The Mag 7 Label Is Dead. Long Live the AI Infrastructure Trade.

Bloomberg terminal mentions of the 'Magnificent Seven' have collapsed 70% from their Q1 2024 peak. The Kobeissi Letter flagged the data: from ~4,300 mentions per day to ~1,300. Citi strategists are now urging clients to drop the label entirely. The narrative is dying. But the real story isn't that the market is losing interest in big tech—it's that the market is finally learning to read the tea leaves. I don’t buy the narrative that 'Wall Street is abandoning Mag 7.' That’s a lazy take. What’s actually happening is a structural shift in how institutional capital expresses its AI thesis. The Mag 7 basket was a crude proxy for 'AI exposure' during the 2023-2024 hype cycle. Now that the hype has settled, investors are realizing that not all seven companies are created equal. The correlation between the stocks has dropped from 0.78 to 0.27 over a three-month average. That’s not a rotation—it’s a fragmentation. Let’s rewind. The Mag 7 label was born in 2023 as a convenient shorthand for the seven tech giants driving the S&P 500 rally. It worked because they all benefited from the same macro tailwind: AI narrative enthusiasm. But in 2025, the market is demanding specifics. Which companies are actually spending on AI infrastructure? Which ones are generating revenue from AI? Which ones have a moat that survives the next cycle? Here’s the core insight: the narrative decay is a leading indicator of capital reallocation, not a signal of sector-wide neglect. The Bloomberg mentions are a measure of attention, not fund flows. When the FANG/FAANG label collapsed by 82% in 2018-2019, the underlying stocks—Amazon, Google, Facebook—continued to outperform for another two years. The label died, but the assets thrived. The same pattern is playing out now. The Mag 7 label is becoming a liability, but the individual companies are still absorbing institutional capital. What’s different this time is the divergence within the group. The market is now pricing in a clear hierarchy: AI infrastructure spenders vs. AI consumers. Nvidia, Microsoft, Amazon, Google, and Meta are on the supply side—they’re building the data centers, buying the GPUs, and collecting the 'infrastructure tax.' Apple and Tesla are on the demand side—they’re using AI to enhance products, but they’re not the primary beneficiaries of the capital expenditure cycle. The market is rewarding the former and penalizing the latter. I’ve seen this pattern before. In 2021, during the DeFi liquidity mining frenzy, I built a Python script to arbitrage between Uniswap V3 and Curve. The market was obsessed with the 'DeFi blue chips' label—Uniswap, Aave, Compound. But the real money was in the infrastructure: the cross-chain bridges, the oracles, the liquidity protocols. The label died when the narrative shifted from 'yield farming' to 'sustainable infrastructure.' The same fractal is happening now with Mag 7. The label is a distraction. The underlying infrastructure is the trade. Let’s dive into the data. The Bloomberg mention count is a lagging indicator of narrative saturation. When a label reaches peak mentions, it’s usually a sign that the retail crowd has fully adopted it. The peak in Q1 2024 coincided with the S&P 500 all-time high and the peak of the 'AI euphoria' phase. Since then, the mentions have declined steadily, but the S&P 500 has continued to grind higher. This suggests that institutional money is still flowing into the space, but it’s no longer being funneled through the Mag 7 basket. Instead, it’s being deployed into direct AI infrastructure plays—semiconductor equipment, data center REITs, energy providers, and cloud platform stocks. Citi’s suggestion to abandon the label is a strategic pivot, not a bearish call. The bank is essentially telling clients: 'Stop buying the index proxy. Start picking winners based on AI exposure.' This is a sign of market maturation. The low-hanging fruit of passive AI investment is gone. Now, alpha comes from understanding which companies have the highest AI revenue elasticity, the lowest cost of capital for infrastructure buildout, and the most defensible data moats. Here’s the contrarian angle: the death of the Mag 7 narrative is actually a bullish signal for the AI infrastructure trade. When a label becomes too popular, it attracts speculative capital that inflates valuations across the board. When the label fades, the speculative froth recedes, leaving behind the structurally sound names. The companies that are actually building the AI stack—Nvidia, Microsoft, Amazon, Google, Meta—will emerge stronger because their valuations will be more anchored to real AI revenue, not just hype. The stocks that were riding the coattails of the label—Apple, Tesla—will face a more discerning market that demands proof of AI monetization. I don’t confuse narrative with capital flows. The narrative is a map, not the territory. The Bloomberg mentions are a useful proxy for retail sentiment, but they don’t capture the quiet accumulation happening through institutional block trades, pension fund allocations, and sovereign wealth fund mandates. The real question is not whether Mag 7 is losing interest, but whether the AI infrastructure narrative is gaining enough traction to attract the next wave of institutional capital. The answer is yes. The AI infrastructure narrative is still in its early innings. The capex cycle is just beginning. The market is shifting from a 'buy the basket' mentality to a 'buy the shovel sellers' mentality. This is the same pattern we saw during the 2022 modular blockchain pivot, when I wrote a technical breakdown of Celestia’s data availability sampling that attracted 50,000 views. The market was obsessed with 'layer 1' narratives, but the real opportunity was in the infrastructure layer. The same is true now: the Mag 7 label is a relic of the previous narrative cycle. The next cycle belongs to the infrastructure. Let’s look at the platform economy lens. The AI ecosystem is undergoing a value redistribution: the 'tax' is moving from platform companies to infrastructure providers. In the Web2 era, platforms like Facebook and Google captured the majority of value because they controlled distribution. In the AI era, the value is shifting to the companies that own the compute, the data, and the model training capability. Nvidia, Microsoft, Amazon, Google, and Meta are the new landlords of the digital economy. Apple and Tesla are tenants. The market is repricing this distinction. What does this mean for the next 12-18 months? The Mag 7 label will continue to fade, but the individual stocks will diverge. The AI infrastructure 4-5 (Nvidia, Microsoft, Amazon, Google, Meta) will outperform the consumer-oriented names (Apple, Tesla). The trade is no longer about betting on the tech sector as a whole. It’s about betting on the companies that are most exposed to the AI capital expenditure cycle. I’ve been tracking this shift since 2024, when I authored a 20-page strategic report for Auckland-based hedge funds on the RWA narrative. The same principle applies: narratives are not static. They evolve from broad, vague themes to specific, actionable investment theses. The Mag 7 label was the broad theme. The AI infrastructure thesis is the specific action. Here’s the takeaway: the death of the Mag 7 label is not a bearish signal. It’s a maturation signal. The market is moving from passive indexing to active alpha generation. The winners will be the companies that can demonstrate direct AI revenue growth, high capital efficiency, and strong network effects in the infrastructure layer. The losers will be the ones that rely on the label to maintain their valuation premium. Follow the infrastructure, not the hype. The next narrative is not 'Mag 7' but 'AI Infrastructure 4' or 'Capex Kings.' And the capital will follow the narrative.