Usman Salis
AI still dominates the news cycle, but Coatue Management's recent analysis paints a more complicated picture. More interestingly, the report identifies a changing of the guard — with infrastructure players like Palantir, Broadcom, and TSMC emerging as the next generation of AI leaders.
What Coatue Found
The investment firm's report, shared by tech analyst Molly O'Shea, shows that while Big Tech continues pouring money into AI, actual adoption is slowing down and valuations are starting to look uncomfortably similar to the 1999 dot-com bubble.Coatue analyzed what they call the "Top 7 in Tech" — NVIDIA, Microsoft, Apple, Google, Amazon, Meta, and Broadcom — and compared their current state to historical market benchmarks. The findings tell a story of a market in transition. Today's tech giants are trading at valuations that mirror the late 1990s peak, raising red flags about potential overheating. The once-dominant "Magnificent 7" are showing cracks in their armor, with performance starting to diverge after nearly two years of synchronized gains. Meanwhile, market concentration has reached extreme levels, with the top 10 AI firms representing roughly 77% of GDP-equivalent market value.
Perhaps most significant is the adoption slowdown. Despite record spending on chips and data centers, enterprise uptake has plateaued, likely due to infrastructure limitations, integration challenges, and mounting costs. Coatue also flags vendor financing as a potential concern, suggesting some companies may be inflating their revenue numbers by offering generous payment terms that mask weaker underlying demand.
Why AI Growth Is Slowing
Several factors explain the deceleration. Infrastructure constraints remain a bottleneck — even major players struggle to secure enough capacity for their AI ambitions. Enterprise integration is harder than expected, requiring extensive work on data governance, staff training, and regulatory compliance. Economic headwinds haven't helped either, with high interest rates forcing companies to prioritize profitability over experimental spending. And after the early wins with chatbots and automation, the next generation of AI applications requires deeper, more complex innovation that takes time to develop.
Coatue's report points to a strategic shift in where AI value is being created. The first wave belonged to consumer-facing giants, but the next chapter looks like it'll be written by infrastructure providers. Palantir is building AI platforms for defense and government analytics. TSMC and Broadcom control the semiconductor supply chains that make model training possible. GE Vernova shows how traditional industrial companies can transform legacy sectors with machine learning. This rotation suggests we're moving toward a more diversified AI economy where the real money flows to enablers rather than just platform owners.
Coatue compared current conditions to over 30 historical bubbles spanning four centuries and concluded that while things are definitely overheated, we're not quite in classic bubble territory yet. Unlike past manias, today's valuations are backed by genuine technological progress. The real risk isn't hype itself but execution gaps — companies that promise transformative AI results without delivering actual productivity gains or revenue growth. And investor patience for unprofitable AI stories is running thin.
For investors, the message is clear: reassess your exposure to mega-cap tech, look toward infrastructure players and mid-caps in the AI supply chain, and scrutinize any revenue that depends heavily on vendor financing. Enterprises need to manage expectations, strengthen their data infrastructure before scaling AI solutions, and focus on ROI-driven adoption rather than chasing trends. Policymakers should monitor market concentration to prevent systemic risk and work on regulations that foster responsible but scalable AI deployment.
Usman Salis
