Big Tech’s $650 Billion AI Bet Enters a ‘More Dangerous Phase’

When Greg Jensen, co-chief investment officer at Bridgewater Associates, looked at the capital expenditure plans coming out of Silicon Valley this month, he saw something that made him uneasy. The numbers weren’t just big—they were unprecedented. And the risks that came with them, he warned, were entering a “more dangerous phase.”

The figures are staggering. Alphabet, Amazon, Meta, and Microsoft are collectively projected to pour approximately $650 billion into AI-related infrastructure in 2026. That’s a 58% jump from the $410 billion expected this year. To put it in perspective: this single-year increase alone rivals the entire GDP of countries like Poland or Sweden.

“The AI sector is entering a more dangerous phase. We’re seeing unprecedented spending on physical infrastructure, and the reliance on outside capital is increasing.” — Greg Jensen, Co-CIO, Bridgewater Associates

The Hyperscaler Arms Race

The individual numbers tell their own stories. Amazon alone expects to spend roughly $200 billion on capital expenditures in 2026—a figure that would have been unthinkable just a few years ago. Microsoft, Google, and Meta are all scaling at similar velocities, each racing to secure the computational horsepower they believe will determine the winners of the AI era.

Demand is outstripping supply at every turn. Jensen noted that the hunger for computing power currently exceeds what’s available, creating a scarcity dynamic that’s driving prices up and timelines out. Data center construction timelines have stretched from months to years. Nvidia’s latest chips, despite record production, remain backordered. The physical constraints of building AI infrastructure are becoming as important as the software innovations they support.

But this isn’t just about hardware. The investments span data centers, specialized chips, power infrastructure, and the talent required to build and operate it all. It’s a capital-intensive transformation that makes the cloud computing build-out of the 2010s look modest by comparison.

“We’re at a point where fear of falling behind can fuel a bubble. Companies are treating AI spending as a must-have, which can stretch the boom further than fundamentals justify.” — Bridgewater Research

Wall Street’s Growing Skepticism

Markets have reacted with caution rather than enthusiasm. When Amazon, Alphabet, and Microsoft detailed their spending plans during recent earnings calls, their shares fell. Investors are increasingly asking the same question: where’s the return?

Analysts have started running the math, and the results are sobering. A 10% return on $1 trillion in AI capital expenditure would require generating an additional $100 billion in free cash flow annually. For context, that’s more than the annual profits of most Fortune 500 companies. Some analysts view this target as unlikely, if not impossible, given current AI monetization rates.

The financing model is shifting. Tech giants have historically been cash machines, generating more money than they could productively reinvest. Now, some are becoming significant debt issuers to fund their AI ambitions. The capital requirements are so large that they’re attracting infrastructure funds—investors who typically finance bridges, ports, and power plants, not software companies.

This shift has broader economic implications. According to Apollo Global Management, hyperscaler capital expenditure in 2026 will approach 2% of U.S. GDP. Without AI-related investment, overall U.S. corporate equipment investment would be negative. The AI build-out is essentially propping up American business investment.

Pressure Mounts on AI Startups

The ripple effects extend beyond the tech giants themselves. Jensen specifically warned that AI startups like Anthropic and OpenAI may need significant breakthroughs to justify their current valuations and future funding rounds. The bar for success is rising as fast as the capital being deployed.

Other sectors face disruption too. Software companies and data providers are experiencing volatility as markets try to price in how AI might reshape their businesses. Recent stock selloffs in these sectors suggest investors are uncertain whether incumbents will be disrupted or enhanced by the AI wave.

The competitive dynamics are also shifting. As hyperscalers build their own AI capabilities, they become less reliant on third-party providers. This creates a challenging environment for AI startups that have built their businesses on providing infrastructure or models to larger companies.

“This reminds me of the Barnes and Noble moment in retail—companies treating spending as a defensive necessity rather than an offensive opportunity. That mindset can extend booms well beyond their natural endpoints.” — Bridgewater Analysis

What Comes Next

The critical question isn’t whether this investment will happen—it’s already committed. The question is what it will yield. Will 2026 be remembered as the year AI infrastructure reached sufficient scale to unlock transformative applications? Or will it mark the peak of a capital cycle that outpaced real demand?

Several scenarios could play out. If AI applications achieve widespread enterprise adoption and demonstrate clear ROI, the spending will look prescient. If adoption lags or if AI capabilities plateau, the infrastructure overhang could weigh on the sector for years.

For now, the bet is placed. Alphabet, Amazon, Meta, and Microsoft are collectively wagering that AI will reshape their businesses and the broader economy. The scale of that wager—$650 billion—makes it one of the largest capital allocation decisions in corporate history. Whether it pays off will shape the technology landscape for the next decade.

As Jensen’s warning suggests, the danger isn’t just that the bet might fail. It’s that the fear of missing out could drive spending even higher, extending a cycle that may already be stretched thin. In the race to build the future, nobody wants to be left behind. But history suggests that not everyone can win.

This article was reported by the ArtificialDaily editorial team. For more information, visit Reuters and Tech in Asia.