Duarte

A reação dos sócios do Sport Lisboa e Benfica à subida de Luís Filipe Vieira para discursar na Assembleia-Geral deste sábado.

MEET THE NEW MAG 7 OF AI’S NEXT ERA We’re in one of those rare windows where the market gives you a second chance. Not because the fundamentals have changed, but because the noise has drowned them out. What we’re seeing now -- with $SPY down 15% from its highs, $QQQ working its way out of a bear market, and even the most resilient names selling off -- isn’t a reflection of these companies failing. It’s a market pricing headlines, not horizons. And that’s exactly what makes this moment so compelling. Zoom out, and it’s clear: the next industrial base is already being built. And it’s being led by a new class of incumbents -- what I consider the “Mag 7” of AI’s next era. These aren’t hype-cycle beneficiaries. These are companies laying the foundation for how the digital economy will operate over the next decade. Each is building an ecosystem, not a product. Each controls its lane so decisively that competition is either irrelevant or already embedded. And right now? You’re getting them at prices that completely ignore the scale of what they’re building. $TSLA is the prime example. The narrative still tries to pull it back into the EV bucket, but that’s yesterday’s framework. What Tesla is really building is a fully integrated real-world AI loop -- where software, hardware, and energy infrastructure feed each other in continuous feedback. FSD is not a driver assist feature. It’s the front end of an autonomy stack built on vertically owned data, custom silicon (Dojo), and a robotics platform that’s learning from billions of real-world edge cases. When Optimus comes online at scale -- and it will -- we’re not going to be debating margins on car sales. We’ll be talking about Tesla as the physical operating system for intelligent infrastructure. From vehicle networks to power grids to autonomous logistics, it’s not about selling units. It’s about embedding compute into motion. $PLTR, in a parallel but equally important lane, is doing for institutional decision-making what Tesla is doing for real-world autonomy. AIP isn’t a toolkit -- it’s the execution engine. When enterprises plug into Palantir, they’re not just gaining visibility -- they’re delegating action. AIP can ingest, interpret, and execute across workflows, at a level of complexity that makes traditional business software look like glorified spreadsheets. Its adoption curve in defense, healthcare, manufacturing, and energy isn’t slowing -- it’s compounding. And once it’s in, it doesn’t come out. Because in high-stakes, regulated environments, trust isn’t a feature -- it’s the whole product. $SNOW, for its part, is solving what might be the biggest bottleneck in the AI economy: data liquidity. Everyone talks about model performance. But models are nothing without clean, governed, high-velocity data moving through them. Snowflake’s Data Cloud isn’t a warehouse. It’s a cross-cloud, real-time platform that lets enterprises federate data and deploy intelligence across teams, borders, and partner networks -- without breaking architecture. As AI moves from experimentation into production, Snowflake becomes the connective tissue. Not the spotlight. The bloodstream. $CRWD is turning that bloodstream into a protected environment. In a world of autonomous agents, synthetic users, and adversarial AI, the attack surface isn’t just broader -- it’s live. Falcon doesn’t just react. It learns. And it’s already scaled. Every endpoint it protects becomes another signal in a global detection network that trains continuously. Its move into identity and observability makes the platform even more defensible -- converging what were previously fragmented security functions into a single, AI-native architecture. It’s not just about stopping breaches. It’s about building the security layer of the intelligent enterprise. $NET is enabling something even more profound: distributing compute to the edge. As the AI economy matures, speed becomes everything -- and centralization becomes a bottleneck. Cloudflare is already running models directly in its global edge network, which spans hundreds of cities and processes over 45 million HTTP requests per second. This isn’t about making the internet faster. It’s about making intelligence instant. When you need a fraud detection model to run in 2ms or a personalization engine to respond before a page loads, Cloudflare is the only infrastructure already there. And as AI inference moves toward decentralized execution, Cloudflare is positioned to become the low-latency layer for every modern app, device, and system. $AXON, while less obvious, might be the most vertically entrenched of all. It owns public safety infrastructure -- full stop. There is no second choice. Body cams, evidence, real-time operations, AI-driven video analytics -- it all flows through Axon. And because it built the full stack in-house, there are no handoffs. Just lock-in. What makes Axon unique is that its moat isn’t tech. It’s institutional capture. Agencies don’t trial Axon. They adopt it. And once they do, they’re locked into 5–10 year cycles with no viable alternatives. That control is now expanding into new verticals -- corporate security, critical infrastructure, and even civilian use cases. Axon is becoming the trusted interface for ethical, accountable AI deployment in real-world environments where latency, compliance, and reliability are paramount. And finally, Databricks is tying all of this together. Because the companies building AI-native systems still need to build, and they need to do it fast. Databricks eliminates the silos -- between engineering, ML, and operations -- and allows continuous loops of data ingestion, training, and deployment. It’s the only platform that lets companies treat AI development like software development -- fast, iterative, and aligned to production. And in the future AI economy, speed is the edge. If you can retrain and redeploy faster than the competition, you win. Databricks isn’t just selling that speed. It’s embedding it. This is why I’m so focused on these seven. Because the future is not some abstract vision anymore. It’s being built, piece by piece, by companies with defensible moats, proven deployment, and massive end-state potential. And yet, they’re trading like they’re just another basket of tech. You don’t need to pick winners inside of the hype. You need to own the infrastructure being laid beneath it. That’s what this group is. The foundation layer. The control layer. The deployment layer. And right now, it’s all on sale. Not because they’ve lost their edge -- but because the market can’t see far enough ahead to price it. That’s the opportunity. This is when you build the portfolio that you’ll look back on in five years and wish you added more to. Not because it was obvious. But because you knew what mattered before everyone else caught up.