Key Takeaways
- • Ray Dalio warns that the AI boom has entered a "more dangerous phase" centered on liquidity risk rather than technology viability, with the financial structure supporting the buildout now under strain as capital requirements accelerate.
- • Big Tech's combined AI infrastructure spending is projected to reach $650 billion in 2026, up 59% from $410 billion in 2025, fundamentally shifting the sector from asset-light software economics toward heavy infrastructure finance with unproven revenue timelines.
- • Meta raised its 2026 capex guidance to $125–$145 billion (from $115–$135 billion), exemplifying how rapidly AI investment is consuming corporate budgets and eroding the high-margin, cash-return model that justified premium valuations over the past decade.
- • Nvidia, the primary financial beneficiary of the AI buildout, reported $81.6 billion in fiscal Q1 2026 revenue (up 85% year-over-year), with data center revenue climbing 92% to $75.2 billion, underscoring the scale of infrastructure spending but raising sustainability questions.
- • The structural risk is not AI's utility but whether corporations can fund $650 billion annual capex, sustain buyback programs, and attract external capital simultaneously—a dynamic that resembles capital-intensive infrastructure finance rather than traditional high-margin technology.
Summary
Ray Dalio's recent warnings about artificial intelligence center on a critical but overlooked risk: not whether AI technology works, but whether the financial architecture supporting the boom can withstand the accelerating capital demands of the buildout. According to Bridgewater Associates analysis reported by Reuters, the AI cycle has shifted into what co-chief investment officer Greg Jensen characterized as a "more dangerous phase," one driven by expanding infrastructure requirements and growing reliance on external funding sources.
The spending trajectory illustrates the magnitude of this structural shift. Bridgewater estimates that Alphabet, Amazon, Meta Platforms, and Microsoft will collectively invest approximately $650 billion in AI infrastructure during 2026, representing a 59% increase from the $410 billion deployed in 2025. This acceleration is not merely a budget line item; it fundamentally alters the financial profile of the technology sector. For the past decade, Big Tech's elevated valuations rested on three pillars: high profit margins, asset-light software economics, and substantial capital returns to shareholders through buybacks and dividends. The AI infrastructure buildout is eroding this model by pushing the sector toward capital-intensive, asset-heavy operations where major upfront expenditures precede proven revenue streams at scale.
Meta exemplifies the velocity and scale of this budget reallocation. The company reported Q1 2026 capital expenditures of $19.84 billion and raised its full-year capex guidance to $125–$145 billion, up from a prior range of $115–$135 billion, citing heightened infrastructure costs and data center capacity requirements. This represents a fundamental repricing of corporate capital allocation priorities, one that will likely constrain cash available for shareholder returns and increase dependence on debt markets and equity issuance.
Nvidia, as the primary beneficiary of the AI infrastructure wave, reported fiscal first-quarter 2026 revenue of $81.6 billion, up 85% year-over-year, with data center revenue reaching $75.2 billion, a 92% increase. While these results validate near-term demand, they also highlight the concentration of financial gains in a narrow set of suppliers. The broader risk Dalio identifies is whether corporations can simultaneously fund $650 billion in annual capex, maintain competitive return programs, and access capital markets at favorable terms—a dynamic that increasingly resembles capital-intensive infrastructure finance rather than the scalable, capital-efficient technology business model that historically commanded premium valuations.
The spending trajectory illustrates the magnitude of this structural shift. Bridgewater estimates that Alphabet, Amazon, Meta Platforms, and Microsoft will collectively invest approximately $650 billion in AI infrastructure during 2026, representing a 59% increase from the $410 billion deployed in 2025. This acceleration is not merely a budget line item; it fundamentally alters the financial profile of the technology sector. For the past decade, Big Tech's elevated valuations rested on three pillars: high profit margins, asset-light software economics, and substantial capital returns to shareholders through buybacks and dividends. The AI infrastructure buildout is eroding this model by pushing the sector toward capital-intensive, asset-heavy operations where major upfront expenditures precede proven revenue streams at scale.
Meta exemplifies the velocity and scale of this budget reallocation. The company reported Q1 2026 capital expenditures of $19.84 billion and raised its full-year capex guidance to $125–$145 billion, up from a prior range of $115–$135 billion, citing heightened infrastructure costs and data center capacity requirements. This represents a fundamental repricing of corporate capital allocation priorities, one that will likely constrain cash available for shareholder returns and increase dependence on debt markets and equity issuance.
Nvidia, as the primary beneficiary of the AI infrastructure wave, reported fiscal first-quarter 2026 revenue of $81.6 billion, up 85% year-over-year, with data center revenue reaching $75.2 billion, a 92% increase. While these results validate near-term demand, they also highlight the concentration of financial gains in a narrow set of suppliers. The broader risk Dalio identifies is whether corporations can simultaneously fund $650 billion in annual capex, maintain competitive return programs, and access capital markets at favorable terms—a dynamic that increasingly resembles capital-intensive infrastructure finance rather than the scalable, capital-efficient technology business model that historically commanded premium valuations.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.