AI AND FREE CASH FLOW: WHAT THE SMART INVESTOR NEEDS TO KNOW

By Tejas Shankar, Equity Analyst

During the gold rush, the people who came on top weren’t the ones digging for gold; rather, they were the ones selling the pickaxes and shovels. Today’s AI boom is creating a different kind of gold rush where the miners are the world’s latest technology companies spending hundreds and billions to build the future. The question isn’t whether AI is a revolutionary technology but instead which companies can survive the cost of building the future. The reason why everyone can’t survive this race to the top is very simple: Money. The Capital Expenditure, or CAPEX, spending commitments that some of these big tech names are throwing out are unfathomable and are leading to some legacy companies that have been known for being cash kings going into negative Free Cash Flow (FCF)

Before diving into this larger trend and the growing concerns hitting headlines with Magnificent Seven (MAG 7) companies, let’s first understand what Free Cash Flow is. Let’s set the stage with you running a lemonade stand. You sell $500 worth of lemonade in the summer, with your costs including $100 on the lemons, $50 on cups & $200 on a new cart. After all of that spending, you are left with $150 in your pocket, and that is your free cash flow. That is the actual money you generated after paying for everything, including the investments you made to grow your lemonade stand.

Wall Street often focuses on the income statement, also known as Profit & Loss (P&L), which tells only part of the story. On the P&L Statement, it doesn’t show the full cost of, let’s say, the cart, for example. This is due to the fact that accountants spread the cost over several years through depreciation. So if you are looking for the actual cash situation, the cash flow statement is where smart investors go. It shows what actually moved in and out of your bank account. The media highlights the profit and loss statement, which skews the general public’s opinion at times.

The reason why FCF is so important is that it tells us what the business has after running the business and making the investments necessary to grow. It is the cash available to pay dividends, pay debts, do stock buybacks, make acquisitions, and more, and if a company doesn’t have that buffer room, it starts to raise red flags. A company can show profit while also burning through cash and vice versa.

Big Changes at the Mag 7
For most of the 2010s and 2020s, the Mag 7, including Apple, Microsoft, Google, Amazon, Meta, NVIDIA, and Tesla, were the best of the best and what all companies aimed to emulate. They were generating free cash flow at a prolific and unprecedented rate. What makes them special is that they have very low marginal costs. Take a look at Google, for example. Once they made their search algorithm, each additional query cost them almost nothing. Once Apple built its app store, adding another app and generating revenue from it cost them almost nothing. When Microsoft wrote Office 365, adding one more user cost almost nothing, and this is what allowed many of these MAG 7 companies to stand out. Building this business model and getting the first couple of customers was difficult, but after that, getting the next millionth customer was essentially just pure cash in their pocket.

Over time, due to these superior advantages, MAG 7 companies were getting uninterrupted FCF growth. In 2025, Google pulled north of $70B in FCF, and Apple almost $100B. Meta had turned its advertising juggernaut into a cash-printing machine. Microsoft’s Azure cloud revenues continue to climb at a strong growth rate. Amazon’s AWS was the most profitable cloud business in the world. All of these companies share the fact that they essentially own a machine that takes their customers and turns them into cash with extreme efficiency. Unfortunately, this golden era seems to be coming to an end due to a capital spending arms race with no end in sight.

Capital Expenditures(CAPEX) are the money that a company spends to buy or build long-term assets such as factories, data centers, servers, and chips, just to name a few. Unlike an operating expense such as buying lemons for the lemonade stand, CAPEX shows up on the balance sheet as an asset and is gradually expensed over time through depreciation. The key thing to note here is that CAPEX comes out of your cash immediately, even if the P&L statement only shows a small percentage of the cost. High CAPEX commitments compress FCF almost instantly.

What happened in the past year is unlike anything corporate America has seen before. 4 of the largest hyperscalers (Google, Amazon, Microsoft, Meta) collectively spent north of $400B in 2025, and that figure is expected to reach north of $725B in 2026. That is a 77% increase in a single year. Goldman Sachs estimates the hyperscalers will collectively spend $5.3T on Capex between 2025 and 2030. That is an astronomical number that will be spent on data centers, AI chips, energy infrastructure, and more over the next 5 years. But at what cost are they scaling their operations this much? The cost is their free cash flow. These companies are already tapping into the debt market at alarming rates to keep up with their spending commitments, and it should be worrisome for investors.

Most figures are approximations. Source: Company earnings releases, Yahoo Finance, Forbes, Finbox. Date as of August 2026

If it hasn’t already been made clear yet, the trigger for this spend is Artificial Intelligence. Every one of these companies realized that whoever develops and builds out the most AI infrastructure will dominate the next decade. They are building data centers faster than any technology in capital markets’ history. They are buying Nvidia chips in quantities that turned them into the most valuable company in the world at one point. They are signing 20-year power purchasing agreements, and they are borrowing tens of billions of dollars to fund their infrastructure since their FCF is depleted. Most notably, Google, which was known for being one of the best cash-generating companies, just had its first quarter of negative FCF in its history.

The fears and headlines are starting to turn into realities. When a company has negative FCF, it means they are spending more in real cash than they are generating in operations, and to survive, they must withdraw from cash reserves, issue new debt, or raise equity, and we are seeing this desperation come to fruition as Google continues to raise the share they are listing.

Red Flag, Green Flag
With all of these red flags being raised about CAPEX with Mag 7 companies, there is definitely one that could stand alone as triumphant in this chaos. Before picking that, it is important to address Oracle as a vivid warning of what happens when a strong company with strong FCF tried to compete in the AI CAPEX arms race.

Oracle spent years as the gold standard for disciplined buybacks and returning cash to shareholders. Then the company announced $50B in CAPEX guidance for FY26 to fund AI data centers for its clients such as OpenAI and Meta. The problem was that Oracle’s underlying cash generation can’t support that level of spending commitment, with its FCF deficit hitting $23.7B, a massive turnaround from near breakeven just a year ago. Its debt-to-equity ratio skyrocketed to 500%, and banks were warning that its debt could be downgraded to junk status. The stock fell 12% on this news in a single session, wiping out billions in market value. Oracle is an example investors should look back to when analyzing hyperscalers’ capex commitments, not to show that the AI opportunity is wrong but rather to show that hopping into a capital arms race without the balance sheet to back them could drive them to the ground. As Charline Munger once said, Ladies, liquor, and leverage drive men broke.

Of these MAG 7 companies, Apple has been taking the back seat. They have been notorious for their failure with Apple intelligence and outsourcing their AI integration into their products to peers such as OpenAI and Google’s Gemini. Their CAPEX, as seen in the table above, came in the low teens. Just because their AI implementation is a failure doesn’t mean the company as a whole is. Their core business is still promising, and they are much more focused on their hardware and development of new products. Whoever wins the AI race, the Apple iPhone is still the de facto “gateway” to the internet. Almost everyone accesses AI through an Apple hardware platform.

They just hired a hardware-minded CEO to replace former CEO Tim Cook to guide the direction the company is taking. They have been trading spots with Nvidia as the world’s most valuable company for a couple of months now and currently sit at the top. While its peers are in an arms race for AI compute, they have gladly accepted to outsource their AI needs and are focusing on the moat that they have with their Apple products. So of the MAG 7 companies, Apple is probably the most distant from this race and can comfortably stay out of the concerns of tapping into debt markets and going into negative FCF to fund their CAPEX.

So what’s the takeaway? One or two of the hyperscalers will eventually generate returns that justify the investment. Microsoft’s Azure AI business has already crossed $37B in annualized revenue run rate. Google’s cloud grew 63% YoY in Q2 of 2026. These are real businesses with real momentum, and the question isn’t whether the revenue will come. Rather, it is whether it comes fast enough to service the debt and close the FCF gap before investors lose hope. The companies most at risk are those borrowing most aggressively on a thin FCF buffer.