Good morning.
Futures bid, tape green, AI infra news flow doing the heavy lifting. Brent +1.8% on US-Iran tail risk — matters for the European economics. Asia closed with Korea memory complex leading on a Goldman re-rating.
Memory got re-anchored, not nudged. Goldman lifted SK Hynix 2027 OP +22% to $266B and Samsung 2027 +21% to $352B, pulling the HBM peak into 2028+ and lifting aggregate memory capex — TSM, MU, 000660.KS, 005930.KS all trade off this.
Europe is finally real on the ground. SoftBank committed €75B for 5 GW of AI DCs in France by 2031, Schneider as lead industrial partner at Dunkirk — first concrete test of whether Europe breaks its grid-constrained positioning.
NVDA crossed 20% of TSMC revenue as Customer A, displacing AAPL. Jensen teasing Feynman and N1X at GTC Taipei; combined with Win10 EoL and MacBook Neo (5M → 10M), the multi-vendor Arm+AI PC window just opened. New bottleneck named: inference $66B → $292B by 2029 at 45% CAGR — capacity tightness is the four-year supply story, not a 2026 digestion risk.
The GW race is up for grabs. META may bring the most incremental hyperscaler capacity online in 2026, potentially ahead of OAI/Anthropic's 6 GW EoY target — pair that with "demand ahead of buildout" universal across every channel check and constraint is the dominant tape driver.
Watch list: SMCI still soft, LITE/COHR/AAOI parabolic with technical weakness (40-week MAs the late-summer target), APLD stacking the ARW feed quietly.
We'll hit up NVDA and the memory complex (MU, TSM, SK Hynix/Samsung) first, then get to META, the optical names (LITE, COHR, AAOI), and the AI infra adjacencies (VRT, CRWV, Schneider proxies).
Verdict: Six buy-side initiations in two days and the stock can't even hold $200. Street is all in, tape isn't — that's the trade. The action tells you the market is worried about something. The notes tell you it shouldn't be. Pick a side.
The set-up is unusual: CBRS IPO'd at $185, ran to $386 on day-one euphoria, and has now given back 48% to $201. Down 5.75% last week. Into that weakness, the Street dropped a coordinated launch — SEVEN initiations, no sells, all Buy/Outperform/Overweight. Nobody wants to be the bear on a wafer-scale AI pure-play with an OpenAI pre-pay, but the price action says the market is pricing execution risk, customer concentration, or both. Worth noting the OpenAI deal was disclosed in January and the AWS deal in March — so by the time these notes hit, the catalysts were 3-5 months stale. Tape's not buying it yet.
Seven initiations, $250-$300 PT range, cluster sits $280-300. All buy-equivalent ratings. Consensus roughly $290 (~44% upside from $201).
Bull case (steelmanned): CBRS is the only company commercially shipping wafer-scale silicon. The OpenAI deal is $20B+ with 750MW committed and an OPTION for 1.25GW additional — that option alone, if exercised, is a re-rating event. AWS is the optionality nobody's modeling. Fast-inference TAM is exploding (Mizuho sees 291% CAGR to $550B by 2030; Barclays calls it a $300B end-of-decade market). The WSE has orders-of-magnitude more SRAM and memory bandwidth than any other AI processor, purpose-built for the low-latency, high-token-throughput workloads that define the agentic era. 76% LTM growth. Profitable on LTM basis. Only supplier shipping to OpenAI under a pre-pay.
"The compute cycle pivots from training to inference, where speed dictates output value." — Wedbush
"Unique chance to invest in an AI processor company with a first-mover advantage against NVIDIA." — Morgan Stanley
Bear case (steelmanned): $44B market cap on $510M revenue. P/E 225x. That's the bear case in two numbers. The stock got cut in half for a reason — either the market doesn't believe the OpenAI conversion math, doesn't trust AWS materialization, or is worried about a single-customer architecture story. If OpenAI doesn't exercise the 1.25GW option, the 2028 revenue base for the $300 PTs looks aggressive. Cerebras' gross margins on a brand-new fab process at sub-scale economics are unproven at $11B revenue. And the 750MW deal — while large in headline — is small relative to total OpenAI compute, meaning this is a niche supplier, not a NVDA replacement. The fact that Arm/SoftBank allegedly approached pre-IPO and got declined tells you even the strategic buyers weren't sure the unit economics worked at scale.
Not much new in the facts — OpenAI deal (Jan), AWS collab (Mar), both already disclosed. What IS new is the coordinated Street launch and the magnitude of the bull case being uniformed across seven desks. That's rare and tells you the analysts have had prep access. The incremental info: Needham flagged that "should OpenAI exercise its option for the additional 1.25GW or Amazon take delivery of any meaningful amount, we see significant upside to our current model" — meaning even the Street's own numbers are conservative vs. contract option value. Mizuho's 4-pillar framework (AI capex beneficiary, fast-inference lead, hardware moat, 122% revenue CAGR 2025-2029) is the most detailed bull case on the tape.
CBRS is the cleanest pure-play on the inference pivot narrative — and if the bulls are right, the read-through to NVDA, AVGO, MRVL is "inference TAM is bigger than training TAM, not smaller." The Street consensus framework literally uses NVDA and AVGO as valuation comps. If CBRS holds $200 and the 1.25GW option story starts getting priced in, it becomes a sympathy leader for the entire AI complex on inference days. Conversely, if it breaks $200 on no incremental news, it drags the whole "AI 2.0" cohort — particularly any name trading on TAM-not-revenue stories. Watch the tape at $200. That's the line.
VERDICT: STREET IS CONSTRUCTIVELY BULLISH BUT RANGE IS TIGHT — $22.50-$26, NO SELLS, WITH THE SPREAD DRIVEN BY EXECUTION CONFIDENCE. BXDC is the data center trade dressed up as a "wait and see" — the asset class is white-hot, Blackstone's sponsorship is gold-plated, but you're buying a blind pool at a 9% premium to forward NAV with zero assets on the books. Stock at $21.84, ~5% off 52-week high of $22.90, $1.91B market cap. Spreads on first deployment news.
Consensus PT sits in the $23-24 zone with one outlier bull. Three Buy/Outperform (Barclays $24, RBC $24, Bernstein $26), two Neutral (BofA $22.50, JPM $23), one Market Perform (BMO $23), and DB at Buy $24. No sells on the tape. Bernstein's $26 is the high; BofA's $22.50 is the floor — that $3.50 spread (16% dispersion) is all about how much credit you give Blackstone to execute on a $25B pipeline. RBC tagged it "Speculative" risk despite the Outperform, which is the honest read: this is a momentum/sponsorship trade, not a yield play yet (no dividend).
BULL CASE — The setup is genuinely unique. Blackstone has thrown $200B+ at data centers since 2018; BXDC gives public investors a vehicle to buy stabilized, already-leased hyperscale DCs without taking development or power risk. The company reviewed $25B of potential deals at IPO — 13% deployment to hit $3B+ of total capital at 40% LTV. Bernstein sees progress "as early as this quarter." Tenant quality is investment-grade hyperscalers on triple-net with 2-3% escalators. First-mover advantage in a $90B+ TAM with limited public comps outside DLR and EQIX. JPM — typically the most cautious bank on the Street — called it an "attractive opportunity" for AI/DC equity exposure with "significantly reduced" development risk. That's a meaningful concession from the BX crowd.
"Management, supported by Blackstone, has a proprietary pipeline of deals and is a first mover in the AI/HPC stabilized data center marketplace." — JPMorgan
BEAR CASE — You're buying a SPAC-like structure with a 100bps tiered management fee and no cash flow. Trades at 22.9x 2027 AFFO — basically in line with data center REITs (23.3x) but at a massive premium to net lease comps (13.1x), and those comps have actual buildings generating rent. BMO's balanced read is the right one: industry tailwinds are real, but near-term the path is binary on capital deployment pace, cost of capital, and whether Blackstone actually feeds the vehicle good deals vs. keeping the best ones in-house. The external management drag is real — that's the single biggest overhang on the multiple.
"Strong industry fundamentals, premium tenant quality, and 2-3% escalators as supporting attractive long-term growth... balanced by near-term uncertainties around capital deployment pace, external management structure, and reliance on a favorable cost of capital." — BMO Capital
The 7-firm initiation wave is the story — this is the Street formally pricing a new public vehicle, not an update to an existing thesis. The $25B pipeline disclosure (vs. $90B+ TAM) and the 40% LTV math are the numbers to anchor on. Trade is in price discovery mode — shares are ~5% off the post-IPO high, up 10% in 6 months, and the analyst PT cluster suggests ~5-10% upside to consensus with a 19% upside to Bernstein's bull case. Watch for first acquisition announcement — that will re-rate the multiple or expose the bear case.
This is the public-market proxy for the AI infrastructure buildout that institutions have been funding through private markets for 18 months. If BXDC can demonstrate 2-3 accretive deployments at attractive cap rates, expect capital to chase it as a substitute for the private fund queues. Pairs naturally long BXDC / watch DLR and EQIX for sentiment beta. Risk: if hyperscaler capex guidance softens in 2H prints, this name catches a falling knife with no revenue cushion. Tighter-than-typical PT spread tells you the Street is mostly waiting on the same catalyst — first deal close.
VERDICT: RBRK IS THE TAPE RIGHT NOW. Stock up 34% in a month, prints a clean Q1, raises FY27 guide by MORE than the beat, and 11+ firms are racing to lift PTs. The trade is crowded long. The only cut on the Street? Jefferies on cloud growth. That's your bear case.
PTs are clustering in a tight $87-$95 band post-print, with Guggenheim the high outlier at $110 and the consensus range now $70-$126 (per Guggenheim's note). Nearly every firm raised:
BULL (Scotiabank, BMO, Stephens, Guggenheim, et al): This is a best-in-class cyber resilience franchise with 30%+ subscription ARR growth potential in an upside scenario. Mythos release is already driving customer conversations in the same pattern CRWD and PANW saw. Hardware cost fears? Immaterial. Market share consolidation in backup is a tailwind. Stock trades at a discount to high-growth software peers on FCF basis — that gap closes as they execute through FY27. Agentic cyber resilience is a free call option. Identity went from 0 to $50M in ARR in 12 months — that's the product velocity that justifies a premium multiple.
"Rubrik has a best-in-class product in the mission-critical area of cyber resilience and could benefit from the coming release of Mythos, which is already driving an increase in customer conversations." - Scotiabank
"We expect healthy, sustained demand for Rubrik's core offerings and new solutions." - BMO
BEAR (Jefferies, plus the valuation hawks): Cloud Net New ARR grew just 5% YoY. That's the growth engine and it's barely ticking. Non-Cloud growing sequentially for the first time in 13 quarters is a nice tell on the on-prem side, but cloud is where the multiple lives. Stock up 34% in a month — priced for Mythos to be a hit, for cloud to reaccelerate, for Identity to keep ramping. Multiple at 9-10x revenue, 46x FCF per KeyBanc's math. Plenty of execution already in the price.
RBRK is the cleanest data security / cyber resilience tape right now. Read-through to CRWD, PANW, S, OKTA — the whole cyber complex has a bid into this print. The Mythos comparison to CRWD's platform evolution and PANW's product cadence is the narrative the Street wants to own. If Mythos lands well at the release, this is a multiple-expansion story on top of the execution story. If cloud growth stays at 5% net new, the Jefferies cut looks prescient. Watch the F2Q print for cloud re-acceleration — that's the binary.
Positioning note: 34% in a month, every sell-side shop bullish, only one cut on the Street. This is a momentum tape, not a value setup. Size accordingly — the easy money's been made.
The thesis is back. ServiceTitan just printed a beat-and-raise quarter that re-anchored the bull case on three vectors at once: growth reacceleration, AI monetization emerging as a real line item (not a slide), and enterprise + PE as a durable engine. Street responded with a broad PT cluster shift to $100-125. Stock's at $74 — that implies ~35-70% upside to consensus, and even the lowest published target ($67) is roughly in line. Setup is clean.
Q1 FY27 was the cleanest print in quarters. Revenue $268.8M (+25% YoY vs +19% guide, reaccel from +21% in Q4 FY26). EPS -$0.24 vs -$0.29 consensus. USAGE REVENUE ACCELERATED TO 29% — highest growth rate in nearly 4 years. Subscription revenue beat of 3.5% was the largest in TTAN history. GTV +23%. LTM revenue ~$961M, gross margin 70-71%, non-GAAP op margin >15%.
Guide: FY27 revenue raised to 18.1% growth (from 16.5%) — and the RAISE EXCEEDED THE Q1 BEAT, which is the line PMs should care about. Q2 also above expectations. Incremental margin assumption bumped to 29% from 25%.
Six firms in our coverage set raised PTs post-print. New cluster: $100-125, all Buy/Overweight, with $67 as the only notable low outlier. Standouts:
New:
"Business is firing on all cylinders. The rollout of Max is progressing well and ServiceTitan continues to gain traction with enterprise customers." — KeyBanc
Bull case: This is a clean compounder re-emerging. 25% growth, expanding margins, guide raise > beat = under-shipped Q2. Max doubled in size in one quarter. Virtual Agent is no longer a deck slide — it's converting to revenue. Enterprise + PE is the engine that scales regardless of SMB churn. At 6x CY27 sales on a path to 20%+ growth and improving op margins, the multiple compresses fast on any beat. Piper calling it a top-2 pick, KeyBanc a top idea — those aren't throwaway labels.
Bear case: The 25% print was juiced by an extra business day and weather dynamics (Needham's read; underlying closer to 24.6%). The 6x sales multiple isn't free — TTAN is still GAAP-unprofitable, and a single digit deceleration in usage revenue would compress the multiple fast. The $67 PT outlier represents a cohort that doesn't buy the AI-as-feature narrative and views vertical SaaS as structurally threatened by general-purpose AI agents. Fair concern in theory — but mgmt's "orchestration layer" framing is the cleanest answer to that bear case we've heard from a trades-vertical name.
This is the cleanest data point for the vertical SaaS AI-aggregation thesis in weeks. The argument: GPT-style models are commoditized, but the workflow data, customer relationships, and integration depth in trades/homeservices make TTAN the surface where AI gets applied — not the company that gets disintermediated. If TTAN keeps posting these prints, the cohort (think: small/mid-cap vertical SaaS with proprietary workflow data) gets a multiple re-rate. Watch the read-through to other trades-adjacent and SMB-vertical names this week.
5 PT hikes in the last 24 hours. Consensus PT cluster sits $240-300, with Guggenheim ($400) the high outlier and RBC ($190) the lone bear. Specifically:
Bear camp: RBC still at $190, citing peer multiple expansion as the reason for the raise. The capex is a bloodbath — Guggenheim modeling $75B FY27 and $85B FY28 capex. Levered FCF was -$24.7B LTM. FY27 is the trough before FCF improves in FY29. That's a long time to wait.
Bull case (steelmanned): This is a compute infrastructure story masquerading as a software company. OCI growth re-accelerating from ~84% to 89-92% on capacity adds is the main event. The $553B RPO means the revenue is largely already booked — you're underwriting execution, not new logo wins. Layoffs (~30K jobs, 18% of workforce) combined with GPU pricing power and capex efficiencies should drive operating margin expansion even as cloud gross margin compresses from IaaS mix. Trading at ~25x CY27 EPS (Barclays basis) for what Guggenheim calls a "Decade Stock" — full-stack offering, enterprise data gravity, cloud database migration. The 14% pullback is gift-wrapped entry ahead of a print Street models as a beat-and-raise.
Bear case (steelmanned): You're paying 37.8x P/E for a company burning $24.7B of levered FCF. FY27 capex of $75-85B means the trough FCF year is next fiscal. Industry-wide data center delays and inflation are real — Barclays flagged this explicitly. Cloud gross margin faces structural pressure from IaaS mix and depreciation ramp; only opex discipline (i.e., layoffs) is masking it. New CFO Hilary Maxxon's first call adds transition risk on a guide-the-street print. RBC at $190 implies Street is too generous on multiples once you adjust for FCF dilution from the $45-50B debt/equity raise.
The layoffs are the freshest data point — 30K jobs, 18% of workforce, explicitly framed as redeployment toward AI data center buildout. This isn't a cost-cut story; it's a mix-shift story. Pair that with TD Cowen noting the "AI backdrop appears more favorable compared to three months ago" and Oppenheimer's proprietary regression showing 100bps upside to OCI guide and 400bps upside to consensus, and you have a setup where the bull narrative is getting reinforced, not fading.
New CFO Maxxon on the call is incremental uncertainty but not a thesis-changer.
"Oracle is a 'Decade Stock' based on technology and execution." > — Guggenheim, Best Idea note
"Compelling risk/reward at current levels when considering expectations for Oracle to soon reach 30%-plus total revenue growth." > — Barclays, $240 PT
ORCL is a direct read on AI capex durability. The $75-85B annual capex is a tailwind for NVDA, AVGO, MRVL, MU on the compute side and for power/cooling plays (VST, CEG tangentially). The OpenAI $120B funding round Guggenheim cites is bullish for the whole AI infrastructure complex — ORCL is a leveraged way to play it with actual revenue visibility. If ORCL prints clean OCI acceleration and the guide holds, it takes pressure off the entire enterprise software cohort that took a beating on AI-disruption fears. RBC's $190 is the cleanest expression of the bear case on the comp set — watch if the gap closes post-print. Options market pricing ~12% move on the print; we're positioned for a beat-and-raise clearing the air on capex trajectory.
Verdict: Print goes from "priced for disaster" to "still a show-me story." ADBE prints Q2 Thursday (June 11), stock at ~$250, sits 40% off the 52-week high ($419.82) and 26-29% YTD (worse than software peers at -32% vs IGV -5%). Implied move is 8.7% per options — and ADBE has BEAT the implied move in 5 of last 8 prints. The setup is asymmetric: valuation already reflects the bad news, the narrative is "AI monetization stall," and any green shoot on ARR re-accel gets paid.
Consensus ARR $26.6B. Mizuho modeling +200bps lift from Semrush (closed end of April) for both Q2 and FY26 — so 10.2% YoY ARR guide is the bogey. Piper noted 9.9% revenue growth at the midpoint guide. Options market pricing 8.7% move. Burry added to the position, which is a fun color note but not a thesis.
Three brokers on the tape, split into a clear 1-2-1 pattern:
"ARR is within striking distance of re-accelerating growth on a quarter-over-quarter basis." — RBC Capital
Bull steelman: Cheap on every multiple that matters (15x P/E, 8.5x EV/FCF). Experience Cloud is a real $1B+ ARR business at 20%+ growth — three of them. Semrush closes the SEO/wallet gap. AI narrative pain is in the price.
Bear steelman (TD Cowen): 3p data is the only real read on AI demand, and it's decelerating into the print. If Firefly/Acrobat AI monetization is "minimal," the GenAI re-rating thesis dies. Low-beta value trap until ARR proves it can re-accel — and fading pricing tailwinds mean even in-line prints could disappoint.
Print is the catalyst. Q/Q ARR direction is THE metric — RBC needs re-accel, TD Cowen needs 3p data to stabilize above 2%, Mizuho needs the 10.2% guide to be credible into FY27. Cheap enough that a clean guide-plus-slight-raise gets a 10-12% move, which the options market is not pricing. Skew is long. Size the print, not the position yet.
Verdict: Bullish tape, but the AI story is still mostly TBD. Sell-side is leaning into the Ternus transition and a "deferred AI optionality" narrative — Wedbush's $400 PT is the ceiling, BofA/Evercore cluster in the $365-380 range, and the only real bear (UBS at $296 Neutral) is hanging its hat on App Store deceleration.
Stock at $302.52 after the WWDC print, sitting ~5% off the $316.94 52-week high. $4.43T market cap, up 51% YoY. P/E 37.6x — rich but not crazy for Apple when you back out net cash and treat services as a software multiple.
This was Tim Cook's last WWDC. That alone matters. Wedbush framed it as Cook "passing the AI foundation to Ternus" — a transition story the Street clearly wants to trade. The actual content: deeper Apple Intelligence across iOS/iPadOS/macOS/watchOS/visionOS, hybrid on-device + Private Cloud Compute architecture, a standalone Siri app with a chatbot interface, and — this is the one people will talk about — Apple's rearchitected foundational models now powered by Google's Gemini family. Licensed, not in-house.
BofA's framing is probably the cleanest read on what the Street is actually modeling:
"An important marker for Apple's ability to convert its installed base, silicon roadmap, privacy architecture, and App Store distribution into a differentiated agentic AI platform."
BofA also explicitly said WWDC alone won't settle the "is Apple behind on frontier models" debate. That's honest. Expect that debate to drag into iPhone 18 / fall print.
Bull (Wedbush, BofA, Evercore, Bernstein): The AI/services monetization isn't in the multiple. Wedbush's $75-100 of incremental value from AI is the most aggressive number floating around — and Wedbush is the high at $400. BofA at $380 is doing the work — 37x CY27E EPS of $10.29, in line with current multiple but with optionality baked in. Capital returns, edge AI leadership, new product cycles post-Cook. 26 analysts have revised earnings UP for the upcoming period. iPhone refresh + AI features = upgrade super-cycle narrative.
Bear (UBS at $296 Neutral, plus the Kuo smart glasses delay): UBS is the only firm with a PT below the stock, and it's not even really bearish — just cautious. Their hook: App Store revenue grew ~3% in May, decelerating from April. That's the real tell on the consumer. Kuo pushing smart glasses with display out to 2029 kills a near-term hardware catalyst. And the Gemini licensing is a double-edged sword — Apple gets capability, but it also confirms they don't have a frontier model story of their own.
Stock is near highs, up 51% YoY, multiple already at 37x. The bull case is largely a next year story — CY27 earnings, AI monetization in the fall iPhone cycle, Ternus execution. The bear case is this quarter — App Store is already decelerating, no hardware at WWDC, smart glasses punted. R/R from here is tougher than it was six months ago. PT cluster of $350-400 vs $302 = ~16-32% upside. Not bad, but you've got to believe the AI story accelerates from here, and Wedbush's $400 is asking you to believe a lot.
Watch the September quarter iPhone guidance and any color on Apple Intelligence attach rates. That's the data point that moves this from "deferred optionality" to "real monetization." Until then, this is a narrative stock trading on the Ternus transition and the hope that fall 2026 is the AI inflection.
Dog of the cohort keeps getting cheaper, and now the Street's racing down to meet the stock. ACN -34% YTD at $179, no relief in sight pre-print. Bull case isn't the story — it's the math: 12.5X NTM P/E, 10% FCF YIELD, 3.66% DIVVY. That's capitulation pricing for a compounder with a fortress balance sheet. Bear case is real though — AI tokenomics question, budget freezes, Truist's downgrade to Hold last week captured the structural anxiety. Setup is the FY26 exit rate, not the Q3 print. Everything hangs on whether that 2.5% organic / 5% M&A-included exit number firms or rolls over.
5 brokers touched ACN in the last few sessions. PT cluster has compressed hard: JPM $201 (OW, from $247), TD Cowen $258 (Buy, from $282), Truist $210 (Hold, downgrade), Stifel $270 (Buy), Goldman $270 (Buy). Range $201-270, median ~$258. Dispersion is the story — when PTs are this wide, no one has conviction. Buy ratings still dominate, but the price targets are doing the talking.
"The fiscal 2026 exit rate will be key to recalibrating fiscal 2027 growth views." — TD Cowen
That's the line. Until that exit rate prints with conviction, stock stays in no-man's land. We like the r/r here for a position-sized add into the print — 12.5X P/E AND 10% FCF aren't prices that last for compounders. But this isn't "buy the dip" yet. Need to see bookings hold and FY26 exit firm before we get aggressive.
Goldman just ripped the cover off this name, taking ROHCY to BUY from NEUTRAL and roughly DOUBLING the price target to JPY6,500 from JPY3,300. The trigger was FY25 results where mgmt lifted its FY30 AI/DC sales target to ¥100B from ¥30B, with SiC contributing ¥30B of that. GS thinks even that number is light — the higher-voltage trend in AI server power is creating a pull that's bigger than mgmt is willing to underwrite publicly. Halving the PT to a single number isn't an analyst flex, it's a statement: this is a different business than it was six months ago.
The pivot narrative is the entire ballgame here. Rohm wrote down the SiC biz by just under ¥200B in FY25 because the EV demand curve cratered, and the street spent 18 months treating them as a busted industrial. Now the same SiC capability — voltage tolerance, heat resistance — maps directly onto AI server power supply, where the demand is the opposite of the EV story. GS calls it the best operating leverage setup among the EV-to-AI diversifiers. Their model: ¥20B loss in FY26 (halved YoY), breakeven in FY27 — a full year ahead of mgmt's FY28 timeline.
"Rohm has the potential for earnings recovery with the most operating leverage among these companies [diverting EV parts to AI server applications]."
Stock's already done the work — +179% over the last year, +125% in the last six months. Unprofitable LTM with EPS at -$2.59. So this is not a "discover the value" trade, it's a "is the AI server power re-rating durable" trade. A doubling of the PT after a 179% rip is either Goldman catching up, or Goldman underestimating how much SiC content per server ramps as rack densities push 800V/1000V+. We'd lean the latter, but the easy money's been made. Fresh money wants a pullback into the JPY4,500-5,000 area (where the prior PT sat) for a better r/r — chasing into the print is buying a stock that's already priced for the FY27 breakeven.
Catch-up trade is mostly played. Valuation is the wall. Northland holds Market Perform after the Clearwater Forest Server CPU demo, acknowledging Intel is finally closing the gap to TSMC on process tech and to AMD in server CPUs. But the stock's already done the work — INTC +386% over the past year to $37.19, kissing the $38.99 52-week high. Hard to chase here on a "we're catching up" narrative when the tape's already paid for it.
"Intel is catching up with Taiwan Semiconductor Manufacturing Co. on process technology and with Advanced Micro Devices Inc. in the server CPU market." — Northland
Real alpha might be in the picks-and-shovels. Northland flags Veeco (VECO) and Camtek (CAMT) as packaging beneficiaries, with Alphawave IP getting a heterogeneous integration lift. Intel is dumping serious capex into packaging capacity and pulling in foundry customers — the firm says INTC expects "billions in annual packaging foundry revenue." That's a real number to anchor on, and frankly a cleaner way to play the Intel rebound than the stock itself near highs.
The bear case Northland won't say out loud: 386% in a year on a turnaround that still has years to play out. Multiple expansion did the lifting, not earnings. At ~$37 you're paying for a flawless execution path on 18A and foundry wins that haven't shown up in the P&L yet. R/R on the long here is thin — better to own the supply chain (VECO, CAMT) where the customer concentration risk is lower and the operating leverage is more obvious.
Mizuho lifts PT to $500 from $425 (prior $360) — Outperform maintained, stock at $342.93, so they're pricing in 46% UPSIDE. This isn't a multiple expansion exercise. Mizuho is rewriting the revenue curve: they now think Arm pulls in the $15B AGI CPU target (FY31) and is modeling closer to $20B. Earlier AND bigger. That's the message.
The real swing factor is AI ASIC. Mizuho sizes that TAM at $1T+ with a potential launch late 2026/early 2027. ASPs above $15K (10x AGI CPU), market 5-10x the CPU market. If even a sliver of that materializes, SOTP math gets genuinely interesting.
"Arm is expanding its AGI CPU platform with Oracle and ByteDance while RTX Spark unlocks agentic AI at the edge."
Mizuho just upped their PT to $1,090 (from $875) — that's now the high on the Street, sitting above Evercore's $1,000 and BofA's $900. Stock at $870.60, P/E 81.7, +576% over the past year. The bar is HIGH and the bull case needs HAMR to keep printing.
"HAMR technology mix growth and strong nearline demand visibility extending into calendar 2028 at key cloud customers."
Mizuho is modeling 91% EPS growth in FY27 to $28.35 (street $26.57), and the new PT implies ~38x that number (up from 36x). FY28 EPS jumped to $43.92 (street $38.93) — they're getting aggressive on the back half. Revenue path: $17.5B FY27 → $24.2B FY28. FCF-funded buybacks are the torque.
BULL: HAMR mix shift plus nearline visibility into CY28 is unusual for this space. Cloud capex digestion fears are overdone. FCF + repurchases = compounder at scale.
BEAR: Up 576% in a year. 38x FY27 into a tape that already paid for perfection. Any wobble in HAMR yields or hyperscaler order pacing and this gets cut hard. At 81x trailing you're paying for FY28 to land AND FY29 to be visible.
THINKING: We've been long all year. Story isn't the question — price is. A sell-side PT raise to $1,090 is validation, not a green light. If you have it, you trim into upgrades and reassess into the next print. If you don't, you chase and hate yourself.
Rosenblatt $225 reiteration post-AIPCon 10. Above the Street. Buy stays. The event itself was the catalyst — customer disclosures across legal (Kirkland & Ellis), construction (McCarthy), gov (USDA), enterprise (Hertz, Parts Town), and the expanded Google Cloud partnership (two-way data federation between BigQuery and Foundry). That's a real cross-vertical breadth list, not the same 3 Fortune 500 logos recycled every quarter. The Google tie-up is the underrated one here — that's PLTR embedding deeper into the hyperscaler distribution channel, which matters for land-and-expand at a time when every enterprise AI budget is getting questioned.
"Palantir is one of the most important components of the enterprise AI value chain." > — Rosenblatt
That line captures the buy-side thesis cleanly. Consensus is $200 (Baird lives there), Rosenblatt at $225. Spread's tightening, which means upside from here is going to come from print beats and new logo announcements, not multiple expansion. Trade the name, don't marry it.
TD Cowen $160 from $125, Buy held. ~10% upside to new PT is the wrong number to focus on — the real tell is product revenue +41% YoY in Q1 FY26, a clean reacceleration that reframes FTNT from "secular grower" to "cyclical grower with secular tailwinds." Print was a beat across the board: $1.85B revenue (+20%), billings $2.09B (+31%), 80.3% gross margins held even as memory costs ramp. That last data point is the under-appreciated one.
The ASIC strategy is doing real work. Competitors are getting squeezed on memory pricing; FTNT is largely insulated because they own the silicon. TD Cowen specifically flagged this as a structural advantage, and the margin print validates it. Bull case is pretty clean right now: refresh cycle, AI tailwind (FortiAIGate now integrated with NVIDIA platforms for AI workload security), data center buildout, sovereign SASE. 38 analysts revised earnings upward post-print, so this is consensus catching up to a setup that's been working for a few quarters.
Risk to the r/r: stock already at $146, only 10% to new PT, and the easy money on the reacceleration narrative is largely in. PT cluster from peers (Truist $120, Rosenblatt $125, Scotiabank $110, Cantor $110) is meaningfully below TD Cowen, so there's a wide bid/ask between the Street and the most bullish houses.
"TD Cowen noted that artificial intelligence is providing a next growth layer for Fortinet and that data center buildout and sovereign SASE represent long-term tailwinds."
Truist opens coverage at Hold, $90 PT against a stock printing $81 and bumping up against the 52-week high of $90.80. SOLS is up 74% in six months. Not the easiest timing for a "wait" call — either you're late to the party or you're smartly sidestepping a melt-up.
The thesis: portfolio sits in the refrigerant transition, advanced electronic materials, and nuclear — all secular tailwinds. Truist even concedes the premium multiple vs specialty chem peers is "warranted." The rub is capex stepping up to chase those multi-year organic projects, which crimps near-term FCF. Firm doesn't see growth + cash flow profile warranting further upside from here.
"Truist said it does not believe the company's near-term growth and cash flow generation prospects warrant meaningful upside."
Contrast that with BMO, already at Outperform, which just lifted PT to $101 from $92 on the nuclear/uranium angle post a recent investor event. Q1 print was real — EPS $0.63 in-line, revenue $991M beating the company's own guide. The story isn't broken; it's priced.
Two brokers splitting on whether the nuclear/refrig narrative has another 10-15% in it. Stock is near highs on a 74% six-month rip, capex is ramping, and the bull case is well-telegraphed. We lean Hold — fade into strength, don't chase. That $90 PT is a sensible bogey for fading into, not buying through.
Wedbush finally bites — upgrade to Outperform, $70 PT (from $56). That's a 25% raise on a stock sitting 39% OFF its 52-week high ($87). The call is simple: pipeline converts in H2, narrative flips.
Cognizant is one of those names PMs forgot about in the AI capex arms race. IT services, not AI infrastructure. But the bookings print tells a different story — and Wedbush is leaning into it. Stock trades at 11.6x P/E (the upgrade is roughly 11x 2027), which is value-territory for a TMT name with a credible AI story now.
Q1 bookings: +21% YoY to $29.6B TTM. Book-to-bill hit 1.4X. That's strong for any IT services shop — basically means they're booking $1.40 for every $1.00 shipped. Even more important is the MIX:
That second clause is the kicker. They grew bookings +21% into a soft demand backdrop. If discretionary spend inflects even modestly in H2 (management's base case), the revenue follow-through writes itself. Vertical AI strategy in financial services is where the wins are — health sciences stable despite regulatory noise.
Wedbush isn't early here — they're catching a re-rating that hasn't started yet. Stock needs to digest $70 PT, then fight $87. The bookings data is the real signal. Book-to-bill 1.4x with mix shift to large deals is what re-rates IT services names. Pair trade candidate if you're long the AI infra complex (NVDA, AVGO) and want a services beneficiary on the convert side. Not a high-conviction core long, but the risk/reward into H2 prints is interesting at these levels.
Craig-Hallum steps up with $18 PT and a $100B+ aspiration — that's nearly 8x the current $13.4B mkt cap, which tells you this is a "what if it works" call, not a "what is it" call. Stock at $6.52, up 78% YTD, so the market is already sniffing the autonomous long-haul thesis. The $18 print is now the high on the Street, north of the $15 prior topper and way above the $3.59 low end of the cluster. For context: $100B+ valuation would put AUR in the same zip code as the entire US Class 8 truck market — that's a heroic bogey for a company still burning cash (Q1 EPS -$0.10, only a penny better than consensus).
"the most commercially advanced on-highway trucking product currently available" — Craig-Hallum
R/R here is asymmetric in both directions. Bull case: regulatory green light, hub-to-hub driverless scales, freight margins capture. Bear case: timeline slips again, capex stays heavy, and $6.50 was just a 2024 AI-beta melt-up. Northland also came in this week with an Outperform at $11, so you're getting multiple initiations clustering around a narrative that AUR is the cleanest US-listed pure play on physical AI / autonomous freight. Light coverage day — not enough data to get more granular. Watch the Q2 print and any DOT/FMCSA commentary on the driver-out framework; that's the real catalyst, not PT raises.
Jefferies initiates with Buy, $34 PT vs ~$27 print (27% implied). Stock already up 72% over the past year, so the real question is chase vs. fade. The thesis is a capital-light OFS model with EBITDA margins walking from ~19.8% in 1H26 to ~22.2% in 2027, driven by Eldridge exit, under-absorption cleanup, and low-margin subsea backlog roll-off. FCF + clean balance sheet fund the M&A flywheel — just closed Drilling Innovative Solutions (float valves, cement retainers) to bulk up the consumables book.
"The analyst expects margin expansion beginning in the second half of 2026, driven by the Eldridge exit, elimination of under-absorption, and footprint optimization."
Eldridge exit is the cleanest piece — that's real unlock, not narrative. 2027 margin target (22.2%) is where the multiple expansion lives, and Jefferies isn't paying up for the multiple today. Bull case: 2H inflection prints, bolt-on M&A keeps compounding, the consumables mix keeps shifting toward higher-margin product. Bear case: 72% in a year already discounts a lot of the path, and OFS is a cyclical name trading like it's a compounder. First initiation on the tape — expect 2-3 more shops to follow with bullish reads into mid-summer.
SUOPY (Schneider Electric ADR) — Direct read from the SoftBank €75B France deal: Schneider is the named industrial partner for 5 GW by 2031, 3.1 GW phase one. This is the largest single European AI infrastructure commitment and SUOPY is the lock for DCPOWER, busway, and the 800VDC architecture. Bull: anchor customer with multi-year revenue visibility. Bear: European energy cost structure is the unresolved variable — a Brent spike on US-Iran tail risk is the real project-killer.
CDNS — No fresh signal in today's feed, but the theme stack all reads back to EDA. Glass core substrate ramp 2028, Intel 18A viability as second source, AI silicon design complexity scaling — CDNS sits upstream of every tape-out. Bull: every NVDA/AMD/ASIC chip is a CDN seat/license pull. Bear: 2H26 ODM production cuts -15-18% YoY implies consumer/smartphone weakness dragging EDA volume at the low end.
GOOG — Treated same as GOOGL above. CAPEX DOUBLED TO $190B, Apple Foundation Model on GCP, share gain 16%→22%. Constraint unchanged: equity dilution risk if the raise prints, and 3 weeks straight down says the positioning unwind is already in motion.
APLD — Neocloud caught in the SMCI dilution contagion. Backlog is real, capex financing is the test. Bull: AI inference $66B→$292B by 2029 reads directly to neocloud demand. Bear: SMCI's $7B raise is the template — every neocloud with negative cash conversion is a forced seller eventually. APLD is on the same list.
FFIV — ADC and security layer for the inference era. As inference traffic scales ~4x by 2029, app delivery and east-west traffic management become load-bearing. Bull: defensive infra, high switching cost, AI data growth = structural tailwind. Bear: any cloud-native service mesh or hyperscaler in-house ADC is the slow-burn disintermediation risk.
NVT — Data center power, busway, liquid cooling, rack PDU. Reads alongside SUOPY at smaller scale and US-skewed. Bull: VRT's 34% growth guide + Schneider 800VDC transition is the proof point for the category. Bear: VRT is the cleaner pure-play; NVT is more diversified, less torque to AI.
ROKU — CTV ad platform, no direct signal in the feed. Exposed to ad budget reallocation as TikTok/Amazon Ads consolidate. Bull: CTV share gains structurally intact, ad targeting moat deepening with AI. Bear: AI-native creative tools + Amazon/Google ad dominance = mid-term headwind to take rates.
FCEL — Fuel cell behind-the-meter. SoftBank France deal explicitly cited behind-the-meter as the only path to bypass the 3-7 year interconnection queue. Bull: FCEL positioned for on-site generation contracts at European industrial sites. Bear: BE has the larger installed base and better-known anchor customers (Apple, Equinix, AEP).
WIX — AI website generation pressure is the structural bear case. Hostinger/durable.co style AI-native builders commoditizing the entry tier. Bull: SMB demand still durable, AI tools could be additive if integrated. Bear: pricing power at the bottom erodes every quarter a new AI builder ships.
FICO — Credit decisioning moat, AI underwriting tailwind. Banks and regulators are hooked on FICO score infrastructure. Bull: AI lending models still need a benchmark score — FICO is the rail. Bear: AI-native lenders building proprietary risk models is the slow-burn disintermediation.
WYFI — Light signal in the feed, no conviction on the name. Skipping directional call.
OKTA — Identity layer for the AI agent era. NOW's agentic revenue target raised to $1.5B for 2026 — every agent needs an identity. Bull: agent identity management is a new TAM sitting on top of human IAM. Bear: PANW/CRWD platform consolidation could squeeze OKTA at the layer where the budget is consolidating.
PEGA — Workflow + low-code + AI agents. Competing with NOW, CRM, MSFT in enterprise workflow. Bull: regulated-industry moat (banking, insurance, healthcare) sticky. Bear: NOW's pricing power and agentic traction = PEGA losing share at the high end of the stack.
BE — Bloom Energy. SOFC behind-the-meter for AI DCs. Larger incumbent than FCEL with anchor data center customers. Bull: existing AEP/Equinix/Apple contracts = reference customers for the next wave. Bear: natural gas economics are the tail risk if LNG prices spike on US-Iran; fuel cells don't pencil without cheap gas.
RDDT — Data licensing line is the AI hedge. $2B+ revenue partially offsets search disintermediation as ChatGPT/Perplexity/Google AI overviews eat traffic. Bull: Reddit's conversational data is a unique training corpus, signed deals with OAI and Google are the floor. Bear: traffic decline is real and ongoing; founder voting control = persistent governance discount. Open question: what's the long-run equilibrium between licensing revenue and traffic loss? No one knows yet.
Hearing SoftBank shares are +70% YTD 2026 on AI infra expectations, and the €75B France commitment is the first concrete monetization of that narrative. If equity holders start asking where the FCF is, the stock has the same unwind profile as any other AI-infrastructure-SPV name.
Word is the Vista Equity insurance workflow ($8M manual → $200K with proprietary system) is now being shopped as a reference case for AI ROI in PE portfolios. If this scales across Vista's portfolio, it's the strongest counter-argument to the "AI bubble SPV" narrative currently circulating. 40x cost reduction is the kind of number that travels.
Channel checks from the Asian supply chain: MLCC distributors are hoarding inventory preemptively, PC/ODM ASPs +17% with 2H26 production cuts of 15-18% YoY. Demand destruction is showing up at the OEM/ODM level even as memory contract prices print +200-300%. When the consumer can't absorb, the step-down is violent.
Hearing xAI struck a short-term deal with 90-day optionality to reclaim capacity. Read-through: the big neoclouds are now offering reclaim clauses because the demand is so much ahead of buildout they can re-rent at higher rates. The constraint is real and the optionality is the tell.
Rumor NVDA asked InP laser vendors for 20x capacity scaling, most only committed 12x. The shortfall is the 2027-2028 optical bottleneck nobody's modeling yet. LITE/COHR/AAOI are the names that print if the 20x gets forced through.
Hearing from GSA podcast circuit: Jensen confirmed NVDA as Customer A displacing Apple at >20% of TSMC revenue. Implies CoWoS allocation is structurally NVDA-dominated going forward. Zero-sum for AMD/ASIC competitors chasing the same wafer pool.
Unconfirmed but circulating: Intel 18A yields are "going right up like this" per Lip-Bu Tan commentary, Panther Lake shipping at 7x volume on ~50% yields. If Intel Foundry becomes a real second source for leading-edge, the TSMC concentration risk narrative breaks. Not there yet, but the slope of the curve matters.
Word is OpenAI's Sora team is pivoting to robotics — capability reallocation toward physical-world AI consuming frontier lab talent. Robotics is inference-heavy + edge-deployed, shifts the lab → hyperscaler → rack distribution of compute. Read-through to any on-device inference play (QCOM, NVDA N1X).
Hearing DeepSeek driver-assist tech shipping in BYD Ti7 including voice-activated KFC ordering. Distillation of frontier models into consumer hardware at scale is now mass-deployment, not prototype. Partially offsets chain-level capex growth via efficiency feedback loop.
Channel check suggests GPT-5.6 is crushing the deepswe benchmark, beating Claude Opus 4.8 on agentic coding eval. Coding agent leadership remains with OAI — incremental, not structural shift, but the agentic moat is widening at the top.
Rumor mill: glass core substrate initial production targeted ~2028 for high-end applications, not 2026/27. Means the glass substrate revenue ramp is out of the current model window for ABF incumbents — read-through to COHR/laser-drilled glass substrate suppliers as a 2028+ thesis.
Hearing MacBook Neo shipments revised 5M → 10M units. Validates the Arm+AI PC category formation thesis; reads directly to NVDA N1X and QCOM Snapdragon C as adjacent silicon competitors. NVDA's N1X social with Microsoft + Arm was timed to this.
Unverified but in the feed: AI inference TAM $66B (2025) → $292B (2029), 45% CAGR. If that trajectory holds, the 4-year constraint narrative is structural, not cyclical. Doubles the addressable capex pool for neoclouds/hyperscalers in 36 months.