Bridgewater founder Ray Dalio has sharpened his warnings about artificial intelligence, telling the All-In Podcast that AI is “eating everything” and warning it might “eat itself.” The comments land at a moment when corporate spending on AI infrastructure continues to accelerate, yet the path from investment to profit remains unclear for most firms. Dalio’s argument ties together genuine technological excitement with the mechanics of a bubble, and it deserves closer scrutiny than a single soundbite allows.
From “Change Everything” to “Eat Itself”
Dalio’s thinking on AI has evolved in public over several years, and the trajectory matters. During a dialogue at Tsinghua, he described AI as something that will “change everything,” singling out newer systems’ ability to understand and communicate as particularly exciting. That framing was optimistic, focused on capability rather than risk, and it fit neatly with a broader narrative that AI would reshape productivity, labor markets, and even education. While he acknowledged disruption, his emphasis at that stage was on how societies and firms could adapt to extract the upside rather than on how the technology might destabilize markets in the near term.
As the technology raced ahead, Dalio’s public comments began to carry a sharper edge. In an October 2023 interview, he told Bloomberg that AI can have an “enormous productivity impact”, but he already drew a dividing line: the gains would flow to firms that really know how to use the technology, not to every company writing checks for it. That distinction foreshadowed his later warnings. By the time he joined the All-In Podcast, a transcribed conversation shows him using the phrase “AI is eating everything, and it might eat itself,” tying technological promise directly to bubble dynamics. The progression from “change everything” to “eat itself” is not a contradiction. It is the logical next step for someone who sees a powerful tool attracting more capital than the market can profitably absorb.
Bubble Dynamics and the Spending Trap
On January 5, 2026, Dalio stated on the social media platform X that the AI boom is in an “early bubble phase”, a characterization reported by Reuters and echoed in subsequent coverage. That label carries weight because Dalio built his career studying debt cycles and asset bubbles at Bridgewater Associates. Calling something an early bubble does not mean it will pop tomorrow; it means the gap between price expectations and underlying earnings is widening, and the feedback loop that drives it has not yet reversed. In his framework, early bubbles are defined by optimism, rapid capital flows, and a narrative that seems too compelling to question, even as hard numbers lag behind the story.
The “eat itself” metaphor captures a specific mechanism. Companies race to deploy AI, which drives up demand for chips, cloud capacity, and specialized talent. Those costs get passed through the system, compressing margins for the very firms trying to profit from AI adoption. If enough participants spend heavily without generating sustainable returns, the capital dries up and the buildout stalls. A clip summary of his All-In appearance highlights bubble dynamics, monetization challenges, competitive pressure, and shrinking margins as the core of his argument. The technology does not need to fail for the bubble to deflate; it only needs to cost more than it earns for long enough, especially for late entrants who are paying peak prices for infrastructure and talent while competing in increasingly crowded markets.
Why Most Coverage Gets the Warning Wrong
Much of the reaction to Dalio’s comments has treated “eating everything” and “eat itself” as two separate predictions, one bullish and one bearish. That framing misses the point. In Dalio’s telling, the two ideas are causally linked: AI is eating everything precisely because capital is flooding in at unsustainable rates. The same force that drives rapid adoption also inflates costs, crowds out weaker players, and sets up the conditions for a correction. The social sharing link for the All-In recap points back to the same bundle of concerns about margins and competitive intensity, underscoring that Dalio is describing one continuous process rather than toggling between optimism and doom.
This is where his earlier distinction about which firms benefit becomes crucial. When he told Bloomberg that only firms that really know how to use AI would see the enormous productivity impact, he was implicitly sorting the market into durable winners and vulnerable speculators. The companies with clear revenue tied to AI capabilities—those embedding it into core products, workflows, or customer experiences—stand a better chance of surviving a spending correction. By contrast, firms treating AI as a branding exercise, or as a defensive move to keep pace with rivals, risk becoming casualties of the very boom they feel compelled to join. That gap between skilled deployers and speculative spenders is where bubble pressure builds, and it is where Dalio seems to expect the eventual damage to concentrate.
What the “Early Bubble” Label Means for Investors
Dalio’s “early bubble phase” language deserves careful parsing because it can be misread as a simple call to run for the exits. In his historical work on cycles, early bubbles often run for years before they burst, and they frequently leave behind valuable infrastructure. The dot-com era destroyed trillions in paper wealth but also built the backbone of the modern internet, including data centers, fiber networks, and e-commerce rails. Today’s AI buildout (massive data centers, specialized chips, and software platforms) could follow a similar arc: painful for overstretched investors yet ultimately productive for the broader economy. Even a company like Shopify’s commerce platform, which is not an AI hardware vendor, illustrates how digital infrastructure laid down in one boom can later become the foundation for new waves of innovation.
For investors, the practical takeaway is not to avoid AI entirely but to distinguish between companies with defensible AI economics and those burning cash on vague strategies. Dalio’s comments, read alongside his earlier remarks at Tsinghua’s finance programs, sketch a consistent view: technology revolutions reward disciplined operators who tie investment to measurable productivity, while punishing those who chase narratives. That suggests a focus on firms that can show how AI lowers unit costs, expands margins, or opens new revenue streams, rather than those merely touting pilots and prototypes. In an early bubble, selectivity becomes a form of risk management. Investors who treat “AI is eating everything” as a license to buy anything with an AI label risk being the ones ultimately eaten by the cycle Dalio is describing.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
