Good morning.
MU completely changes the cycle math. Print was a blowout — $41.5B revenue vs $35.7B est, guide midpoint of $50B vs $43.2B consensus. But the $100B FLOOR on multi-year strategic agreements is the real story. Hyperscalers locking memory supply through 2027. NASDAQ 100 futures +2.2% on the reversal.
QCOM +13% AH on investor day. DC revenue target raised to >$15B by FY29. C1000 agentic CPU wins with Meta. Two custom ASIC wins (Amazon speculated). They're spending $14B+ to assemble what NVDA ships integrated — but the market loves the ambition. AVGO +2% on OpenAI's 9-month Jalapeño tape-out. Custom silicon viability debate is dead. CBRS crashed to post-IPO lows — admitted DC shell shortage. Physical deployment lag is the new bottleneck.
Asia green on the MU hangover. Taiwan semis +2-3%, Nikkei flat. DXY at 52-week highs, oil below $70. Gold broke $4k. Risk-off macro colliding with semi euphoria.
Three themes framing the tape today:
1. The memory supercycle just got a term sheet. MU's LTSA structure embeds agentic AI caching requirements into a 5-year pricing floor. Sell-side models now pricing HBM +80% by end-2027. Supply has no line of sight to catch up. This elongates the cycle past everyone's models.
2. Custom silicon is accelerating faster than expected. AVGO's 9-month tape-out with OpenAI proves the ASIC threat to NVDA inference is real. QCOM's DC push adds another vector. The question isn't if hyperscalers diversify — it's how fast they can deploy. MSFT rumored as AVGO's 7th XPU customer.
3. Behind-the-meter power is the new binding constraint. FCEL +16% on 380MW on-site DC deal. Grid interconnection queues are 3-7 years. Co-located generation is the only immediate workaround. This theme gets bigger every quarter.
We'll hit up MU, QCOM, AVGO first, then get to semis and energy plays.
Verdict: BLZE just got its "I'm relevant" badge. The CoreWeave deal is a genuine step-change — $335M over 5 years, multi-exabyte, HDD-based storage for AI. Analysts are scrambling to catch up, and the stock is up 150% YTD. The question now is not whether the deal is real (it is), but how much of the $11-12 stock already prices in the good news, and how fast the revenue actually hits the P&L.
PT range: $14.00 (Needham, Citizens) to $16.00 (Craig-Hallum) , up from a prior cluster of $8.00-$8.50. Consensus is cautiously bullish — all three upgrades/PT raises are based on the CoreWeave deal. William Blair went from Underperform to Market Perform (not Buy), signaling they still want to see execution before getting fully constructive. No one has cut — completely one-way positive revisions.
Bull case (3-4 sentences): This isn't just a random deal — CoreWeave is the poster child for neocloud GPU-as-a-service. They needed a reliable, cheap HDD-based storage layer for cold/cool AI data, and Backblaze proved it can deliver at scale. $335M over 5 years is roughly 2.1x BLZE's current 2026 revenue guide ($162.5M midpoint). Even if only half that hits in the next 18 months, you're looking at a material step-function in growth. Craig-Hallum's analyst wrote it best: "Top line acceleration and meaningful upward revisions to revenue and profits on the way." At current EV/rev of ~4.3x on 2026 estimates, even a modest multiple re-rating to 6-7x (still cheap for a growth story with visibility) gets you to $15-16.
Bear case (3-4 sentences): The contract includes two components — B2 Cloud white-label AND a managed storage model where Backblaze's staff sits inside CoreWeave data centers. That second piece is a brand-new delivery model for BLZE, with zero track record. Margins could be very different (lower) than the core B2 business, and we have no idea how the revenue will be recognized. Needham explicitly says the deal is "not included in Backblaze's 2026 guidance as the company works with auditors." William Blair kept 2026 estimates unchanged. Execution risk is real — BLZE has not run a managed services operation at this scale before. And at $700M market cap, the stock is no longer option-implied cheap; you're paying for a story that hasn't delivered a single dollar of GAAP revenue yet.
New: The CoreWeave deal itself (announced 6/23). Also new: warrant exercise price of $7.60 for CoreWeave (implied ~35% discount to the stock at time of deal). The managed storage/"embedded staff" component is incremental — this is not just a reseller agreement.
Already known: Q1 earnings beat (EPS $0.04 vs -$0.0043 expectation, revenue $38.7M vs $37.79M). Backblaze's HDD-based storage thesis for AI? That was the narrative before. But investors weren't buying it until a name like CoreWeave signed up.
"Landmark $335M Deal With CoreWeave Changes The Game. Top Line Acceleration And Meaningful Upward Revisions To Revenue And Profits On The Way. Shares To Head Higher On EPS-Based Valuations." — Craig-Hallum analyst (Jason Krey?) , most aggressive buy-side call.
"The magnitude of the commitment and the nature of the customer relationship demonstrate that Backblaze's platform can play a pivotal role in emerging neocloud architectures." — William Blair, the cautious upgrade note.
"Backblaze has built a reputation for making complex, HDD-based storage infrastructure reliable and easy-to-consume at scale." — CoreWeave VP Nick Hoover, per Citizens report.
CoreWeave's own stock is up too — Citizens has a $180 PT on CRWV with Market Outperform. This deal reinforces a broader thematic: AI workloads generate massive amounts of cold/warm data that don't need all-flash. The neocloud ecosystem is bifurcating into high-performance compute (GPUs, fast storage) and cost-effective capacity (HDD, tape). Backblaze is the pure-play HDD storage proxy for that second bucket.
Peers/read-through: If this model works, look for other neoclouds (Lambda, Vultr, maybe even smaller players) to sign similar deals. Also watch for incumbents like PSTG (Pure Storage) or NTAP (NetApp) — they have HDD tiers too, but Backblaze's simplicity and pricing are its differentiators. For now, BLZE is the only pure-play NASDAQ-listed name in the "AI cold storage" theme.
Bottom line for PMs: The setup is good, but the stock has already moved 40% on the deal announcement. If you're underweight, you missed the initial pop. The next catalyst is the earnings call (likely late July/early August) where management will provide revenue recognition timing and maybe raise 2026 guidance. Wait for a pullback into the $9-10 range if you want better r/r. If you're already long, trim into strength — $14-16 is the consensus ceiling until we see actual numbers.
Verdict: All three analysts (Needham, Rosenblatt, Mizuho) reaffirm Buy/Outperform with $300 PT after the Q1 print. The story is clean – demand is real, margins are inflecting faster than anyone modeled, and OpenAI deepens the relationship. But the stock is down 27% in six months and the market is pricing in perfection on a $50B market cap with hardware revenue shrinking. Rate of change is positive; positioning is the question.
Q1 revenue $191.3M (+92% YoY), beat. June Q guidance $194M vs consensus $181M. CORE GROSS MARGIN GUIDANCE OF 37% FOR JUNE Q – 12.4 points above the 24.6% consensus (Rosenblatt had modeled 24.6%). Full year margin guide of 39.5% at midpoint vs Rosenblatt’s 29.6%. That’s a massive beat on the margin trajectory.
The driver: cloud pricing is improving as the market puts a premium on fast inference tokens. Cerebras Cloud revenue rose 60% QoQ as OpenAI ramps. OpenAI’s GPT-5.4 is live on Cerebras, with GPT-5.5 integration underway. DISAGGREGATED INFERENCE DEAL WITH AWS (Trainium3 for prefill, CS-3 for decode) ramps in 2027 – that’s a 2027 story but gives the bull case a multi-year horizon.
Three firms, same $300 PT. Needham uses 12.5x 2028 EV/revenue. Mizuho calls Cerebras the “fast inference leader.” Rosenblatt highlights the margin flip:
“Management’s core Non-GAAP gross margin guidance for the second quarter of 2026 stands at 37%, compared to the firm’s estimate of 24.6%. Full year gross margin guidance of 39.5% at the midpoint also exceeded Rosenblatt’s estimate of 29.6%.”
That’s not a small beat. That’s a structural improvement in unit economics as cloud mix shifts and token prices rise. The primary constraint? Data center capacity, not wafer supply. Management named new DCs in U.S., Canada, Europe, and is in talks for Israel, UAE, Australia, Singapore, India, Indonesia. So the bull case is capacity-limited demand with accelerating margins and a sticky hyperscaler relationship.
Bull: Cerebras owns the high-performance inference tier. Token pricing is 50% higher for fast inference (per Mizuho). OpenAI is expanding beyond GPT-5.4 to 5.5. The AWS deal creates a 2027 revenue tailwind. Margins are inflecting faster than any forecast model captured. Data center constraints are a “good” bottleneck – excess demand, not weak order flow. At 12.5x 2028 revenue ($300 PT implies ~$6B+ 2028 revenue), the multiple is reasonable for a company growing >60%.
Bear: Hardware revenue declined 8% QoQ and is expected to keep declining through FY2026. The stock has shed 27% over six months – the market is already pricing in the cloud transition risks. INVESTINGPRO FLAGS CBRS AS OVERVALUED relative to its fair value estimate. EPS loss of $0.14 despite the revenue beat. If the AWS ramp slips or OpenAI goes multi-source, the narrative breaks. At $50B market cap, you’re paying for perfect execution on a 2027 catalyst with hardware revenues shrinking today.
The margin acceleration is real and underappreciated. The 37% guide vs 24.6% consensus is not a rounding error – it’s a proof point that the cloud pivot is more profitable than sell-side models assumed. But I’d flag the valuation context: $226 stock, $50B market cap, still loss-making. The 27% YTD decline tells me PMs are in “show me” mode. If you’re long the name, you’re betting that data center constraints ease, OpenAI stays exclusive-ish, and the 2027 AWS revenue becomes a catalyst not a tease. High r/r, but the signal is improving.
Verdict: The analyst consensus is you ignore the P&L and watch the dirt. Three firms (Bernstein, Piper Sandler, RBC) all reaffirmed Outperform/Overweight calls this morning. Stock near $34, a whisker off the 52-week low of $32.39. Market cap ~$10.3B. The buyside debate isn't about Q1 — it's about whether this is a $47-51 stock (the target range) or a story that needs hard power before it rerates.
All three are saying the same thing with different emphasis: execution on Cape Station is the only data point that matters. The Q1 print ($61k revenue, -$31.8M net loss, -$3.72 EPS) is noise. Bernstein explicitly says stop focusing on near-term earnings. Piper Sandler points to the capital access playbook — $2.2B IPO war chest plus $421M non-recourse project debt for Cape Phase I — creating a repeatable framework. RBC notes Cape Unit I (33MW) is mechanically complete and in commissioning, with Phase II on track.
The collective thesis: this is a pre-commercial infrastructure build, not a SaaS growth story. The stock trades on milestones and cost curves, not revenue multiples.
From RBC (the strongest on project-level detail):
"Cape unit I, a 33-megawatt facility, is mechanically complete and currently being commissioned. Offtake customers are signing agreements without waiting to see Phase I performance results."
That last bit is the key signal. Customers committing before first power is derisking the commercial model. If offtake accelerates, the 1GW by 2030 target (and upside Piper Sandler hints at) looks more achievable. Interconnection queues, not capital, are becoming the bottleneck.
Bull case: The $2.2B IPO is a loaded gun. Cape is on schedule. Behind-the-meter deals with AI data centers bypass grid constraints. Management's confident tone on 2030 upside suggests a catalyst path that isn't priced in at $34.
Bear case: You're staring at a $10B market cap for a company with $61k in revenue and $31.8M in quarterly losses. Pre-revenue geothermal is a high-beta bet on execution. Any delay at Cape — commissioning hiccups, cost overruns, grid interconnection rejections — and the stock gets cut in half. The 3.0 design at Phase II is unproven at scale.
Bottom line: The analyst PT spread ($46-51) implies 35-50% upside from here. The stock is at the low end. That's the setup — but it's a conviction play on physical construction timelines, not a thesis you can prove with an income statement. If Cape Unit I delivers first power in H2 2026 as guided, the narrative flips from "show me" to "tell me how fast you can build Phase II." Until then, it's a waiting game with a very real downside if the drill bits stop turning.
ORCL – STRUCTURAL BULL WITH NEAR-TERM CASH FLOW PAIN. The 10-K confirmed what the Q4 print hinted at: the AI hyperscaler buildout is real, expensive, and creating massive revenue visibility. But burning $23.7B in FCF while raising $20B in equity-linked paper is not a free lunch. PMs need to decide if they trust the backlog curve or fear the dilution.
### THE BACKLOG TELLS THE STORY
Evercore ISI’s deep-dive on the fiscal 2026 10-K dropped the headline number: cRPO HIT $77B, OR 86% OF THEIR NTM REVENUE ESTIMATE (up from ~68% in Q4). That is the kind of visibility most infra plays can only dream of. The firm keeps Outperform / $245, but the real message is that revenue is increasingly locked in.
"The company’s current remaining performance obligation reached $77 billion, representing 86% of our next-twelve-month revenue estimate, up from approximately 68% in the fourth quarter. The figure provides confidence in revenue visibility."
The offset? HEADCOUNT DOWN ~21K Y/Y – that’s the cost control lever pulling operating expenses lower. Combined with revenue growth (LTM +17%), it’s enough to offset gross margin dilution from the heavy COGS of AI infra.
### EXPENSE CLARITY GETS A CHECK
KeyBanc upgraded numbers, not just rating. They raised EPS estimates for FY28-30, now above consensus on FY29-30. The key unlock: OPEX GROWTH MODERATING ENOUGH TO EAT THE GM HIT. They sit at $300 PT and Overweight, and the language is increasingly confident. Mizuho ($320), BMO ($220), and TD Cowen ($300) all chimed in post-Q4 with buy-equivalent calls. Bullish consensus is wide (PTs $220-320) but directionally unanimous.
### THE BEAR CASE IN TWO NUMBERS
CAPEX: $55.7B vs $21.2B prior year. FCF: -$23.7B. Oracle is funding this buildout with debt and equity-linked issuance – they flagged ~$20B in equity-linked financing over the next couple quarters. That’s dilution risk, full stop. The stock trading at $165.16, down 10% in the last week, suggests the market is already pricing in some of that anxiety.
Verbatim the bull case: 1GW of capacity coming online in FQ1’27, demand environment still strong, and the backlog gives a multi-year revenue floor. Bulls say the free cash flow inflection is a 2027 story, not a 2026 one.
### THE TAKE
ORCL is a high-conviction structural name with a messy near-term P&L. The bullish consensus is wide – KeyBanc $300, Mizuho $320, Evercore $245 – but the common thread is that the expense picture is improving and the backlog is king. If you can stomach the cash burn and possible equity-linked overhang, the r/r looks skewed long-term. Just don’t expect a smooth ride in the next two quarters.
Verdict: CPUs are the new AI beneficiaries as agentic AI shifts workloads from thinking (GPU) to doing (CPU). The market is waking up to a $15B+ revenue opportunity by FY31, and the analyst community is repricing this thing hard. Two more PT bumps this week — TD Cowen to $475 (from $265), UBS to $470 (from $260). Both maintain Buy. Cuts across the tape now cluster around $470-500, with Bernstein and Mizuho already at $500. ARM trades $357, down 12.5% on the week — but still up 235% YTD with a $381B market cap. The pullback is noise. The thesis is strengthening.
Agentic AI is the catalyst. As workloads move from GPU-intensive "thinking" to CPU-intensive "doing" (data routing, action execution, orchestration), CPUs get a second act. ARM is the architecture of choice for hyperscalers building custom silicon — power efficiency + latency advantage = natural fit for the "doing" layer.
TD Cowen nails it: "The Doing Behind The Thinking: As agentic AI shifts more work from the thinking GPUs do to the doing CPUs handle, CPUs are becoming an AI beneficiary." They see ARM's $15B FY31 AGI CPU target as reasonable, and flag that the market may be too mechanical in applying a 15% share estimate to Nvidia's $200B CPU TAM.
UBS is similarly constructive. They bumped their internal CPU product revenue estimate to ~$14B by CY2030 (from $13B) and see ARM architectures capturing ~70% of a ~20M unit head-node TAM by 2030. Caveat: first-gen internal product is limited in core count/throughput — not ready for standalone AI racks. But the roadmap of 200-500 cores opens broader use cases in gen 2/3.
Bull case: ARM is the tollbooth for the CPU renaissance in AI. Hyperscalers (Meta, OpenAI, ByteDance, Oracle) are already locked in. The IP licensing model generates high-margin recurring revenue. At $15B+ revenue by FY31, even a 30-40x multiple gets you to $500+. The AGI CPU narrative is still early — attach rates and pricing per core are the key variables, and both are trending up.
Bear case: This is a $400B+ company doing ~$4B in revenue this year. The valuation assumes perfection — that agentic AI actually materializes at scale, that ARM's internal chip doesn't get crushed by Nvidia's own CPU efforts, and that the 70% share estimate holds. First-gen product is underpowered. And the China export restriction story is muddy — CEO Rene Haas himself said distinguishing AI-capable CPUs from general-purpose ones is "like distinguishing oil from oil." Regulatory risk is real.
"The Doing Behind The Thinking: As agentic AI shifts more work from the thinking GPUs do to the doing CPUs handle, CPUs are becoming an AI beneficiary." — TD Cowen
ARM is a long-term AI infrastructure play that just got a fresh catalyst from the agentic AI theme. The pullback this week is pure profit-taking from a 235% YTD run — not a thesis break. The analyst moves are uniformly bullish and raise the floor. If you're underweight TMT semis, this is the name to add for CPU exposure. Risk: valuation is stretched, and the China export dynamic adds headline risk. But the rate of change in the AI CPU narrative is accelerating, and ARM sits at the center of it. Position accordingly.
(Note: I'd be watching the core count/throughput gap on first-gen product — if NVIDIA announces a competing ARM-based CPU at Computex next year, the bull case gets narrower. For now, the path is clear.)
Verdict: ZETA is getting a fresh look from the Street, but the bull case now rests on execution on the Palantir partnership — not just organic growth momentum. The stock sits at $19.51, flat YTD, meaning the 44% upside to GS’s $28 PT is either a gift or a trap, depending on how you read the Palantir deal's near-term financial profile.
Two firms, two different entry points, same underlying thesis: ZETA is a legit marketing tech platform sitting on the intersection of customer data, activation, and measurement. But the divergence in conviction is real.
Bull case: ZETA is compounding at 29% organic (accelerating from 28%) with 19 straight quarters of beats. The Palantir deal adds AI enterprise credibility and a long-term growth vector. GS sees 44% upside. The Marigold acquisition over-delivered by $8M. Political spend in H2 is a real tailwind. Profitability is coming (consensus $1.02 EPS this year).
Bear case: The stock is FLAT YTD in a weak software tape — that's relative outperformance, not absolute conviction. GS's Neutral says it all: the discount to peers has closed. Needham's $25 PT is only 28% upside, and they admit the Palantir deal doesn't move the FY26 needle. Political spend is lumpy. If the macro turns, ZETA's enterprise/agency exposure gets clipped first. At $4.9B EV, this is not exactly a bargain.
ZETA is a real business — 33% revenue growth, product-market fit with big clients, and a credible AI narrative via Palantir. But the setup is messy. You're paying for a H2 political pop and a multi-year Palantir ramp that management itself says doesn't hit FY26 guidance. The risk/reward is okay, not great. If you're long software and need a growth name with a catalyst calendar, ZETA works. If you want clean IRR in a choppy tape, wait for a better entry or see if GS is proven wrong by earnings.
Tale of two PTs. Jefferies cuts to $22 (from $30), Hold. TD Cowen keeps Buy at $35, names it Best Smicap Idea of 2026. Stock at $19.35 — down 47% from its $37.10 high, off 31% in six months. The divergence tells you everything: this is a bet on execution timing vs. near-term noise.
Jefferies cited mark-to-market adjustments and a lower premium vs. peers as emerging competitive threats squeeze the r/r. They acknowledge the underlying business is strong but see a mixed H2 2026 — positive catalysts (1 GW project announcements, NRC update for DOW) offset by AP1000 and EGS headwinds. Management called the NRC filing delay for the Amazon/Energy Northwest project "virtually a non-event" at a fireside chat.
TD Cowen leans the other way. They see the pullback as a gift, citing 109% YoY revenue growth in Q1 (first print since IPO) and a CPA submittal shift to H1 2027 that "should not affect the project timeline." Several catalysts later this year could derisk the story. Valuation looks attractive on both absolute and relative basis.
"Amazon/Energy Northwest CPA submittal shifted from 2026 to the first half of 2027, though TD Cowen said this should not affect the project timeline."
Bull case: You're buying a multi-year buildout option in SMRs at distressed levels. Amazon partnership intact, revenue compounding, and a backlog of project announcements pending. TD Cowen's conviction ($35 PT) implies 80% upside from here.
Bear case: The competitive landscape is tightening (AP1000, EGS), the premium to peers is compressing, and operating losses are deepening. The NRC delay — even if "non-event" — keeps the catalyst queue pushed out. Jefferies sees limited near-term upside.
Bottom line: XE is a pure timeline play. If the next NRC update or project FID comes through, the stock re-rates hard. If another delay hits, you're looking at deeper drawdown. Long-duration optionality with short-duration patience required.
Verdict: Mixed signals, but the floor is closer than the ceiling. TTD prints $17.93, down 75% in a year and kissing the 52-week low. The Publicis overhang is resolved and July is the checkpoint, but Walmart just ripped the exclusivity cork out. Two steps forward, one step back – net zero until we see real recovery flows.
The setup: Benchmark reiterates Buy, $30 PT (from fireside with IR). The key take: Q2 is the trough for both the Publicis dispute AND overall growth. Toth (IR) broke the impact into 25% direct agency-controlled spend (gone) and 75% joint business plan spend (brand-controlled, could ramp). The indirect piece is the recovery lever. Some spend has already returned, but July post-Cannes is the real tell. Benchmark’s Zgutowicz is leaning into the second-half narrative.
The headwind: Citizens keeps Market Perform. The Walmart exclusivity loss is real – Walmart now works with Magnite, Yahoo DSP, and Google DV360, plus bought Vibe.co (its own bidder for CTV/SMB). Analyst Condon: “We view today’s news negatively for The Trade Desk.” He flags that Vibe.co could broaden to other channels and enterprise, reducing TTD reliance further. Compounding factors: Rothschild Redburn initiated Sell with $11 PT, citing structural competitive challenges in the ad supply chain.
Bull case: Publicis dispute is resolved, Q2 is the trough, and the joint business plan spend can come roaring back in H2. TTD’s position with Fox/Roku (Fox bought Roku) keeps the premium CTV inventory story intact. New CFO Nate Olmstead (Logitech, HP) brings operational chops. At $17.93, downside is capped by the $17.21 low and a $30 PT offers 65% upside if recovery materializes.
Bear case: Walmart was TTD’s crown jewel retail media relationship. Losing exclusivity while gaining a rival bidder (Vibe.co) is a structural revenue hit. The Publicis recovery is uncertain – July could disappoint. The Sell-side at $11 says the competitive moat is eroding. Consensus earnings cuts (18 analysts down) suggest the fundamental trajectory is still negative.
Strongest line from the source:
“We believe Walmart could broaden the bidder’s applicability to other channels and extend it to enterprise-level capabilities over time, further lowering its reliance on TTD.” — Citizens analyst Matthew Condon
Bottom line: TTD is a high-r/r recovery play, but the Walmart news adds a new variable that doesn’t get resolved in July. Wait for the July spend data and Walmart’s next move before leaning in.
Verdict: The NHanced deal is the right vertical move, but the price tag requires revenue delivery that the company has never come close to. Rosenblatt is all-in on the narrative; Cantor is sitting on the fence with a $10 PT that implies 60% downside from here.
QUBT closed the acquisition of NHanced Semiconductors for $73.1M in cash/stock, plus up to $72M in earn-outs tied to revenue targets. The asset: a U.S. advanced packaging foundry with 2.5D/3D hybrid bonding, chiplet integration, photonics. This is the Fab 2 they’ve been teasing — a real manufacturing facility, not just a design house.
Rosenblatt loves it. Reiterates Buy, $22 PT. Sees a massive revenue step-function: their standalone forecasts of $30.5M (2027) and $59.5M (2028) jump to $65.5M and $109.5M respectively when NHanced’s targets are included. That’s a 15x-25x leap from the $4.33M TTM revenue base.
“The projections represent significant growth for a company with $4.33 million in revenue over the last twelve months. Additional upside would come from cash earn-outs if higher targets are achieved.”
Cantor is polite but unconvinced. Neutral, $10 PT. They like the strategic logic — “addresses how the company plans to scale volumes with Fab 2” — but that’s about it. They see QUBT capturing 5% of the quantum market by 2035, translating to $375M in sales. That’s a ten-year horizon for a $2.4B market cap stock trading at 500x current revenue.
Bull: The NHanced acquisition gives QUBT real manufacturing capability in a supply chain that’s desperate for domestic fabrication. Revenue kicker from earn-outs is contingent, sure, but the escrow structure ($20M held back, released on hitting $35M in 2027 and $50M in 2028 revenue) aligns incentives. Plus they’ve got $140M in cash (10% of which is the max cash needed), debt-free, current ratio of 66.67. No near-term liquidity risk.
Bear: They’re paying $73M (plus potential earn-outs) for a company that hasn’t publicly reported revenue history and whose earn-out targets are above Rosenblatt’s pre-deal forecasts. The cash escrow structure means if NHanced misses, the money comes back — but that also means the revenue doesn’t materialize. QUBT’s own core business is still sub-$5M in sales. This is a stock priced for perfection on a 2028 story, trading at 2.4x Rosenblatt’s 2028 revenue (including NHanced). Nosebleed territory.
Two other datapoints from the article noise:
Bottom line for PMs: QUBT is a narrative stock with one big binary catalyst (does NHanced deliver?). Rosenblatt’s $22 implies you believe the 2027-2028 revenue ramp. Cantor’s $10 says you’re early. The tape will decide based on next quarter’s organic revenue ex-NHanced and any updates on the Planck Dynamics pipeline. High risk, all optionality. Not a position size unless you’ve got a 3-year lockup.
Verdict: Delivery print is the only narrative that matters for the next week, and the spread between the bull and the house is massive. Stock is up ~7% from the UBS PT and 27% below Baird's. This is a stock that trades on the number, not the nuance.
Two firms, two very different takes. Baird is at 392,900 Q2 deliveries — below Street consensus of 401,100. UBS is at 405,000, which is 1% above Visible Alpha and sits right in the middle of buyside whispers (400k-420k). The spread between the highest and lowest published estimates? Meaningful. Baird sees a Q2 miss; UBS sees a beat.
The divergence matters because TSLA doesn't trade on earnings math right now. It trades on rate of change in deliveries. A print below 395k and this stock goes through $350. A print above 405k and you're testing the $400+ zone.
Baird's full-year 1.68M is slightly above consensus 1.66M, which means they're banking on a back-half ramp. Q3 at 461,500, Q4 at 469,300. That's a 20% H2 boost from Q2 run-rate. Not impossible, but asks a lot from a macro that's softening in Europe and China.
UBS is more cautious on valuation (P/E of 373, per the article) but did bump their Q2 estimate from 380k to 405k. That's a nice catch-up, but they're not buying the multiple expansion story.
Bull case (Baird): Outperform, $522. The merger speculation with SpaceX (post-IPO) is a narrative wild card. FSD approvals in Finland/Netherlands are real regulatory progress. The energy storage business is an underappreciated growth engine — 13.4 GWh in Q2 alone (+40% YoY). If you believe Tesla is an AI/energy play that happens to sell cars, the auto number is secondary.
Bear case (UBS): Neutral, $364. Stock trades above their PT. The NHTSA investigation into the Texas fatal crash is headline risk that never fully goes away. Valuation is silly at current levels unless you're discounting a decade of FSD revenue. UBS sees buyside expectations at the lower end — they're not wrong that the whisper range gives you a clear air pocket below 400k.
Baird's thesis on the delivery estimate methodology is worth reading:
"The firm based its estimates on international vehicle registrations in key markets, third-party trackers, and internal analysis."
Translation: they're not just pulling numbers from hearsay. Baird actually does the boots-on-the-ground work. When they're below consensus, it's a signal — not a guess.
Bottom line: If you're a PM, you're watching July 2 like a hawk. 400k is the line in the sand. Above it, the narrative flips to "demand is fine." Below it, we get a 15% haircut and an even more toxic debate about whether Tesla is a car company again.
RBC bumps PT to $80 from $72, stays Sector Perform. Stock sits at $74.94 — that’s ~7% upside now baked into the target. The firm hosted management (CEO Brace, IR Gill) and came away thinking mobile content bottomed (longer-term 6G/Physical AI/supplier consolidation story intact), while Broad Markets recovers on Wi-Fi, DC, Auto. QRVO deal closure now seen late 2026 vs prior early 2027 — they’re hopeful but note limited visibility on regulatory approvals.
Limited near-term catalyst is the punchline. RBC raised PT purely on sector multiple expansion, not a fundamental inflection. The stock’s already up 22.6% YTD, 20.6% last six months — valuation catching up to narrative. Options activity surged to 32,697 contracts (mostly calls) — highest since Feb 2025. That’s a signal, but not one a PM should chase without a proper trigger.
“Management did not comment on near-term trends, though RBC Capital believes business conditions are stable and continues to model seasonal trends through the year.”
Bottom line: SWKS is a show-me story on the QRVO deal and RF content ramp. The bull case rests on 6G and Physical AI eventualities; the bear case is that near-term revenue visibility remains fuzzy and the deal timeline could slip. At $74.94, r/r feels balanced — not enough to overweight, not cheap enough to ignore entirely.
TWLO is the AI comms play the street is finally pricing in. GS initiates with a $300 PT (63% upside from $184) – a full bogey above the consensus $210-255 cluster from Tigress, Oppenheimer, Needham, Rosenblatt, and TD Cowen. The thesis is straightforward: agentic builders need infrastructure to message end users, and TWLO owns that gateway. Self-service grew 28% YoY in Q4, Voice hit +20% in Q1 – these are not one-off bogeys, they’re rate-of-change acceleration off a restructured base.
“50% of the Forbes 50 AI startups were paying customers of Twilio as of September 30, 2024.” – Goldman Sachs
The GS call is the most aggressive, but the broader analyst crowd has already been creeping PTs higher through the Signal conference and Q1 print. Risk here is that the stock has already run 51% in the past year and 29% YTD – some of that AI premium is baked in. But the margin trajectory (gross profit growth acceleration) is the next catalyst, not just user growth. PMs should watch the Q2 guide for Voice sustainability and whether self-service momentum can hold the 28% clip. If it does, the $300 bogey looks achievable; if not, this is a crowded long that needs a reset.
Verdict: MS is finally pricing in the transition that’s been hiding in plain sight. PT to $130 from $90 (Overweight) — a ~32% implied upside from here. The catalyst is China memory buildout, but the real story is the product pivot.
They’re stepping up ECP adoption conviction for advanced packaging, plus seeing incremental non-mainland backend wins. Combine that with operating leverage as new tool categories scale, and you get a name that’s no longer just a China wet-cleaning play. MS sees a WFE and advanced-packaging equipment platform in the making.
“In our view, the market still underestimates ACMR’s transition from a China wet-cleaning leader to a broader WFE and advanced-packaging equipment platform.” — Charlie Chan, Morgan Stanley
Stock has already ripped 318% in the past year (market cap now ~$6.8B). Valuation is rich at 75x P/E, but the rate of change in revenue visibility is the key. The Q1 beat ($0.34 vs $0.28 est, $231M revenue vs $217M) gives momentum. Only watch-out: the recent $52M registered direct offering dilutes a bit, but Tekne Capital taking it down is a signal. Not sure we can read too much into the Shanghai subsidiary filings — standard STAR Market compliance. Light coverage, but this one has legs if the platform narrative gains traction beyond MS.
ROSENBLATT INITIATES AT BUY WITH $250 PT — THAT'S A 2.5X FROM HERE. They're betting on the purest AI infra play in the public markets, and the numbers are real. $100B backlog, 1 GW active power, 3.5 GW contracted. This isn't vaporware. Revenue doubling this year to $12.8B with a 77% CAGR through 2030, per Rosenblatt.
The bear case writes itself — $35.1B in debt, 7.39x D/E, and the stock is down 12% in the last week to $100.88. But Rosenblatt argues the value per unit of work stays constant while compute capacity per MW increases. They see pricing power, not just volume growth.
"CoreWeave is the largest dedicated AI-infrastructure provider, with over 1 gigawatt of active power and 3.5 gigawatts of contracted power... We forecast $12.8 billion in revenue this year, with the company doubling revenues year-over-year."
At 3.2x 2027 EV/Rev, the multiple isn't pricing in the backlog ramp. The debt load is real (that's the r/r here), but the cash flow profile changes fast if they execute. Initiation from a legit TMT shop with a bogey that implies conviction.
DLR IS THE CLEANEST PLAY ON THE THEME RIGHT NOW. Truist bumps PT to $225 from $208 (Buy), calling it a favorite in comms infra. The catalyst? A $1.6B transaction week (including earn-out) that hits every priority: hyperscale, colocation, and scaling the private capital platform. Stock at $193.84, up 27% in six months, inch from its 52-week high of $208.14.
The growth math is compelling: 10%+ Core FFO CAGR underpinned by an absurdly favorable supply/demand setup. Low vacancy, power constraints everywhere, and Inference is the incremental driver. Not just AI training — inference workloads are stickier and require geographic diversity. DLR is levered to that.
"Digital Realty is one of our favorite names in communications infrastructure coverage with 15% upside to the $225 PT, driven by a favorable supply and demand backdrop for data centers and momentum in both hyperscale and colocation segments, with Inference providing a significant tailwind."
Leverage at 4.7x — clean balance sheet, flexibility for incremental development. 23 consecutive years of dividends (2.5% yield) gives PMs cover for a "yield-plus-growth" narrative.
Bull case writes itself: AI compute demand is real. Inference is the second wave, and it's more colocation-friendly than training. DLR is the largest pure-play DC REIT with scale advantages in power procurement. The ~10% FFO CAGR is achievable even without heroic assumptions.
Bear side worth acknowledging: William Blair just cut its data center index score to 75 — power constraints and NIMBY opposition are real headwinds to supply. If new builds get delayed, DLR's development pipeline gets stretched. Also, 27% in six months means the easy money is in. Consensus is getting crowded.
Other firms are in the same zone — Stifel at $235, Scotiabank at $222. No one is stepping in front of this train. But the rate of change argument cuts both ways — if power delivery timelines slip, the re-rating stalls. For now, the signal is clear: DLR is the right horse for the AI infrastructure bet. Just watch the power narrative.
THE CALL IS BUY. Goldman Sachs initiates with a $34 PT (77% upside from $19.26), joining a chorus of bulls (TD Cowen, Cantor, DA Davidson, Stifel, all Buy/Outperform). The thesis: AI is accelerating legacy tech debt, and Braze’s ability to let marketers orchestrate sophisticated campaigns is becoming table-stakes. Business model improvements (67% gross margins, 27% revenue growth) give Goldman confidence in 20% operating margins by 2029.
“Braze is well positioned to continue taking share from legacy marketing tools as AI increases the strain of legacy tech debt within organizations.”
THE STICKY PART: Recent quarter was mixed. Subscription revenue missed by a hair, professional services beat by $6M. Non-GAAP op margin slipped to 5.0% (cons 5.2%, prior 7.1%). Guidance was the bright spot – raised top-line to +22% YoY (cons +20%) – and EPS printed $0.10 inline. Analysts are leaning on the enterprise win/market share narrative and AI monetization as the rate-of-change catalyst. Stock is -46% in 6 months, so the setup is cheap if the narrative holds. Bears would argue the margin trajectory isn’t there yet and competition from Salesforce/Adobe remains fierce. Right now the street is buying the story, not the ops – watch the next print for margin expansion proof.
Verdict: VRNS is the rare name where private equity buzz and fundamental upgrades are pulling in the same direction. Stephens goes Overweight from Equal Weight, PT to $45 (from $33) — and the trigger is Monday's M&A chatter (Blackstone, Thoma Bravo, Vista circled). But there's real operating leverage too.
SaaS ARR growth hitting 31% YoY last quarter (beat the midpoint by ~$7M) — that's the narrative driving the multiple. Stephens sees durable 20% SaaS ARR growth through FY27 and thinks guidance is sandbagged. Piper Sandler already at $47. DA Davidson at $37. KeyBanc started Sector Weight — cautious but concedes the Microsoft integration moat.
One line from the Stephens upgrade worth lifting:
"Demand for data security remains strong and is increasingly driven by artificial intelligence initiatives."
The PE angle is the wildcard. If a takeout happens, $45 looks low — but even without it, the r/r is decent for a name that just prints 31% SaaS NRR and trades at 6.9x EV/FY27 revenue. Not a home run on fundamentals alone, but the optionality is real.
Piper Sandler is doubling down on the FCF story. They reiterated Overweight with a $60 PT on a stock trading at $48.75. The range they see? $31 to $95 — that’s a fat r/r even allowing for near-term headwinds (indirect hardware exposure, renewal base friction). The market wants AI growth stories, but Piper is betting FCF estimates keep grinding higher.
“Piper Sandler sees a favorable risk-reward profile for the stock with a range of $31 to $95 per share, noting that free cash flow estimates could continue to move higher.”
The broader Street is on board after the Q3 beat. KeyBanc, RBC, Needham, BofA all raised or reiterated PTs in the $58-65 cluster. Consensus is constructive — term-license sales and bookings strong, guidance raised — even with supply chain noise hitting hardware. Net income expected to grow this year on 87% gross margins. Not a growth rocket, but the cash generation story is real. If the market pivots back to value/FCF narratives, NTNX gets re-rated fast.
IBM picking up incremental positive signal from the OpenAI Daybreak Cyber Partner Program. RBC reiterates Outperform, $300 PT (current ~$252). The thesis here isn't transformational — it's about adjacency economics. IBM is taking its consulting/governance muscle (Project Lightwell, $5B commitment) and layering in OpenAI's frontier models for real-time vulnerability detection. Smart, boring, and ownable.
"View[s] the announcement favorably as IBM continues to invest in security by focusing on adjacencies to its core competencies." — Matthew Swanson, RBC
The service runs through IBM Consulting Advantage — read-only, governed, enterprise-grade. Starts as point evaluations, scales to continuous monitoring. This isn't a new revenue line tomorrow, but it's a narrative off-ramp from "old IBM" into "practical AI deployment." 30-year dividend streak gives the base case. This adds an incremental bull hook.
Citizens lights a match under KEEL with Market Outperform and a $10 PT — 51% upside from here. The call is all about the pipeline: 2.2GW gross, with 341MW energized and 648MW secured. Miller’s math says full lease-up on the 1.7GW critical IT load is worth $44+/share. That’s the bull case — a data center developer disguised as a former bitcoin miner.
“Miller described Keel as a vertically integrated company transitioning from cryptocurrency to artificial intelligence infrastructure.”
The stock has already ripped 725% in the past year and 147% in six months, so the easy money might be priced. But Citizens is looking at 2028 EBITDA (28x) — this is a long-duration infrastructure story, not a momentum trade. H.C. Wainwright ($5.50) and Chardan ($4.50) are more conservative, but both still Buy. The redomicile to the US and shift to GAAP reporting should clean up the narrative. Pipeline execution is the only thing that matters from here.
Hold initiation with no price target — that’s a tell. AmerX straight-up says they can’t model this thing: $0.03M LTM revenue, -15% gross margins, and zero commercial sales expected before 2027. The $30M raise (5M+ shares at market) buys runway but dilutes the base. Risk/reward is a coin flip until someone signs a real production contract.
"The shares are fairly balanced between upside and downside risks based on available information."
Translation: there is no fundamental edge yet. Timing risk on B-TRAN commercialization is the whole debate — and until that changes, the stock is a binary options trade, not a compounder.
TEO GETTING DOUBLE UPGRADES — Morgan Stanley to Overweight ($17 from $13), JPMorgan to Overweight ($16 from $12). Both see the 11% sell-off on M&A remedy fears as overdone.
The thesis is simple: TEO acquired TMA in Feb 2025, but the stock sat flat vs MSCI Argentina — the market priced in zero synergies. Now that TEO is complying with regulator remedies, the path to moving TMA EBITDA margins from 29% (Q1) toward organic levels of 37% is clearer. Morgan Stanley explicitly moved M&A benefits from bull case to base case. JPMorgan adds that the antitrust review is nearing completion (final report expected May 2026 per La Nacion).
"By complying with regulator remedies, TEO can unlock M&A synergies and move TMA EBITDA margin from 29% in the first quarter closer to organic margins of 37%." — Morgan Stanley's Cesar Medina
R/r finally tilting positive after months of dead money. Not a home run, but a clean tactical upgrade in a market that was too cautious.
Take: UBS just ripped the band-aid off — PT to $670 from $455, Buy maintained. The call is all server CPU share, but the real bet is on x86’s role in agentic AI workloads. This isn’t just a GPU narrative anymore.
Arcuri is arguing that the standalone CPU rack (not tied to GPU clusters) is gaining traction. AMD’s core count and multithreading edge, plus x86’s software ecosystem, give it a structural advantage over ARM in the “traditional” workloads that underpin agentic AI. UBS now models a 60/40 x86/ARM split in standalone server — that’s a big swing (and a direct bet on Intel’s continued supply pain).
The numbers are aggressive: CPU server rev estimates for CY26-28 go from $16B/$21B/$27B to $16B/$23B/$29B. And CY30 jumps to $50B from $41B. That implies AMD is taking the lion’s share of a growing pie, not just share from Intel.
“The firm is more constructive on AMD as standalone CPU racks gain traction, given the company’s advantage in core count, multithreading, and x86’s software ecosystem for traditional software workloads that are part of agentic AI workloads.”
The 35x multiple is unchanged — so the PT bump is purely from revs, not multiple expansion. That’s disciplined, but it also means the stock’s r/r hinges on execution against a very optimistic share capture thesis. Last note: the Wolfe Research piece flagged capacity constraints at TSMC as the real share driver, not pure performance. Worth watching if AMD’s wafer allocation becomes a bottleneck.
UBS staying constructive here — reiterated Buy and $170 PT as DTM holds near its 52-week high ($152.88 vs $146.99). Thesis is simple: gas-fired reliability wins vs intermittent renewables as power demand surges. DTM’s pipeline footprint and direct connectivity to demand centers let it monetize incremental volumes tied to new generation buildout. Not a new call (PT unchanged), but the rate of change in power demand narratives keeps this name sticky for PMs looking for clean gas exposure.
"The firm cited the company’s positioning to benefit from natural gas-based power solutions as reliability and uptime remain priorities over intermittent power sources."
HOLD INITIATION FROM AMERX is the headline, but the real signal here is capital structure risk. No PT set, which tells you everything about their conviction. AmerX sees KDK needing to raise cash in 2027-28, and at 11.9x FY2028 revenue on just $4M in trailing sales, the premium is pricing in perfection on timing. Stock down 44% in six months — the market is already sniffing this out.
"AmerX expects Kodiak will need to raise capital to support growth in 2027 and 2028, which could pressure equity valuations over that period."
THE BULL CASE (Cantor Fitzgerald still Overweight, $12 PT) leans on Q1 EPS beat of $0.10 vs -$0.18 consensus and DaaS segment traction. But that’s a single quarter of positive surprise in a business burning cash into a multi-year horizon. Rate of change on revenue? Not enough to justify the multiple unless the 2028 ramp materializes exactly when they need it. Positioning feels fragile — light coverage, one cautious initiation, one stale price target cut. Not a buy-the-dip candidate until the balance sheet path is clearer.
No one's picking a fight here. UBS and Rosenblatt both Neutral, but the $10 spread between their PTs tells you how wide the range of outcomes is. CMCSA at $22.42 — basically 52-week lows — and the debate isn't about a turnaround; it's about how ugly the next few quarters get before the second-half "improvement" shows up.
Broadband is the anchor. Rosenblatt cut to $24 blaming connectivity headwinds. UBS's $32 assumes new pricing/packaging and the NBA lap plus political ads in H2 actually work. They just trimmed Q2 estimates — parks softer (Orlando/Japan attendance), EBITDA now seen -6.6% vs prior -6.3%. That's not a big miss, but it's the wrong direction. Content side is the offset: fewer NBA games means better comps, playoff ad revenue, Peacock hitting profitability. The UK theme park £5bn investment (opening 2031) does nothing for this year's P&L.
"We continue to model gradual improvement in the second half as new pricing and packaging at connectivity and the NBA deal is lapped while political advertising ramps." — UBS
That's the bull case. The bear case: broadband losses accelerate, parks stay weak, and the "gradual improvement" turns into a hope-and-a-prayer. At $22, the market is pricing in the bear case. UBS's $32 implies 43% upside — but that's based on a thesis that keeps getting pushed out. Not a lot of urgency either way.
Verdict: Sleepy 5-7% CAGR utility with a hidden growth catalyst that MS thinks pushes it to 7.5-8%. Stock at $48.37, target $50 — not a lot of upside baked in, but the optionality is real.
Morgan Stanley starts Equalweight, $50 PT. The base case is boring: 5-7% long-term EPS CAGR, management guiding to the upper half through 2028. That's fine, nothing special vs. regulated peers. BUT — the real story is the 1.8 GW of incremental generation needed by 2032, driven by a Google data center and other large loads in negotiations. Plus a big interstate transmission line with routing to be decided later this year.
MS thinks the extra capex from these projects adds at least 1% to the CAGR. Funny thing: consensus EPS already implies 7.8% CAGR through 2030 — ahead of management's 6-7% medium-term target. So the market is pricing in some of the upside before the projects are even finalized. If the timing/routing gets delayed or the large loads don't materialize, that consensus number gets sticky.
"OGE will be involved in building a portion of a significant new interstate transmission line, with timing and routing to be finalized later this year… several additional potential large loads in negotiations… could represent further upside." — Morgan Stanley
Caveat: Q1 EPS was a miss ($0.24 vs $0.37 est), but revenue smashed ($752.6M vs $613.6M). The miss might be noise — they're a utility, earnings lumpiness happens. The revenue beat is the numbet that matters for the load growth story.
Bottom line: OGE is a low-conviction name for PMs unless you want to play the AI/data center theme in a sleepy wrapper. The $50 target gives you ~3% upside. The real torque is if the transmission/generation capex gets approved sooner/bigger than expected — that could re-rate the multiple. But for now, it's a show-me story with a nice yield (19 consecutive years of dividend hikes).
DA Davidson doubling down on the ChatGPT ad thesis despite the broader PT carnage. Reiterates Buy, $24 target (45% upside from $16.57, which is basically the 52W low at $15.57). The key: 2,000+ BRANDS now advertising on ChatGPT via Criteo’s platform – double the 1,000 cited May 5. That ramp is accelerating, not stalling. Conversion rates beating other referral channels, CTRs consistently 2-3x HIGHER than comparable formats. Valuation is the safety net – 7.97x P/E, 21% FCF YIELD, InvestingPro health score "GREAT". Other shops (Benchmark, Bernstein SocGen, even DA Davidson themselves) trimmed PTs to $24-35 on reduced FY26 outlook and ad-spending headwinds (geopolitical, oil). But on the ChatGPT vector, this is a rate-of-change story that the PT cuts aren't capturing.
"Criteo reported that conversion rates continue to outperform other referral channels, consistent with commentary from early ChatGPT advertisers."
The bull case is simple: OpenAI is the new walled garden, Criteo is the traffic-monetization layer, and the brand pipeline is doubling every 6-8 weeks. Bear case: the core retargeting business is getting squeezed, FY26 guidance was cut, and $16.57 is cheap for a reason. For a PM with a 12-month horizon, the r/r tilts long if you believe ChatGPT ad spend goes from "experiment" to "budget line."
BOLD CALL FROM GOLDMAN. Initiated Buy, $26 PT — that’s 93% upside from $13.49. KVYO is down 58% YTD, sitting near its 52-week low ($12.53) after the CFO departure and the Q1 print that spooked the Street. Valenti sees a clean entry: high-20s revenue growth, 75% gross margins, multiple vectors (Shopify ecosystem, upmarket, international, non-eCommerce, new product cycles in Service and AI). The stock has been punished for visibility issues, not structural decay.
Consensus still cautious but not bearish. Stifel cut PT to $28 (from $35) but kept Buy — decel in guidance is the bogey. Piper Sandler went to $26 (from $30), Overweight, worried about margins. Canaccord stuck at $32, Buy, calling it “an underappreciated growth story in mid-cap software.” Management is leaning into “improved forecast precision” as they scale. That’s the narrative pivot.
“Goldman Sachs cited Klaviyo’s exposure to multiple growth vectors and revenue growth in the high 20% range. The firm expects the company to continue expanding with its existing customer base and within the Shopify ecosystem.”
The bear case: FY26 guidance decel, CFO uncertainty, margin compression scare. At $13, the market is pricing in a lot of bad news. The bull case: 30% grower at <5x EV / next-twelve-months revenue (rough math) with best-in-class eCommerce workflow stickiness. R/r is asymmetric — if the narrative stabilizes, this doubles. If it doesn’t, $12.53 support breaks and you’re looking at a $10 handle. Not for the faint of heart, but PMs who love mean reversion plays on high-quality SaaS should take a look.
UPGRADE TO BUY AT KEPLER – 2027 FCF INFLECTION IS THE HOOK. Kepler Cheuvreux just moved KPN to Buy from Hold, PT to EUR5.30 from EUR4.30 (stock at EUR5.03, ~4% div yield). Thesis: soft 2026 start but momentum builds. The Odido data breach is a tailwind for KPN’s premium/security positioning – lower churn as trust becomes currency. The real story is FCF: fiber capex peaks this year, so 2027 FCF grows ~25% YoY. That puts the stock on an ~7.5% FCF yield in ’27.
"The investment case is shifting towards the 2027 free cash flow inflection." — Kepler Cheuvreux
Not a barnburner, but the r/r works for a slow compounder with a catalyst in sight.
SERV is getting a fresh Buy initiation from AmerX with a $12.35 PT – ~93% upside from current $6.41 near 52-week lows. Thesis: Serve sits in two massive TAMs (food delivery, healthcare) but trades at a steep discount to autonomous software/delivery platform peers. AmerX uses a 7x FY28 revenue multiple (vs peer 9.3x) on their $132.6M estimate, implying the stock is undervalued.
"Serve is undervalued and that a price-to-sales multiple of 7.0 times its fiscal year 2028 revenue estimate is appropriate."
Not all roses – Freedom Broker just downgraded to Hold (PT $18) on dilution risk after the company raised ~$91M via equity offering. Two conflicting signals: initiation sees value at these levels, downgrade flags capital structure overhang. The market's at the low end of the $6.34-26 range – risk/reward tilts if delivery ramp continues, but dilution overhang is real.
UBS reiterates Buy and $220 PT after Chevron announces Project Kilby – a 2.7 GW co-located natgas power facility in West Texas dedicated to a Microsoft data center under a 20-year PPA. This is a structural demand unlock for CVX’s own gas output (fueled by local production), not just a merchant power play. The modular build (GE Vernova turbines + Solar Turbines) means capital is deployed in chunks, reducing execution risk. Stock at $175, 4% yield, $346B market cap – the multiple compression vs. the S&P is hard to ignore if oil stabilizes.
"Project Kilby is expected to deliver approximately 2.7 gigawatts, making it among the largest co-located natural gas power and data center developments within the U.S."
The bigger picture: CVX is also putting chips on the table in Greece (70% of an offshore gas block) and Argentina ($13.8B Vaca Muerta project). But the near-term headwind is real – the Iran war ceasefire sent oil down on the Strait of Hormuz reopening. Morgan Stanley sees Middle East production recovering 75% in four months. That's a tax on cash flow for the whole sector, but the power deal is a multi-year growth vector that doesn't depend on oil prices. R/r still favors the bull case on a total return basis.
Benchmark cut PT to $11 from $13 post-special dividend ex-date — maintains Buy. The stock sitting at $6.28 (implied upside ~75% to the target, not the consensus 5% bogey). Analyst Harrigan says the Russell/MSCI/S&P 1500 index-driven selling plus macro noise exaggerated the downside. His real thesis: the market is pricing in WAY too much Starlink fear.
"Cumulative Liberty LiLAC free cash flow through 2030 exceeds 150% of the company’s current equity market capitalization."
FCF YIELD OF 25%. That's not a typo. Q1 EPS printed $0.029 vs consensus -$0.0683 (142% beat) — revenue in line at $1.1B but flattish y/y. The Starlink risk is real but largely complementary on mobile; satellite broadband is a longer-term headache, not a near-term value destroyer. At this price the r/r is skewed heavily toward the bull case if management executes on the core LatAm/Caribbean footprint.
Verdict: BTIG kicks off coverage with a Buy and $290 PT, parking right in the middle of the growing GTA VI consensus. The play is simple — the Nov 19 release of Grand Theft Auto VI is the single biggest catalyst in gaming this decade, and the stock is still pricing it with a healthy dose of skepticism.
BTIG analyst Clark Lampen sees the title driving a "sustainable, multi-year improvement in earnings power" — projecting $10 in average EPS for FY27-29 (consensus is $6.60 for FY27 alone, so that's big). The firm points to historical multiple expansion during pre-release marketing cycles for Rockstar titles. TTWO currently trades at ~$242, market cap $45B.
"We’re launching coverage of Take-Two Interactive with a Buy rating and a $290 PT. Later this year, Take-Two is scheduled to release the next installment of its most important and commercially relevant global gaming franchise Grand Theft Auto VI (11/19 release date)."
The rest of the Street is already piling in — BofA went to $368 (bullish on GTA Online persistence), Piper Sandler at $280 with Reddit data confirming strong interest, DA Davidson at $300. Consensus PT cluster is roughly $280-300, with BofA as the clear outlier. The bull case: pre-order ramp + GTA VI itself drives a massive step-change in revenue, followed by GTA Online recurring tailwinds. Bear case? It's all in the timing — any slip past November kills the thesis quickly. But right now, the rate of change in sentiment is accelerating, not slowing.
KEPLER CHEUVREUX DROPS TO HOLD ON PURE VALUATION. Stock hit their EUR 8.00 PT after a 220% RALLY in 18 months — gap to peers is now closed. No obvious catalyst to push higher from here.
"None of these scenarios appears very plausible at present." — Kepler on Poste partnership, higher bid, or business plan surprises.
Downside risks: offer withdrawal or flow-back pressure. R/r skews symmetric at best. Pass.
Stifel cuts PT to $400 from $415, stays Hold – the margin unwind is real and it’s coming faster than consensus expects. The firm sees FY27 gross margins compressing ~450bps YoY to ~63%, which is more than 300bps below the street at 66.5%. That’s not a blip – that’s a structural mix shift to Azure (growing 3x faster than the rest) plus accelerating capex eating into gross profit. Stifel models 100-150bps of QoQ Azure gross margin compression in FY27 alone.
"Even if Azure gross margin remains stable at Q4FY26 levels of ~47.5%, FY27 gross margins would still decline ~300bps YoY."
The offset? Opex efficiencies – headcount declines, mid-to-upper single-digit opex growth guidance. But that’s a band-aid on a compound fracture. The catalyst for this reexamination? Oracle’s recent print and margin guide. Not sure we can read too much into a single competitor data point, but the trendline is clear: the cloud buildout is expensive and the margin payoff is back-loaded. At $365.46, just above the 52-week low of $356.28, the stock is pricing in some of this – but not 300bps of consensus error. Hold is the right call until the sell-side models catch down.
Verdict: UBS raises PT to $92 from $80, Buy maintained — cloud usage trends accelerating, AI token optimization a non-event for revenue. FROG + Snowflake + Datadog = top 3 infra longs among investors. This is real demand, not narrative.
Check from UBS analyst Radi Sultan comes after in-person meeting + industry contact calls. Key takeaway: Q2 cloud usage improving, and AI token optimization fears are a nothingburger for FROG’s top line. Nearly every contact cited healthy spending growth on core Artifactory. No competitive pressure surfacing.
“Nearly all contacts pointed to healthy spending growth driven by core Artifactory usage.”
UBS modeling 25% revenue growth in FY26, then 23% in FY27. Gross margins at 77% — not a concern. Cantor Fitzgerald reiterated Overweight at $80, TD Cowen raised to $100 and added FROG to Best SmidCap Ideas list — both citing AI tailwinds. Russell 3000 addition effective June 26 is a mechanical tailwind for flows, not thesis-changing but clean support.
The bull case writes itself: strong demand, multiple AI vectors (especially from ML model management / artifact distribution), no displacement pressure from DevSecOps consolidation. If ARR re-acceleration materializes, this stock has room above $100. Bear case? Valuation — InvestingPro flags potential overvaluation at current $78.71, and 21 EPS revisions up doesn't mean multiple expansion is safe. But for now, the rate of change on fundamentals is positive and the narrative is gaining converts.
Wedbush is sticking with PLTR. Outperform and $230 PT, even as the stock sits at $116.70 — basically the 52-week low after a 40% six-month slide. The catalyst du jour is the partnership with Zeta on AI-powered marketing infrastructure. Expected to generate $100M+ in revenue for Zeta (multi-year), the tie-up connects operational and customer intelligence inside agentic AI. Wedbush’s thesis: enterprise AI is still early innings and PLTR’s 84% gross margins plus 68% revenue growth prove it.
“We believe many do not fully understand the company’s capabilities.”
UBS and Wolfe also chiming in (Buy $200, Peerperform upgrade) — so the Street isn’t running for the exits despite the price action. Risk/reward looks asymmetric if you buy the long-term platform story.
Verdict: B.Riley stays Buy at $250. The thesis is simple — ad stack investments are compounding. The Cannes Lions update showed ROAS above industry averages in key verticals, and the numbers back it up: REVENUE SURGED 71% TO $2.47B LTM with a 91% GROSS MARGIN. At 15.9x FY27 EV/EBITDA, the r/r still works if growth sustains.
"Reddit’s latest update... indicates management’s investments in the ad stack are delivering a boost to ad performance."
Other analysts are mostly on board too — DA Davidson and Citizens both positive (DA pointing to 20.3% DAU growth vs 18% consensus, Citizens highlighting 129% app install growth and 15% lower acquisition costs). Only BofA sits neutral at $190, flagging the AI data licensing value but not enough to get bullish at current levels.
Bear case? Meta launched "Forum" — a direct Reddit clone for Facebook Groups. That's a risk to watch, but not a near-term derailer. Reddit’s shopping integration (20% of shoppers use Reddit in searches, 50% of US shoppers verify AI recs on Reddit) gives it a unique utility that’s hard to replicate overnight. Keep the position.
Verdict: KBW pounding the table on NCNO at 12x EBITDA and a 52-week low is the kind of contrarian signal that gets attention. They see accelerating revenue growth from the platform pricing transition plus direct monetization of vertical AI in banking. The market is pricing in fear of AI disruption — management says that's disconnected from actual engagement and pipeline.
Stock is $14.57, down 44% in six months, nearing its 52-week low of $13.80. KBW points out it trades at 12x FY28 adjusted EBITDA and 16x FCF (both after SBC). For a company posting 11% revenue growth (Q1 FY27 print of $159.4M vs. consensus $155.8M) and accelerating from 6% last quarter, that multiple is pricing in a lot of pain.
"Management highlighted a disconnect between market concerns around AI disruption and the strong engagement and pipeline activity the company is experiencing."
Citizens reiterated Market Outperform at $23, JPMorgan bumped target to $17 on AI defensibility overblown. Street broadly constructive but waiting for execution proof. The agentic banking platform adoption is the hook — KBW is buying the narrative that NCNO owns the vertical AI opportunity in banking, and the current price is the entry point for that thesis.
AXIS CAPITAL DOWNSHIFTED WIPRO TO REDUCE – slashed PT to INR160 from INR220. The call is straightforward: acquisitions haven’t delivered, organic growth is falling behind, and the FY1Q27 guidance implies more pain. Stock already down 22% in a year, but the PEG ratio screaming 43.38x tells you earnings aren’t coming to the rescue anytime soon.
The thesis hinges on capital allocation and execution risk. Capco and other deals didn’t generate the client mining lift expected. Now financial services and consumer verticals are rolling over in FY26, and the revenue decline guide for next quarter suggests Wipro will lag its peers into FY27.
“Major industry segments including Financial Services and Consumer have deteriorated in fiscal year 2026. Wipro’s revenue decline guidance for the first quarter… indicates organic revenue growth will likely trail competitors.”
Buyback (₹15,000cr at ₹250/sh) isn’t moving the needle – it’s a Band-Aid on a structural growth problem. The acquisition of Alpha Net contracts (undisclosed terms) adds more execution noise. Hard to see a catalyst here until the organic story stabilizes.
STIFEL GOES BOLD – BUY $12 ON THE ARM OF PHYSICAL AI. Market is still pricing BB as a legacy auto supplier. Stifel says that’s wrong. The shift to runtime royalties, 76% gross margins, and partnerships with NVDA/QCOM/AMD make this a capital-light royalty annuity story reminiscent of ARM. Stock up 133% YTD, but Stifel sees 36% more upside from $8.82.
"The company is no longer an auto-supplier story but rather a mission-critical software layer in the physical AI stack."
Rest of the Street less excited – Baird Neutral at $5, Canaccord Hold at $4.40. The divergence is the trade. Rate of change on margins and revenue quality is accelerating. If the margin expansion thesis plays out, the re-rating has legs.
MU — The print resets the memory cycle math. $41.5B revenue, $50B Q4 guide, and — crucially — a $100B floor on multi-year LTSAs. That's not incremental; that's structural. 84.9% gross margins on what was a commodity six months ago. The debate shifts from "is this peak?" to "how long can the shortage last?" Supply response takes years (Idaho, NY, Japan fabs). Agentic AI workloads caching execution states in DRAM expand memory intensity per task. The bear case is supply catch-up; the bull case is visibility extends through 2027 at 80% higher HBM prices. Market hasn't fully priced the duration shift.
QCOM — Raised FY29 datacenter revenue target to $15B. That's a 2x increase from $22B total non-handset to $40B. Bold for a company whose DC revenue is effectively zero today. The C1000 agentic CPU with Meta as first customer validates the pivot. Two additional custom wins (Amazon speculated). The narrative changes from "can they?" to "how fast?" — but 5-year targets have low visibility. Stock +13% AH on the confidence, not the numbers. Integration risk on $10B+ M&A spree is real. Key question: can the assembled stack (Ventana, Alphawave, Tenstorrent) cohere into a platform?
AVGO — OpenAI's "Jalapeño" tape-out in 9 months is the signal. 50% cost savings vs merchant GPUs for inference. Hock Tan now designs custom chips for both Google (TPU) and OpenAI — universal custom silicon partner. MSFT likely becoming 7th XPU customer. The ASIC acceleration thesis gets validated in real time. Stock only +2% AH — market may not be pricing the strategic accretion. Custom margins are lower than merchant, but revenue growth and relationship depth compound. The multi-generation roadmap (next version 2028) locks in design cycles.
FCEL — 380MW on-site power deal for DCs is the play. Behind-the-meter generation is the only way to bypass the 3-7 year grid queue. Stock +16% on the news. The thesis moves from "if" to "at what scale and margin." Execution risk in building fuel cell manufacturing capacity remains. But the deal confirms power is the next binding constraint and operators will pay for dedicated solutions. Competing against Bloom Energy, natural gas, solar+storage. FCEL's clean baseload pitch is differentiated for 24/7 DC ops.
NVDA — The bear case is building. OpenAI's 5o% cost savings, QCOM's agentic CPU, Cerebras AWS deal, Google TPU — all point to custom silicon fragmentation for inference (70% of AI compute spend). Black market prices in China doubled (B300 from 4M to 8M RMB) — demand still extreme but not adding to reported revenue. H100 lease rates rebounded 40% since October — merchant GPU demand robust. NVDA underperforming semis by widest margin in 2.5 years. Training remains a monopoly. Inference share is the debate.
META — Zuckerberg reshaping surface area for agents to operate on — recreating the internet via wholly-owned first-party inventory. Long-term architecture play. Glasses going viral — "Wall Street is asleep" per social arb. AI driving $20-25B incremental revenue at 90% margins (reaffirmed). SHOP data point (all-time low quota attainment on AI-first GTM) shows legacy sales motion struggles, but that's a different business unit. Capex growth without near-term revenue is the concern. But Zuck is playing for the next decade.
GOOGL — Brain drain is accelerating. Jonas Adler and Alexander Pritzel (Gemini core) leaving for Anthropic. Follows Noam Shazeer and John Jumper. Top researchers choosing equity upside and mission over Google stability. Product advances (Gemini 3.5 Flash computer use) are fine, but the innovation engine may decelerate if key architects keep leaving. This is a people problem, not a capital problem — harder to solve. TPU and infrastructure remain strengths, but model-side talent is the moat.
AMZN — AWS combining Trainium3 (prefill) with Cerebras CS-3 (decode) for disaggregated inference by 2027. Validates the split-workload thesis — different chips for different inference phases. Differentiates AWS on latency. Reduces NVDA dependency for inference. Long-term thinking consistent with custom silicon strategy (Graviton, Trainium). 2027 is distant, near-term revenue is GPU-based. But the architecture signals where inference is heading — specialized, not general.
INTC — Using advanced packaging (EMIB, Foveros, hybrid bonding) as a side-door attack on TSMC. Rather than compete on leading nodes, capture silicon value in HBM4 base dies where packaging dictates performance. Server CPU shipments expected to grow 30% in 2027 — recovery from AMD share loss. Pelosi disclosed INTC calls — "Pelosi effect" auto-tracking triggered $500M+ in mirrored trades. Retail momentum. Stock more than doubled YTD. Packaging narrative provides a new growth angle. Fundamentals improving but still challenged.
TSM — Raising all advanced-node prices 5-10% for N7 and below. "Find ways to charge more" per senior leadership. Confirm tight capacity and pricing power. Margin positive for TSMC, headwind for chip designers. CoWoS remains bottleneck — SPHBM4 standard could bypass interposer, easing packaging constraints but reducing TSMC packaging revenue. MediaTek investing in GUC for Google's 3nm Axion. TSMC remains foundry of choice. Price hike signals confidence in demand durability.
ASML — SK Hynix $29.4B capital raise includes EUV tool purchases. Backlog visibility strong. MATCH Act in NDAA could ban immersion DUV to China — ~15% of revenue at risk. Netherlands lobbying against expansion. EUV monopoly intact. The only risk is TAM reduction from export controls. Otherwise, demand from leading-edge fabs is durable. Regulatory binary outcome is the swing factor.
BABA — Anthropic publicly accused Alibaba of the largest distillation attack — 28.8M Claude accesses via ~25K accounts. First time a US company has called out a Chinese company at scale. Potential US policy response (export control expansion). Stock -3%. The implicit admission that Chinese labs need to extract capabilities validates US leadership. But distillation helps close the gap faster. Tail risk is real — policy timeline unclear. Diversion from core business.
AAOI — AMD reportedly negotiating for CW laser chips and ELSFP for optical connectivity. If real, AAOI becomes a supplier to AMD's AI infrastructure. Price target range $57.50 - $220 — extreme uncertainty reflecting binary outcome. Optical networking booming on 800G/1.6T DC interconnect demand. High-beta play on CPO and silicon photonics scaling. Competition intense (Coherent, Lumentum). Rumor-driven volatility typical.
TSEM — $290M deposits for silicon photonics foundry, $1.3B total booked. Customers prepaying for capacity before fabs are built. Extreme scarcity in optical component manufacturing confirmed. Positions TSEM as critical enabler of AI networking boom. Prepayments demonstrate inelastic demand and pricing power. Margin profile likely attractive. Execution risk is building to spec on time. Competitors exist (Intel, GlobalFoundries) but prepayments size signals competitive technology.
SNDK — AI servers overtaking smartphones as largest LPDRAM demand source per TrendForce. Elevated consumer pricing the new norm. Capacity routing to high-margin AI. Micron's DC SSD revenue >$5B (2x sequentially) confirms AI driving NAND demand. SNDK pure-play NAND. Positive but NAND is more commoditized than HBM — supply response faster, margin expansion less extreme. Stock more than doubled YTD. AI memory theme driving multiple expansion.
WDC — NAND and HDD beneficiary. Micron's DC SSD data shows demand. HDD nearline for AI data lakes also benefits. Restructuring to focus on flash and HDD. Prices firming across both segments. Competing with Seagate in HDD, Samsung/SK Hynix/Micron in NAND. Strategic review underway — potential separation of HDD and Flash. AI tailwind supports both. Valuation reflects the cycle.
STX — HDD pure play. AI data lakes require exabyte-scale nearline storage. HAMR technology differentiator for 30TB+ drives commanding premium. AI buildout is multi-year tailwind. HDD pricing stable to improving. Duopoly with WDC. Risk of flash substitution for performance tiers. HDD remains cost-effective for cold storage. Stock more than doubled YTD. Upside capped by cycle pricing.
WIT (no specific new signal from feed)
Hearing MSFT may become Broadcom's 7th XPU customer. Currently deploying OpenAI's Jalapeño chip in own DCs for OpenAI workloads. If performance holds, MSFT evaluates for proprietary deployment by 2028-2029. Not in any model.
Word is Anthropic accused BABA of "largest distillation attack" — 28.8M Claude accesses via ~25K accounts April-June. First public accusation by US company against Chinese company at scale. Policy response timeline unclear but tail risk real.
Channel checks suggest AMD negotiating with AAOI for high-power CW laser chips and ELSFP for optical connectivity. If confirmed, AAOI becomes AMD supplier for AI infrastructure. Binary outcome — not in numbers.
Hearing Tower Semiconductor (TSEM) took $290M deposits for silicon photonics foundry, $1.3B total booked. Customers prepaying for photonic chip fabrication slots before fabs built. Extreme scarcity in optical component manufacturing.
Word is SHOP experiencing all-time low quota attainment on AI-first GTM mandate per field checks (N=962). AI product-market fit strong but agentic sales workflows struggling to land. Not in street models.
Channel checks suggest Nantero (NRAM) is gaining traction in bespoke chip applications for AI inferencing on edge. Custom silicon orders supporting 12+ month backlog. Not yet at revenue recognition stage but growing.
Word is Cerebras (CBRS) hit post-IPO low after admitting US shortage of operating datacenters. Orders mean nothing without energized shells. Physical deployment lag is the binding constraint.
Hearing Black market NVDA B300 prices in China doubled from 4M to 8M RMB. Confirms unmet demand despite export controls. Doesn't hit reported revenue but signals pricing power.
Channel checks suggest TSMC raising all advanced-node prices 5-10% for N7 and below per senior leadership directive to "find ways to charge more." Tight capacity confirmed.
Word is Pelosi disclosed INTC calls — "Pelosi effect" auto-tracking triggered $500M+ in mirrored trades. Insider signal, but retail momentum driver, not fundamental.
Hearing FuelCell 380MW deal with Fit Energy is just the first of multiple behind-the-meter DC power announcements expected in coming quarters. Others in the pipeline unnamed.