For more than a decade Alphabet, Microsoft, Amazon and Meta Platforms, the so-called hyperscalers, have been lauded for their ability to generate free cash flow, while simultaneously investing in growth.
Free cash flow is the cash a company generates after covering all its expenses and capital expenditures. It demonstrates an ability to self-fund expansion and indicates rude financial health.
The big four technology firms have, until the appearance of ChatGPT in late 2022, benefited from asset-light business models which required little capital to grow their businesses.
For example, Microsoft has grown its earnings by around 25% a year, while growth in capital expenditure has lagged.
This ‘asset light’ model allowed Microsoft to grow free cash flow from $25 billion in 2016 to $56 billion in 2021, which was used to reward shareholders with share buybacks to the tune of approximately $100 billion.
Those ‘asset light’ days appear to be receding fast as Microsoft ramps-up AI capital expenditures to compete with spending by other hyperscalers.
As the table shows, the combined capital expenditures of the four firms have grown appreciably over the last three years with Alphabet planning to double AI spending in 2026.
This has resulted in higher capital intensity (capital expenditure as a proportion of revenue) which has surged from around 12% in 2021 to around a third in 2026.
It echoes the dotcom era when capital intensity of telecoms firms reached 25% to 35% as they raced to lay down fibreoptic cable to build capacity.
The difference is that hyperscalers can afford to self-fund their rollout, but as we discuss shortly, they are increasingly turning to debt to finance spending.
One important aspect to consider for shareholders is that higher capital intensity tends to reduce returns on capital, everything else being equal.
Increased AI sending has also put the brakes on share buybacks with Alphabet, Microsoft and Meta reducing their combined repurchases to $12.6 billion in the final quarter of 2025, representing a drop of 74% from their 2021 peak.
Amazon has not repurchased shares since late 2022 with little prospect of it restarting buybacks given the company’s record $200 billion capital expenditure plan for 2026.
Despite some rotation, investors still appear to be backing AI
As the chart shows, although the combined market value of the big four tech firms has pulled back a little in recent weeks, it remains around 50% higher than in 2021.
This means the free cash flow yield (free cash flow divided by market value) has dropped to 1.6%, compared with 3% in 2021.
This can be interpreted in one of two ways; either investors expect investments in AI will eventually pay off, or less generously, there is further room for share prices to adjust downwards.
Why are hyperscalers using debt?
Despite sitting on a combined $300 billion of net cash the hyperscalers have recently turned to the debt markets to finance their AI spending.
The hyperscalers’ strong balance sheets mean they are seen as good borrowers, allowing them access to low rates of interest. Using debt frees up cash for other uses like acquisitions and paying dividends.
Notable debt issues include Google-owner Alphabet’s recent upsized a $15 billion debt sale to $32 billion due to strong investor demand. It included a rare 100-year bond.
The combined issuance of the big four recently reached $116 billion, which is roughly four times the annual average debt issuance over the prior five years.
Despite their sudden appetite for debt, projected financial leverage remains modest among the hyperscalers. The huge $200 billion of planned capital expenditure in 2026 means Amazon is the most likely of the hyperscalers to continue tapping the debt markets.
Martin Gamble: Shares and Markets Writer
Martin Gamble is Shares and Markets writer at AJ Bell. He was previously the Education Editor of Shares Magazine. He has been with the business since 2019.
Martin graduated from the University of Kent in...
