The Quiet AI Trades Powering the Next Decade VIEW IN BROWSER BY ANDY SWAN, FOUNDER, LIKEFOLIO Our most profitable AI trade of 2025 wasn’t an AI company at all. It was Oklo (OKLO) – a nuclear power developer we recommended buying in April… and then selling half the initial stake for a +461% gain just five months later. AI didn’t show up in Oklo’s marketing. It showed up in the math: exploding compute demand, strained grids, and a race for always-on power. Web traffic exploded. Enterprise demand surged. Oklo’s nuclear power became impossible to replace. That trade wasn’t about hype or headlines. It was about understanding where AI demand has to go next. That’s what MegaTrends is built for. Our proprietary Social Heat Score monitors millions of real-time consumer and web data points – search trends, app traffic, product interest, and online engagement – to uncover which companies are gaining momentum before Wall Street catches on. We track real-world behavior to identify stocks like OKLO with soaring Social Heat Scores – OKLO’s was a bullish 79 out of 100 at the time we recommended it – positioned to benefit from massive shifts while they’re still flying under the radar. Nvidia (NVDA), Meta Platforms (META), Microsoft (MSFT): These powerhouses helped fuel one of the biggest technology runs of the past decade. But their valuations already reflect the hype. The easy money is gone. The real opportunity has shifted beneath the surface. So today, I want to show you where our data sees the next wave of AI winners. And why it’s in places most investors aren’t even thinking of looking. Hidden Opportunity No. 1: The AI “Compute Crunch” Analysts once assumed AI would stall for lack of quality data or diminishing returns in the algorithms. Instead, the obstacle is compute capacity. Think of it like the raw horsepower behind AI. It’s what all those semiconductors are for. They’re converting energy into AI responses and results. The more compute, the higher the potential. Sam Altman framed the stakes directly in a September blog post: With 10 gigawatts of compute, AI could work on curing cancer. Or it could deliver personalized tutoring to every student on earth. Ten gigawatts of compute is roughly equivalent to the output of 10 nuclear reactors. To put this in perspective, AI compute capacity has increased 100x in just three years. In the past year, it tripled again. Future AI demand is projected to require data centers consuming as much electricity as entire cities. So scaling AI isn’t a software problem – it’s a physical one. And it’s not just about chips. Chips need power… Power generates heat… Heat needs cooling… Cooling requires infrastructure. Every link in that chain is under strain. And every link is an opportunity. The Obvious Play: Advanced Micro Devices (AMD) Advanced Micro Devices’ newer chips make AI workloads cheaper and more efficient. As compute and power needs grow, innovations like this will be essential. The Under-the-Radar Play: Vertiv (VRT) But before a single AI chip can be switched on, data centers need power systems, cooling equipment, and racks. That’s Vertiv’s domain. Rather than build AI models, this company makes sure the machines running those models don’t melt down – allowing more compute capacity to come online safely. That makes it essential to the AI infrastructure buildout. VRT is up 48% over the past year. And its Social Heat Score sits at a bullish 66.4, suggesting web traffic and enterprise interest continue climbing. 
Vertiv is a classic “hidden AI” play. Boring on the surface, critical underneath. Hidden Opportunity No. 2: The Invisible Workforce The next phase of AI isn’t about chatbots answering questions. It’s about autonomous agents executing work. These systems plan, decide, and act – across multiple steps – without human intervention. Think of it as software that doesn’t just think. It does. The implications of this “Invisible Workforce” are massive. - Private AI giant Anthropic estimates agentic AI could double U.S. productivity growth over the next decade
- MarketsandMarkets projects the agentic AI market will grow from $7 billion in 2025 to $93 billion by 2032
- Precedence Research pushes that number to $199 billion by 2034
- Agentic commerce alone could reach $1.7 trillion by 2030
Every leap in technology has rewarded whoever removed friction. Agentic AI will remove the entire thought. The Obvious Play: Alphabet (GOOGL) As shopping moves toward AI-driven search and automated recommendations, Alphabet controls the entry points agents rely on – structured data, reviews, and product discovery. Scale matters here. And Google has it. We like this stock as much as the next guy. But at a $4 trillion market cap, the hype is largely priced in. The Under-the-Radar Play: Innodata (INOD) Innodata currently sits at a fraction of that market cap ($1.38 billion), but this relatively tiny company has quietly become a key partner for five of the “Mag 7” tech giants. AI agents need training. That training needs to happen on massive, high-quality data sets to get the best possible results. Data like that doesn’t exist in the public domain – but it’s exactly what Innodata provides. Training on bad data is costly, especially in high-stake sectors like finance, healthcare, or government where the tasks are complex – and the consequences are steep. Companies need AI agents to do real work safely and reliably. Innodata supplies that “action-ready” data. The market hasn’t bought the story yet. INOD is up just 10% year over year, but its Social Heat Score is teetering on bullish territory at 57.7 out of 100. (Scores above 60 meet our threshold.) Any sustained uptick in web traffic, search interest, or otherwise could soon push it to a buy – making INOD a must watch. 
Expectations are low. And that’s often where upside begins. Hidden Opportunity No. 3: The Quiet Robotics Boom AI advancements and lower costs have pushed robots from science fiction to scalable reality. Consider that the cost to build industrial robots has been slashed in half over the past decade. Meanwhile, the global factory robot workforce has more than doubled over the same period. With a setup like that, robots could soon be as common as smartphones, transforming industries and minting new stock market winners. Big tech sees it coming… but retail investors have yet to catch on. Nvidia (NVDA) has been quietly writing checks to robotics startups like Figure AI (participating in a $1 billion+ Series C round in September). That’s a strong vote of confidence. And we’re following its lead. The Obvious Play: Tesla (TSLA) Tesla is leading the charge in humanoid robots. It’s now essentially a robotics company that happens to sell vehicles. Just last week, the company announced plans to decommission Model S/X lines at its Fremont factory to make room for Optimus Gen 3 production. MegaTrends members are already up over 100% on TSLA – and we believe robotics is the next chapter. But it’s not exactly a secret. We’re looking for the next winner, where the upside could be even bigger. The Under-the-Radar Play: Lattice Semiconductor (LSCC) Every autonomous system – whether it’s a delivery drone or a smart surgical tool – needs real-time decision making. That need means there’s no time to query a server thousands of miles away and wait for a response. The AI will need to operate directly on the device, without lag or risk of disconnection. That’s where Lattice (LSCC) shines. Its ultra-efficient FPGAs (field-programmable gate arrays) are designed to sit right next to sensors, allowing AI models to run directly on devices so the intelligence happens locally. The stock is up 45% year over year, and its Social Heat Score is a bullish 65.6 as of this writing. 
This is how “AI everywhere” actually happens. And frankly, LSCC covers more than just robotics: - Compute crunch: Lattice’s chips are power efficient, operating where GPUs can’t – and reducing the load on data centers.
- Agentic AI: Edge-based agents will need real-time processing. Lattice enables it, especially where low-power, real-time computation is required.
- Robotics: Autonomous robots require instant response time to operate in the real world. Lattice’s FPGAs enable these systems to run inference locally.
Better yet, Lattice’s underlying web traffic is up 56% year over year – a signal that demand is building quietly beneath the surface: 
Rather than chase what already worked, we’re positioning for what’s next. And that’s where I believe the smartest AI opportunities are hiding right now – just beneath the surface. The Social Heat Score spots the next AI winners early by identifying where real-world demand is headed next. Our system tracks company and brand-level web traffic… search interest… app usage… social media chatter… and even scans the web for press releases and media coverage to detect spikes in mentions and to measure sentiment. Those metrics are baked into the Social Heat Score algorithm so investors can spot emerging opportunities as the data shifts – from right inside TradeSmith Finance. This tool is available with a MegaTrends subscription. It covers more than 250 stocks – and we’re expanding that coverage constantly. To learn how you can unlock it for yourself, go here now. Until next time, 
Andy Swan
Founder, LikeFolio | 
