Big Tech AI Capex Spending Surge Threatens Market Stability
Until recently, capital expenditures were something rarely discussed outside of CFO summits, accounting conferences and maybe a budget meeting at a heavy manufacturing company. But lately, capex has been on everyone’s mind.
Thanks to the surge of AI and the need for massive data centers to support it, Big tech’s capex as a share of revenue has risen to its highest level in over a decade. That’s got many investors and their advisors worried. If earnings can’t keep up with the massive investments companies are making, that could stifle the “AI trade” in big tech, which has largely driven the strong bull market over the past several years.
When it comes to big tech, we think about nimble, non-capital-intensive businesses with high margins. The Mag 7 and other hyper-scalers make mountains of cash thanks to their brainpower—intellectual property, software, algorithms, advertising, platforms and network effects—not on physical assets.
A business that relies heavily on physical assets can still generate strong returns for shareholders. Just look at how value stocks have historically outperformed growth stocks. But capex can cause more ups and downs in earnings. That’s because asset-intensive companies are vulnerable to demand fluctuations, pricing pressure, and the risk that equipment becomes outdated and needs to be replaced.
Investors today are wondering whether the huge surge in capex could reduce stock buybacks or force the AI sector to lean more heavily on debt financing. Either outcome could introduce more volatility and pressure on valuations.
That means we could see more volatility and some downward pressure on sky-high valuations of U.S. equities. Before going further, let’s make sure we’re on the same page when talking about capex. Simply, capex refers to the money a company spends to acquire, upgrade or maintain long-term productive assets. These are investments that create capacity for future growth, not day-to-day operating costs. For example:
-
Building or expanding data centers.
-
Purchasing servers, GPUs and networking hardware.
-
Constructing factories or warehouses.
-
Buying equipment, vehicles or machinery.
-
Long-term software or infrastructure investments.
These expenditures appear on a company’s cash flow statement under “Investing Activities,” but most of your clients aren’t spending much time there.
Why Capex Matters
Capex tells you:
-
How aggressively a company is investing in future growth.
-
Whether free cash flow will tighten.
-
How much external financing (debt or equity) might be needed.
-
Whether returns on capital are likely to rise or compress.
When a company is showing high capex in its financial statements, it could mean they’re feeling confident and investing heavily in the future. Or it could mean they’re really worried about the future, and they’re investing heavily just to keep up with the competition. The Mag 7 and other “hyperscalers” are spending so heavily on AI infrastructure (data centers, GPUs, custom silicon, power buildouts) that their free cash flow could be turning negative. They may need massive debt issuance to fund the buildout, and because of their massive size, this could stress the corporate bond market. And the bond market is already showing signs of strain. Spreads are widening for software and tech IG debt. Leveraged loans tied to software are down significantly so far in 2026, and private credit lenders are selling off. One of the biggest problems with the AI mania is that the numbers don’t seem to add up—yet. For instance, Microsoft, Amazon, Alphabet (Google), and Meta are spending an estimated 100% of operating cash flow on Capex in 2026, compared to their historical average of 40%. That doesn’t give them much margin for error or allow them room for stock buybacks. Based on Goldman Sachs analysis (as of early 2026), the massive capital expenditures in AI by hyperscalers—projected to exceed $500 billion to $600 billion annually in the coming years—are putting significant pressure on profitability. To maintain historical returns on capital, these companies would need to generate over $1 trillion in annual profit, which is more than double the 2026 consensus estimate of roughly $450 billion in income.
This level of spending isn’t sustainable, and the payback period may be longer than expected – if at all. Big Tech’s Capex on AI has become so large that some players are at risk of going cash-flow negative, according to analysts at Evercore ISI.
Another potential problem is that the revenue gap must be filled with debt, such as corporate bonds, private credit, ABS/CMBS, and sovereign financing, among other sources. If left unchecked, AI could be bringing us into another debt bubble.
How Advisors Can Help
As advisors, it’s important to help clients diversify risk to avoid market bubbles. For instance, the 2000-2003 tech crash did not extend to other sectors of the market if you study the returns by asset class. The tech market and associated companies fell nearly 40%, but a well-diversified portfolio dropped only 10% during that same time period.
Advisors who chase returns will always be behind the market. The only way to track the market is to be the market. This is accomplished through wide diversification and by committing to holding one’s position through the market’s ups and downs. The biggest problem with getting out of the market is knowing when to get back in. You’re never going to get a memo or receive an “all clear” text that tells you the danger has passed and it safe to get back in. Wealth managers must look beyond the moment and keep an eye on historical performance and the likelihood that current conditions will extend into the future. Looking only at the present can cause you and your clients to extrapolate short-term changes into long-term mistakes. The core of an excellent wealth manager is to help the clients stay calm and certain, even in times of great chaos and volatility. It is not a job for the faint of heart. But I know you can do it.