newsio aggregates and links to original sources. We do not own the original images or content. If you believe content infringes on intellectual property rights, contact us — it will be removed at first notice.
business/news//Forbes
The five largest hyperscalers are on track to spend somewhere between $700 billion and $900 billion on capital expenditures in 2026.
Hyperscalers' capex is projected to hit $700-$900 billion in 2026, driven overwhelmingly by AI infrastructure.
KEY POINTS
A $600 billion annual revenue gap exists between AI infrastructure spending and ecosystem revenue, and it's widening.
Big tech is now issuing new debt to fund capex, with $108 billion raised in 2025 alone.
NVIDIA captures 90% of AI accelerator spend, earning around $180 billion in annual GPU purchases.
MIT found that 95% of enterprise GenAI pilots had zero measurable P&L impact as of July 2025.
The numbers coming out of Silicon Valley this earnings season are, by any historical measure, staggering. The five largest hyperscalers — Amazon, Microsoft, Alphabet, Meta, and Oracle — are on track to spend somewhere between $700 billion and $900 billion on capital expenditures in 2026, a 36% increase over 2025 according to CreditSights estimates. Amazon alone has guided for $200 billion in capex this year, more than doubling its 2025 outlay. Meta raised its full-year guidance to as much as $145 billion, citing higher component costs and additional data center buildout. Microsoft is tracking above $120 billion for its fiscal year. And Alphabet has roughly doubled its guidance to $175-185 billion, with Google Cloud backlog surging to over $460 billion.
These are not incremental expansions. They are commitments at a scale that, as one analyst noted, rivals Sweden's entire GDP. Roughly 75% of that spend — or about $450 billion — is directly tied to AI infrastructure: GPU clusters, custom accelerators, data centers, and the power and cooling systems to keep them running. NVIDIA's data center revenue hit $62.3 billion in Q4 alone, up 75% year-over-year, and the company's networking segment grew 263%. Jensen Huang called it "the agentic AI inflection point." He may be right. But the financial markets are beginning to ask a harder question: where is the matching revenue?
The Gap That Won't Close
Sequoia's David Cahn laid out the arithmetic bluntly in his widely-circulated analysis: there is approximately a $600 billion annual revenue gap between what hyperscalers are spending on AI infrastructure and what the AI ecosystem is generating in actual sales. That gap, which Cahn calculated in 2025, is widening in 2026 as capex has accelerated faster than revenue projections. According to Allianz Research, the divergence between AI capital expenditure and revenue growth is running at roughly 46% — already exceeding the 32% divergence observed during the 2001 telecom excess cycle, a period that preceded a brutal multi-year market correction in tech.
The revenue side is not zero, and it deserves fair treatment. AWS is running at roughly $150 billion annualized and growing 28% year-over-year. Google Cloud surged 63% in Q1. Microsoft's AI business crossed a $37 billion annual run rate, up 123% year-over-year. These are real numbers. The problem is not that revenue doesn't exist — it is that the investment is scaling 50% faster than revenue, which means the payback period is being pushed further out with every passing quarter. Analysts at Evercore and Bank of America now project hyperscaler capex could exceed $1 trillion in 2027. Free cash flow among the big five is under significant pressure; projections from Morgan Stanley and JPMorgan suggest the tech sector may need to issue $1.5 trillion in new debt over the next several years to finance continued construction.
Debt Enters the Picture
This is the part of the story that tends to get underreported. For most of the past decade, hyperscalers funded their capital programs internally. Their cash generation was so robust that debt was largely optional. That has changed. CreditSights documented that aggregate capex for the big five, after buybacks and dividends, now exceeds projected cash flows — meaning they are leaning on debt markets to bridge the gap. In 2025, the group raised $108 billion in new debt. Capital intensity — measured as capex as a percentage of revenue — has reached 45-57% for these companies, a ratio that looks less like a technology business and more like a capital-intensive utility or industrial firm.
MORE FOR YOU
That shift matters for investors who own these stocks based on their historical multiple and business model characteristics. The market has long assigned premium valuations to big tech on the premise that it is asset-light, throws off abundant free cash flow, and compounds earnings with minimal reinvestment. Those assumptions are being stress-tested in real time. Meta shares fell 9.25% in a single session this spring after the company raised its capex guidance — described by Mark Zuckerberg as investment for "personal superintelligence to billions of people." Investors flinched. That reaction may be the first real tremor of a broader repricing.
What the Stack Actually Looks Like Now
One underappreciated dimension of this story is how the investment is distributed across the AI value chain — and what that means for where returns eventually accrue. NVIDIA captures approximately 90% of AI accelerator spend, which at current scale represents something in the range of $180 billion in GPU purchases annually. The infrastructure layer — NVIDIA, data center operators like Equinix, power and cooling providers — is being compensated immediately and generously. The application layer, where revenue ultimately justifies the entire edifice, is still being built.
Enterprise software companies, cybersecurity firms, and workflow automation vendors are the logical beneficiaries of AI monetization — but they are downstream from the infrastructure spend, and many are only beginning to convert AI integration into pricing power. An MIT Project NANDA study from July 2025 found that 95% of enterprise GenAI pilots produced zero measurable P&L impact on roughly $30-40 billion in corporate spending. That figure is a year old and conditions are improving, but it illustrates the lag between infrastructure buildout and genuine enterprise ROI. Agentic AI — autonomous systems capable of multi-step task completion — is the mechanism most widely cited for closing that gap, but meaningful enterprise adoption is still, by most credible estimates, 12-24 months away.
Where This Leaves Investors
The AI trade is not over. The structural demand for compute is real, and dismissing the infrastructure build as pure speculation misreads the competitive dynamics at play. No hyperscaler is going to unilaterally cut spending and cede ground to rivals; the capex cycle is, in that sense, self-reinforcing regardless of near-term ROI. But the market is no longer in the phase where enthusiasm alone justifies valuation. EV/EBITDA multiples for U.S. tech and AI equities are near 25x according to Allianz Research, close to historical extremes and not far from telecom valuations that preceded the 2000 peak.
The more useful frame for 2026 is to separate the infrastructure layer from the application layer in terms of portfolio exposure. Infrastructure names — semiconductors, data center REITs, power and cooling plays — are collecting cash now and face the least execution risk. Application-layer and software plays carry higher upside if monetization arrives on schedule, but also more binary risk if enterprise adoption continues to lag. The investors who will navigate this cycle best are probably those who hold both, size positions with discipline, and resist the temptation to extrapolate last year's Nvidia returns into the next chapter of a story that is meaningfully more complex.