interest means cheap leverage. If demand keeps accelerating, the convert is just fuel. But at 316% Y/Y, positioning is crowded. Not sure we can read too much into one analyst reiteration — the real signal is whether Verizon’s dark fiber push becomes a repeatable hyperscaler gateway. Early days, but the narrative is shifting from “optical recovery” to “AI infrastructure backbone.”
RBC staying bullish after a clean 2Q26 beat. They're lifting PT to $227 from $207, Outperform. Stock already up 13% in the past week and 30% YTD – market pricing in the double-digit FFO growth narrative.
The core thesis: MULTI-YEAR DOUBLE-DIGIT CORE FFO/SH GROWTH IS VISIBLE. Management explicitly guided 2026 to $8.15-8.20 (+10% CC) and said similar visibility extends into 2027. The Blackstone NoVa assets trickle in NOI through 1H27, and asset recycling (new stabilized hyperscale fund) is a 2H26 story.
"Double-Digit Core FFO Growth Visibility Extends Into 2027; DLR posted 2Q26 results ahead of consensus on all key financial metrics."
That's the whole bull case in one line. Not much to argue with on the operational front – record leasing, CapEx turning into cash flows, hyperscale demand unrelenting. Bears would point to the stock already pricing a lot of this at $194 and the eventual supply glut risk, but that's a 2028 conversation. For now, the rate of change is your friend.
Verdict: battleground name that’s been cut in half — but the setup is still messy near-term. Piper Sandler (OW, $250) holds the bull flag, pointing to core growth drivers and credit monetization, even as Q2 checks lean slightly negative. The stock at $223 is down 61% from its 52-week high, and the analyst community is split: UBS also at $250 Buy, Cantor Neutral at $200, Macquarie cut to Neutral at $190. The divergence is about timing on the AI monetization and sales retraining hangover.
“Investors need to see early evidence that sales changes were a one-time issue, that reduced AI pricing drives usage and volume, and that customers pay for AI tools after the free trial ends.”
That’s the near-term hurdle. The Q1 reversal (net new ARR growth falling below constant currency revenue growth after a long streak) broke the narrative. Until we see that curve re-steepen, HUBS stays a show-me story — but at 1.5x forward revenue (vs 8x+ at the peak), the r/r gets interesting if AI usage actually converts.
Baird staying constructive into the Q2 print (Aug 3). Reiterates Outperform, $200 PT. Thesis is straightforward: 12th consecutive quarter of accelerating revenue growth, upside in both US Commercial and Government, and estimates moving higher. They see the recent pullback/digestion as a buying opportunity in what they frame as a growth asset that's still growing into its valuation.
"Baird projects that 2026 and 2027 financial estimates continue to move higher. The firm is anchoring to $8 billion of free cash flow in 2027 as a reasonable upside case, compared to its current estimate of $6.7 billion."
The PEG ratio of 0.48 is the bull case anchor — the 147x P/E is the headline bogey everyone's staring at. Options market pricing 8-10% implied move for the print. Rackspace partnership for regulated AI is a nice narrative add but not a needle-mover this quarter.
Verdict: Citizens sees RKLB at an inflection point – Neutron first flight by year-end, Iridium spectrum + recurring revenue, and broad-based launch demand. Stock down 24% over six months to $66.94, but PT stays at $130 (94% upside). Key risk is execution on Neutron manifest, but the payoff if it fires is a floor-valuation reset.
Channel checks with component suppliers show launch volumes exceeding internal expectations across legacy and emerging providers. That demand backdrop is real. The $266M Space Force contract (12 launches, 6 more optional) adds backlog visibility into late-2026.
The $130 PT implies 63.5x CY27E EV/revenue – a massive premium to the A&D peer mean of 9.7x and even the high-growth peer mean of 14.8x. This is not a value call; it's a binary de-risking bet on Neutron and Iridium closing.
"Booked Neutron launches and a first flight on track for end of year suggest the program is approaching a key inflection point. A successful debut followed by execution against the manifest should raise the floor valuation for the stock by de-risking the platform."
Iridium acquisition (announced June 29) adds L-band spectrum, an operational constellation, and high-margin services revenue – a smart asset grab that shifts RKLB's revenue mix toward recurring. But the stock's already pricing in a lot of that. Watch the Neutron launch date – any delay and the 24% six-month drawdown gets worse before it gets better.
Needham stays at Buy / $83 after the print, but the setup is more nuanced than the headline beat suggests. Revenue crushed — $258.7M vs $95.3M consensus — but that was driven by pass-through fit-out services (one-time, low-margin). Strip those out and base rent aligned with estimates; Bitcoin mining was a modest beat. The real signal: HPC NOI margins hit 91%, above the mid-80s guidance, bringing lease economics in line with public peers. That’s the kind of rate-of-change that matters for the re-rating narrative.
“The company’s high-performance computing net operating income margin reached 91%, above its previous guidance of mid-80s.”
The expansion story is alive. Management is in advanced talks with two existing tenants for 250MW total (100MW at Polaris, 150MW at Delta Forge). Needham is now modeling 150MW of near-term Delta Forge expansion into its estimates. The stock is down 12% last week after a 149% year — feels like a positioning reset, not a thesis break. The r/r here is about execution on that 1.7GW pipeline, not the Q beat.
DA Davidson sticks to Buy with $225 PT — stock at $177, $6.3B market cap, and TTM revenue just crossed $1B for the first time. The firm hosted a webinar with VP finance Bill Schroeder post Q2 and came away more constructive. Thesis: premium tier adoption is accelerating, new offerings are adding juice, and the growing unit portfolio is a durable monetization runway. Not cheap at 24x 2027 FCF (consensus PT range $185–260), but the rate of change on revenue and ARPU is positive — and that’s what matters for a growth name in a risk-on tape.
“We see multiple avenues for AppFolio to monetize its growing unit portfolio.”
Q2 revenue printed $281M (+19% YoY), a beat vs consensus $277M. DA Davidson’s revenue est is already above Street — they see upside even from there. The key driver: accelerating adoption of the premium tier, not just unit adds. That’s a mix shift story with expanding margins. Stock is still off ~20% from its 52-week high, so the re-rating is not baked in. PMs watching for sustained execution into H2 — if premium attach rates hold, this thing grinds higher.
BE — Paradigm shift, not a quarter. Revenue doubled in a year ($1.065B, +166% YoY). Full-year guide raised to $3.9-4.2B. Customers are cancelling turbine and recip engine orders for Bloom. Power availability is now a direct AI revenue multiplier — 1GW DC = $12-24B annual revenue. The $25B Brookfield shelf gives them scaling fuel. This is the most important energy infrastructure print of the year.
KLAC — Broadest equipment cycle confirmation in years. Raised 2026 WFE to low-$150B with a 2027 planning case near $190B. Advanced packaging revenue forecast raised to +70% (was high-50%). Backlog at $12.5B with lead times 18-24 months — they're taking orders for late 2027. If they could build more, they could ship more. Supply constrained, not demand constrained.
STX — Demand-constrained, price-led phase. Mass-capacity storage has contracted visibility through 2029. Nearline exabytes allocated into 2028. Price-per-exabyte grew ~10% YoY in June, ~20% implied in September. Customers paying above contract for incremental exabytes. HAMR at ~40% of nearline run rate. Mozaic 4 ramping with 2 largest CSPs. Pricing power accelerating.
GLW — Scale-out optical exceptional before any scale-up or photonics contribution. Optical Comms $2.07B (+32% YoY), enterprise +65%, GenAI nearly doubled. Multi-year agreements with META (up to $6B), AMZN, and NVDA partnership de-risk capacity expansion. Stock down 12% post-print — positioning issue, not fundamental. Expectations ran ahead of reality.
SKHY — Record profit levels but expectations were even higher. Q2 operating profit 60.5T won (miss vs 64.2T est). HBM4 mass shipments began. LTAs with ~10 customers. Memory demand visibility into 2026 remains firm. Korea market crashed 15% — positioning unwind, not fundamental. Q3 pricing data (TrendForce raising DRAM to 13-18%) will resolve the debate.
NVDA — Vera Rubin Ultra uses 576 GPUs across 8 racks. CPO switches in mass production. 800 VDC validated. Network content per GPU increasing with cluster size. SSI shift from TPU-only to Vera Rubin is a significant platform commitment. Circular finance narrative is the key risk vector — is Nvidia Fannie Mae or are these real end-customer deployments?
ASML — China mass-producing immersion DUV tools (5 units 2026, 20 by 2027). Performance gaps remain far below displacement thresholds. Strategically significant, commercially irrelevant near-term. No EUV threat. Chinese WFE is additive, not substitutive. WFE path to $300B+ by 2030 unchanged by this news.
AMD — Warrant deals with OpenAI and Meta function as compute rebates (up to 105% discount for OpenAI). Structurally lowers effective GPU pricing to buy deployment share. Margins under pressure. Fubon pessimistic on CoWoS allocation. MI series competing but facing warranty cost headwinds. Inference shift may not favor Instinct line.
MU — Seagate independently confirmed NAND pricing tightening. Edge IoT revenue +20% YoY attributed to "ongoing tight supply conditions." DDR4/DDR5 pricing strong. Memory supply shifting to long-term contracts, reducing spot volatility. CXMT emerging as fourth DRAM force absorbing conventional demand — MU benefits from overall tightness but faces capacity competition.
SANM — Q3 operating margin 8.0% supported by high-margin pre-production engineering. Core coms and cloud ex-ZT grew ~33% YoY. Cerebras confirmed as additional platform. Heavy pre-production engineering obscuring true cash conversion. Underlying demand robust.
AVGO — Hyperscaler custom-silicon program counts accelerating. Each hyperscaler pursuing multiple workload-specific designs. Faster lane rates (200G, 400G) increase switching silicon and SerDes complexity. Larger clusters require third optical switching layer — 50% more content per GPU above 130K. VMware integration adds software revenue.
MRVL — Dragged down by Corning's 12% drop — positioning unwind, not fundamental. Faster lane rates increase DSP and SerDes content. Larger clusters require more optical switching. Custom silicon engagements with hyperscalers expanding. Optical interconnect demand accelerating.
VRT — 800 VDC adoption driven by both compute (GPU density) and generation (native-DC fuel cells like BE). Entire power train from source to rack being redesigned. Legacy AC-centric UPS faces displacement risk. Vertiv has announced dedicated 800 VDC architectures. Data center power infrastructure cycle accelerating faster than silicon cycle.
ETN — 800 VDC requires new power conversion and distribution equipment. Opportunity larger than single converter — includes entire power train. Data center power infrastructure is the binding bottleneck for AI growth. Electrical components critical for both AC and emerging DC architectures.
AMKR — Expected to crush earnings as AI CPU boom beneficiary. Early advanced packaging ramp driving structural margin expansion — FY26 GM at 17.5% (two years ahead of FY28 target). US facility ramp is a headwind but revenue scaling offsets. KLA raised AP revenue forecast to +70% — OSATs with differentiated AP capabilities benefit.
CXMT — Surged on STAR Market IPO debut. Nomura initiated with Buy and CNY116 target, forecasting revenue and net profit compounding at 63% and 74% CAGR 2026-2028. Expanding wafer capacity at 95% utilization. Absorbing conventional DRAM demand neglected by incumbents focused on HBM. Reliance on foreign equipment remains geopolitical risk.
GOOG — GCP at 82% YoY (~79% ex-Wiz) is the first hard evidence the capex cycle is clearing an ROI hurdle. Supports the bull case that AI infrastructure spending is demand-backed, not speculative. FCF turned negative $6B after capex guide increase. If AMD gains more CoWoS capacity, GOOG will need Intel foundry capacity by 2028.
ORCL — Power availability directly determines AI revenue realization. Bloom delivered fuel cells for an Oracle data center within 55 days — time to power is time to token revenue. 1 GW DC can produce $12-24B annual revenue. Market values entire ~$630B AI compute backlog at zero after the stock round-trip. Circular financing debate impacts valuation multiple.
Hearing Bloom customers are cancelling not just reciprocating engines but also combustion turbines from CAT and CMI. If true, substitution risk for large-turbine OEMs (GEV) is higher than the market prices in. Bloom's 800 VDC native-DC value prop is validating the entire electrical infrastructure cycle.
Word is GPU spot rental pricing is at least 2x contracted rates. Hyperscalers are underearning — anyone who signed a GPU compute contract in 2024-25 is overearning. Morgan Stanley modeling GPU cloud businesses generating 70% incremental margins and 30%+ ROIC. Operating cash flow growth accelerating from 31% in Q1 to 50% in Q2 per one sell-side model. Contract repricing is the miss.
Channel checks suggest SK Hynix's Q2 miss is a positioning artifact — one analyst called it "record profits but expectations were even higher." The real debate is whether HBM4 pricing is peaking. TrendForce raising Q3 DRAM to 13-18% Q/Q argues against. Korea selloff looks like a positioning unwind, not a fundamental change.
Hearing Sam Altman admitted OpenAI paused training after the Hugging Face security incident — "the first security incident I felt very viscerally." Multiple zero days chained together. He's now talking about pacing AI development. If true, this introduces a new risk vector for the entire AI compute demand thesis.
Word is Kimi K3 at 2.88T total / 104B active parameters makes local frontier inference impractical for individuals. Recommended 64-chip supernode. Running K3 on an NVL72 without stuffing the batch can cost 5x+ vs an API provider. Negative read-through for Dell's enterprise GPU server narrative — GPU rental structurally more attractive than owning.
Hearing OpenRouter data is capturing sub 1% of global inference usage. Do not use it to claim Chinese models have surpassed US models. The data is not representative.
Channel checks China's DUV tools replicate 15-year-old ASML technology. Performance gaps in throughput, overlay, and reliability remain far below commercial displacement thresholds. BofA calls it a "modest threat." No EUV threat. Chinese WFE is additive to global capacity, not substitutive for ASML's monopoly.
Word is CXMT using Chinese lithography tools for some layers. If they can commercialize without Dutch/Japanese equipment, the self-sufficiency narrative accelerates. But current reliance on foreign photoresists and equipment remains the binding constraint. Big milestone in 2030, not 2026.
Hearing AMD's warrant deals with OpenAI and Meta function as compute rebates up to 105% discount for OpenAI. Structurally lowers effective GPU pricing. If true, AMD is buying deployment share at the expense of margins. Partnering with Core Scientific for 500 MW (expandable to 2.5 GW) suggests they need capacity fast.
Word is Both OAI and Anthropic have been holding back frontier models. Scaling laws are holding. Base case should be a significant capability leap and ARR growth acceleration into year-end. The variable is US government pursuit of anti-distillation.
Channel checks suggest SSI's shift from TPU-only to Vera Rubin marks a significant compute-platform commitment. $5B investment gives them 10x compute scaling. The $750B deal pipeline ($500B Hynix, $250B OpenAI, SSI, Thinking Machines) crystallizes the circular finance concern. Market split on validation vs vendor-financed bubble.