Chris Camillo made the most bullish case on $AMZN. Everyone worried when Amazon raised their capex to $200 billion Camillo believes they might make that $200 billion back on their Anthropic investment alone when Anthropic IPOs. What Amazon has done in the last 18 months is remarkable in scope. They didn't try to win the foundation model race against OpenAI and Google directly. They went a completely different direction, positioning themselves as the infrastructure underneath everything that is AI. Trainium chips, cloud buildout, advertising integration and equity stakes in two of the three most important AI companies on the planet. First Anthropic. Now $50 billion into OpenAI on top of that. And that OpenAI investment isn't just a financial bet. It ties OpenAI into Amazon's cloud ecosystem, probably permanently. Two of the big three AI labs now run on Amazon infrastructure. Google is the only one that doesn't. The market has spent two years debating who builds the best model. Amazon decided to own the plumbing those models run on and bought equity in the winners while they were at it.

$LLY has joined the trillion-dollar club and the market still hasn't figured out how to price Eli Lilly correctly (Save this). They are pulling in roughly $65 billion in annual revenue with about $22 billion in earnings. The old mental model treats Lilly as a straightforward weight loss drug driving revenue. But they're actually taking the revenue from their thriving drug business and investing it into various AI plays: - Built LillyPod, its own AI data center with 1,000+ NVIDIA GPUs - Launched TuneLab, an AI platform for drug discovery - Partnered with Google's Isomorphic Labs (Google's AI research division) - Committed $75B to acquisitions and partnerships since 2023 A stock already at a trillion dollars in market cap on $65 billion in revenue is not a cheap stock on a simple multiple basis. But we believe that the market is still treating Lilly like a one-dimensional pharma company. Bullish on Eli Lilly as a multi-layered growth story.

Neoclouds like $CRWV & $NBIS are the most durable businesses according to @GavinSBaker. The reason CoreWeave can charge a significant premium for its GPU hours is that not all GPU hours are the same. CoreWeave's GPUs are being utilized two to three times more per hour on average than a lower-quality provider's. That utilization gap is the product for $CRWV. In 2005, all you had to do to build a $50 billion market cap retailer was run 1,000 stores across 50 states with different climates and different consumer preferences. That's all you gotta do but in all of American history, roughly only ten companies have managed it. Running a GPU cluster at CoreWeave's utilization rates is harder than that. And don't forget about their impressive customers. $CRWV have over $63 billion in disclosed customer commitments this year alone: - OpenAI alone committed over $21B through 2031 - Meta committed $35B+ - Nvidia signed a $6.3B deal as a paying customer - Jane Street dropped $6B in April. - Anthropic signed a multi-billion dollar deal as well

Cathie Wood: "$SPCX could become the most important company in global history." Our PRO team called this exact price on SpaceX before the stock ever got here (Save this). Back on July 13 with $SPCX still above $139, our PRO team's Kyle put out a prediction. His buy zone sat around $115 and he said the stock would likely keep chopping into the August unlock before that level showed up. SpaceX closed at $118.24, down 47% from a June peak above $225. Part of this drop is ordinary post IPO unlock jitters. August 6 brings an insider unlock that could double the tiny public float. @CathieDWood continues to buy the dip in $SPCX. ARK has bought more than $475 million of SPCX since the IPO across its ETFs buying on the way up and adding again as the stock fell through the $119 to $154 range. That position now sits $60 to $85 million underwater given how far the stock kept falling after ARK bought in. The operating business underneath all this volatility has not changed. Our PRO team's Melvin frames SpaceX as a full stack infrastructure company across launch, satellites, broadband, defense, AI compute, power and manufacturing. There is also a decade long proof point sitting in the background. Alphabet's roughly $900 million investment from 2015 is now worth more than 100 times what it paid. Melvin's own price target sits at $225 to $250 by year end and our PRO team already holds a position. View both their portfolios for $1 (link below).

When does the first AI-created drug arrive from Eli Lilly? "Don't know, 2028, 2029, maybe 2030, or even later." But that's not really the point. AI is already compressing drug approval timelines and boosting manufacturing efficiency right now. The real question is when the market reprices the story "and I think this will come way earlier" than the actual AI drug appears. FT @WhiteCollarExit @LgDoucet.

Jensen Huang has put roughly $2 billion into $CRWV & $NBIS each. Neoclouds like Nebius and CoreWeave need to finance $16 to $20 billion in capex this year alone. Historically, that meant either torching the balance sheet with debt or diluting shareholders through equity raises. Jensen Huang is essentially telling banks "lend to these Neoclouds to buy our GPUs and we'll make sure the revenue is there to back it up." Nvidia is removing the single biggest constraint on Neocloud growth: access to cheap capital. That solves the single biggest bottleneck for names like Nebius and CoreWeave.

$UBER just closed the largest deal in food delivery history but its stock barely moved. @m0xt_ took advantage of this and bought in. The deal: $14.8 billion to acquire Delivery Hero at €41.50 per share. That's roughly 10% of Uber's current market cap. This deal will cause: - Trips and gross bookings to go up roughly 20% - Cross-platform markets nearly double from 34 to 58 countries - Over 50 million new users land inside Uber's cross-sell machine - Uber One subscriptions, advertising and financial services all get a much larger pool to work with. The deal prices at approximately 8x 2027 estimated EBITDA, comes with $1.2 billion or more in targeted annual synergies and keeps leverage below 2x. On pure deal mechanics, this is a clean acquisition at a reasonable price that immediately makes Uber's business larger and more profitable. The stock sat at roughly 22x forward earnings before the deal and it's still there after. A business that just got meaningfully larger and more profitable is trading at the same price it was before. That's the setup @m0xt_ is buying. There's one more reason this fits Martin's portfolio specifically. His portfolio skews heavily crypto and tech-cyclical assets that tend to move together when risk sentiment shifts. Uber runs a 0.25 correlation to his total portfolio. Adding it isn't just a business bet. It's a hedge against the rest of what he owns moving in lockstep during a drawdown. The base case price target sits at $93. Want to see exactly what else he owns in his portfolio? Check it out for just $1. Link below.

$AAOI's biggest risk: customer concentration. Their top 10 customers make up 96.6% of revenue. Just two customers made up 82% of total revenue last year. But that's true for almost every company in the AI bottleneck trade. There are really only five or so hyperscalers buying at this scale. The other risk is Chinese vendors who can make these modules much cheaper. "That's the biggest risk AOI has."

Our PRO team is adding to both $CRWV & $NBIS at these prices. Nvidia put $2 billion into CoreWeave in January 2026, a 9% equity stake at $87.20 a share. Two months later it put another $2 billion into Nebius through pre-funded warrants tied to a target of 5 gigawatts of AI capacity by 2030. Jensen Huang has a reason to write those checks: Hyperscalers building their own custom AI chips is a direct threat to Nvidia's growth. Independent neoclouds that buy Nvidia GPUs instead of designing their own silicon are what keeps Nvidia's install base expanding. For a while, the obvious bear case on neoclouds was cash burn. Build massive GPU clusters on debt and if customer demand does not show up fast enough, you are stuck holding depreciating hardware nobody is paying to use. But that case got weaker on July 1. Nvidia launched an AI Compute Partnership, a revenue sharing financing model where Nvidia takes a cut of a neocloud's cloud revenue in exchange for backstopping idle GPU capacity at a guaranteed floor rate. Sharon AI and Firmus are already deploying GPUs under it. Nvidia is now the one underwriting utilization risk. That makes GPU clusters easier to finance and more bankable for companies Nvidia already owns a piece of. Jensen has turned himself from a one time chip seller into a recurring revenue partner with real skin in the game on whether Nebius and CoreWeave actually succeed. Bullish $CRWV & NBIS.

"Meta is a threat to Neoclouds" narrative doesn't make sense. Morgan Stanley estimated that if Meta leased out just 250 megawatts of capacity to third parties at $40 per megawatt, it would generate roughly $3 in EPS upside by 2028, an 8% earnings boost. Scale to 1,000 megawatts and you're looking at $12 additional EPS. Here's the kicker: The cost to build a gigawatt of AI compute doubled from $50 billion to $100 billion in just three to four months. Meta locked in capacity last year before the price spike so nobody else can afford to compete anymore. That timing is the entire thesis. Any new entrant trying to build comparable capacity today faces a $100 billion price tag just to get in the door. That effectively locks out new competition and hands the biggest structural advantage to whoever already has capacity like $NBIS and $CRWV. Bullish on Nebius and the established Neoclouds as the barrier to entry compounds against anyone still trying to build from scratch.