CV Research

$QCOM just told customers prices are going up by double digits, effective September 1. Here's what that actually means for the business. Start with the mechanics. Qualcomm sent a letter to its entire customer base, per Bloomberg, raising prices on products shipped after September 1, citing supply chain costs. The driver is memory. AI data centers consumed the world's DRAM supply, memory prices exploded, and every company building devices is absorbing it. HP's CFO said memory and storage went from 15-18% of a PC's bill of materials to roughly 35% this year. Qualcomm is passing that through. What it means for revenue: a double digit price increase across the handset chip portfolio flows almost directly to the top line, because volumes in premium tiers are relatively inelastic. Phone makers can't design out Snapdragon in a product cycle. The chip is qualified, the software stack is built around it, and the launch dates are fixed. That's why the increase gets announced with a 5-week lead time instead of negotiated over quarters. What it means for margins: and this is the part that matters, the question is whether the hike offsets input costs or exceeds them. If Qualcomm is raising double digits to cover a single digit cost increase, margins expand. If they're covering cost dollar for dollar, revenue rises while margin percentage compresses even as gross profit dollars grow. Watch gross margin guidance on the next call, that's the tell on whether this is defensive or opportunistic. The bigger signal is pricing power. You don't raise prices across an entire customer base unless customers have nowhere to go. And the timing says everything, days after Qualcomm swept Samsung's entire Galaxy lineup, phones, watches, and the new intelligent eyewear. That's a company with the leverage to reprice into a supply crunch. The risk is real too, higher chip costs pressure OEM build plans, and Chinese manufacturers have already been trimming volumes on memory costs. If unit demand drops more than price rises, the math flips. Watch handset shipment forecasts alongside the pricing. Net read, revenue tailwind, pricing power confirmed, margin direction is the open question. The AI memory shortage is now repricing the entire device supply chain, and Qualcomm has the position to pass it on.

$RKLB is attempting something only SpaceX has ever pulled off, owning every layer of the space economy at once. The rockets. The satellites. The components inside them. And now, with the $8 billion Iridium acquisition, the layer that changes everything, spectrum and recurring revenue. Peter Beck told you the plan himself. Rockets are valuable, but spectrum and recurring services are what turn space infrastructure into a real platform. That's the founder of a rocket company saying the rocket isn't the prize. Look at the shape of what's coming. Revenue at record highs for five straight quarters, up 63%. A $2.2 billion backlog that doubled in a year. Over $2 billion in cash. Neutron targeting its debut in Q4, with the second stage having just passed a full duration burn and flight hardware arriving at the launch site. And on the street's own models, free cash flow swings from deeply negative to positive $222M by 2028, the construction bill ends, the rocket starts billing, and the Iridium rent starts arriving, all inside the same window. But this isn't a clean story, and anyone selling it as one is hiding the file. Neutron has missed four dates. The Iridium deal has a collar, a spread, and a financing window nobody can underwrite from here. The founder's financial promises always land, and his calendar promises almost never do. Knowing which ledger you're betting on is the entire position. We wrote the whole thing up. The deal mechanics nobody's actually read, the tripwire that tells you if this becomes transformation or just accretion, the honest Neutron grade against every peer rocket in history, and the exact dashboard to watch into earnings August 5. Full deep dive below.

@xMarketPlumber Hey Chris, There is something called "Skin in the game" on autopilot which is our personal capital that we allocate to the fund. We currently have capital in there and plan to add consistently.

@CVResearch_ I have a question about this fund Does this mean the fund “owners” have no actual capital invested and just provided you with a paper portfolio?

The CV Research portfolio is now live on Autopilot. You can connect your brokerage and automatically copy the portfolio. Every entry, every exit, mirrored in real time. If you’ve ever said “I wish I could just copy the port,” this is literally that. The thesis behind it: we’re living through one of the largest infrastructure buildouts in history, and the market keeps mispricing the layers underneath it. Everyone crowds into the obvious names while the companies that actually supply the buildout, the interconnect, the optics, the power equipment, the compute landlords, trade at a fraction of their role. CV Research positions one tier below where consensus is looking, in the layer beneath the layer the market just discovered. We track how the bottleneck rotates through the stack, compute to memory to interconnect to power, and position one rotation ahead. And when this buildout matures, the mandate rotates with it, robotics and physical AI are next. Every name enters after a full fundamental teardown, with the thesis written down before a dollar goes in. We target monopolies, sole qualified suppliers, and toll booth businesses that get paid no matter which architecture wins. And the discipline is the real product. Positions exit when the thesis breaks or completes, never because the price dipped. No panic selling, no chasing, held through the volatility, because that’s where the returns come from. You get the whole CV Research system, not just the tickers. Live now. marketplace.joinautopilot.com/landing/5377/1…

$QCOM just swept the entire Samsung Galaxy lineup, confirming the thesis we laid out. At today's Unpacked, Qualcomm announced that Snapdragon is powering everything Samsung unveiled. The Galaxy Z Fold8 Ultra, Z Fold8, and Z Flip8 run the Snapdragon 8 Elite Gen 5 with the world's fastest mobile CPU. The Galaxy Watch9 and Watch Ultra2 run Snapdragon Wear Elite, Samsung's first smartwatch platform ever powered by Snapdragon. And the new Intelligent Eyewear runs Snapdragon AR1. Phones, watches, and glasses. One chip vendor across the entire ecosystem. Two details matter most. First, the watch win is new territory. Samsung has never built a Galaxy watch on Qualcomm silicon before. That's a socket taken from their own in house chips, in the device that sits on your wrist capturing signals all day, exactly where "Personal AI" lives. When your biggest Android partner hands you a category they used to keep for themselves, the moat is widening, not holding. Second, the eyewear is the sleeper. Qualcomm, Samsung, and Google are expanding Android XR into smart glasses, with Gentle Monster and Warby Parker designs launching this fall. Glasses are the first genuinely new device category in a decade, the one Meta already proved demand for, and Samsung's entire entry runs on Snapdragon. Amon's framing says it plainly: intelligence built into the devices people rely on all day. Every one of those devices needs high performance, low power AI silicon, and every new form factor keeps making the same choice. Handsets fund it. Wearables and glasses extend it. The same compute DNA scales into cars, PCs, data center inference, and robotics. The full stack keeps compounding.

New Deep Dive dropping soon!! Any guesses?

New Pilot Alert 🚨 @CKCapitalxx built 85,000 followers on X finding stocks nobody was talking about yet He runs @CVResearch_, which specializes in reading filings and going down the supply chain to find the companies supplying all the major tech companies. Full deep dives on Substack for his members He just brought that strategy to Autopilot Top holdings: • Credo Technology $CRDO *Makes the chips that connect everything inside AI data centers • Applied Optoelectronics $AAOI *Builds the hardware that moves data between AI servers at scale • Nebius Group $NBIS *Rents out GPU power to AI companies • Synopsys $SNPS *The software every chip company uses to design its chips Check out all 10 holdings below

Kimi just paused new subscriptions because they can’t get enough compute to serve demand. The hottest model in the world is rationing paying customers. That’s not a demand problem, it’s a supply wall, and it flows straight down to both of our recent deep dives. $FPS. Every GPU that gets deployed to fix this shortage needs power delivered to it, and that means switchgear, transformers, and distribution gear. Forgent’s backlog already jumped $400M in two months before headlines like this. The compute shortage is a power equipment shortage one layer down. $CRDO. More racks getting built means more links inside them. And the open-weights angle makes it better: when Kimi’s weights drop on the 27th, thousands of companies self host on rented GPU clusters, and every one of those clusters is wired with the exact copper and optical connections Credo sells. They don’t care whose model wins. They sell the wiring. One headline, both theses confirmed. The shortage isn’t slowing the buildout. It’s the reason the buildout accelerates. Both deep dives are up if you haven’t read them.

$QCOM might be the most complete stack in semis, and the market still prices it like it’s not. Look at what’s actually stacked inside this thing. Today’s cash machine: Snapdragon runs the premium Android world, and the licensing business collects a royalty on essentially every 5G device sold on earth. That’s the engine that funds everything else, throwing off billions in cash a year. Automotive: the Digital Chassis has quietly built a multi-billion dollar pipeline. Cars are becoming computers on wheels and Qualcomm is the compute inside a growing share of them. PCs: Snapdragon X pushed them into laptops, taking a real seat at the AI PC table against Intel and AMD. Data center: this is the new leg. Their AI inference chips are aimed straight at the fastest-growing workload in compute, with rack scale systems coming. Inference is where AI economics actually live long term, and Qualcomm’s whole DNA is performance per watt, the exact metric that decides inference. And then the one nobody models: robotics. Humanoids need exactly what Qualcomm has spent 30 years perfecting, high performance compute at low power, on device AI, and connectivity, all in a small thermal envelope. A humanoid robot is closer to a smartphone with limbs than a server with legs. When that market ramps toward the trillions banks are projecting, the edge compute inside it is Qualcomm’s natural turf. Phones fund it. Auto and PC diversify it. Data center scales it. Robotics is the free call option on top. Every layer of the future needs a chip, and Qualcomm has a product aimed at all of them.

$CRDO's board published its own price target in an SEC filing last month. Almost nobody read it. Buried in a June 8-K: a one-time CEO award. 1.44M shares that only vest as revenue climbs from $2.5B to $7.5B and the stock clears hurdles from $244 to $489 by 2031. The CEO gets nothing unless shareholders triple first. Boards don't structure comp like that for companies they expect to grind. Consider who built this. The CTO spent 11 years leading analog design for $MRVL's storage chips, recovering impossibly faint signals billions of times per second without errors. Same physics as the engine inside every Credo cable. He didn't pivot into a hot market. He transplanted a mastered craft. His stake: $1.3 billion. The seed check came from Marvell's own co-founder. $3M, now worth about $1 billion. Lip-Bu Tan, now running $INTC, led the round. Founded, funded, and governed by Marvell royalty. Today, Marvell is chasing them with a copycat cable program. The students became the masters. The smart money kept coming. To win $AMZN in 2021, Credo granted it a warrant on 4M shares at $10.74. Fully vested, now worth ~$1B. Amazon didn't just pick its cable supplier. It took a billion-dollar equity stake in the outcome. And the setup nobody connected: the optical supply chain is short of lasers through 2027, bottlenecked by a substrate under Chinese export controls. $CRDO just spent $1.3B on a photonics architecture that needs far fewer lasers, and its FY28 cable line replaces them entirely with microLEDs made outside the bottleneck. In a two-year famine, the vendor selling the workaround collects scarcity rent. The famine is documented everywhere. The connection is not. The connection is the trade. $NVDA is the one logo missing from the customer list, and that's exactly why the Street can't model this company. Revenue tripled. Guided +80% again. 68% gross margins, zero debt, half of revenue becoming profit. Full deep dive out now! 👇

$FPS might be the only AI trade that asks you to predict nothing. You don’t need to pick the winning model, chip, or cloud. Every one of them ends at the same wall: power comes off the grid at hundreds of thousands of volts and a chip drinks it at one. Somebody has to build the machines in between. Forgent builds all of them, in America, faster than anyone else can. Revenue up 103% last quarter, all organic. For every dollar of gear shipped, customers ordered two more, three quarters running. Backlog near $2 billion, up 157% in a year. And it trades around $44, below the price of a stock deal that closed Monday. A company growing like this priced under its own offering. The market is pricing the seller. Policy is pricing the asset. Full deep dive link down below.

Deep Dive coming soon on retails favorite humanoid stock right now. Can you guess what it is?

$QCOM investor day delivered, and it validated the core of the thesis we laid out in our deep dive. Here is what they actually said. The headline. Qualcomm raised its fiscal 2029 revenue target from $22 billion to roughly $40 billion. Eighteen months ago they guided FY29 to $22B. Now it is $40B, and the data center is the main driver of that jump. They also guided to more than $18 in non-GAAP EPS by fiscal 2029, which is even higher than the $15.50 to $16 bridge most bulls were modeling. The number that moved the stock. A data center revenue target of more than $15 billion by fiscal 2029, built from essentially zero today. The stock jumped over 10% after hours on it. For context, BofA went into the day arguing that even $10 billion by 2028 was already priced in. Management just guided well above that bar. The product reveals backed it up. The new Dragonfly C1000 CPU is a 250+ core chiplet design running at 5GHz with PCIe Gen7, CXL, and LPDDR memory. That last part matters. It confirms the no HBM, LPDDR based architecture Qualcomm has been pitching as its cost and supply edge, sidestepping the expensive memory every other accelerator fights over. And the marquee moment. Mark Zuckerberg confirmed $META entered a multi generational agreement to use Qualcomm’s CPUs. A named hyperscaler committing across multiple generations is exactly the win the socket, hold it for years playbook that took Qualcomm into the car. Here is the one thing they still did not do. They gave revenue and EPS targets, but they did not put a clean gross margin number on the data center business. That was the single most important number in my deep dive, the quality of those data center dollars, and management still would not disclose it. So the bull case on revenue got stronger while the one real question, whether the market keeps paying a premium multiple as the mix shifts, is still open. Net read. The revenue story is now concrete, not a slide. $40 billion FY29, $15 billion of it data center, $18 plus EPS, and a multi generation Meta CPU deal. The thesis is playing out. The margin question is the only thing left to answer, and that is what decides whether this re-rates or grinds.

Full deep dive. cvresearch0.substack.com/p/qualcomm-dee…

Six months ago Qualcomm was a stock everyone had written off. A boring phone chip company you trade around the handset cycle and forget. Today it is quietly shipping AI racks to a Saudi sovereign and building custom silicon for ByteDance, and almost nobody has repriced it. $QCOM might be one of the most mispriced large caps in all of semis. Underneath the phone chip label sits a patent toll booth running over 70% margins that collects a royalty on nearly every 5G device on earth. The market prices it like it dies the day Apple ships its own modem. The contracts say something very different. Then stack on a record auto quarter, a Saudi sovereign scaling toward 1.9GW, a $20 billion buyback, and a data center business the market is barely pricing in because six months ago it was just a slide. I broke the whole thing down in my latest deep dive. Why the toll booth survives Apple, the one number management refuses to put on a slide, and what June 24 actually reveals. $QCOM Full breakdown on my Substack, link in below and in bio.

New deep dive dropping next week. Can any of you guess what stock it is??

Lovely trading day for $STM.