By Philip van den Berge
Every large technology company is currently making a bet about artificial intelligence, and most of them are making it with capital expenditure. Apple is the conspicuous exception, and the gap is far wider than the coverage suggests.
Here is what the most recent annual filings show.
| Company | Revenue | Capital expenditure | Capex as % of revenue | Free cash flow |
| Apple (FY2025) | $416.2bn | $12.7bn | 3.06% | $98.8bn |
| Microsoft (FY2026) | $331.8bn | $115.9bn | 34.94% | $67.0bn |
| Alphabet (TTM) | $445.9bn | $132.4bn | 29.70% | $53.3bn |
Apple spends roughly one dollar in thirty-three on property, plant and equipment. Microsoft spends one in three. Every figure in this article is computed directly from the companies’ own SEC filings using Intrinsiqq, so each one traces back to a document you can open.
The cash consequence is already visible
Capital expenditure is not an expense on the income statement, so it only reaches reported profit gradually, through depreciation. It comes out of cash flow, and that is where the divergence shows.
Microsoft’s revenue grew from $245.1bn in FY2024 to $331.8bn in FY2026, an increase of more than a third. Over the same period its free cash flow fell, from $74.1bn to $67.0bn. Alphabet shows the same pattern: revenue up from $307.4bn in 2023 to $445.9bn on a trailing basis, free cash flow down from $69.5bn to $53.3bn.
Apple’s free cash flow across its last four fiscal years ran $111.4bn, $99.6bn, $108.8bn and $98.8bn. Flat, and lumpy, but not eroding. Across the past five years Apple’s reported fundamentals show free cash flow running between 88% and 116% of net income. Almost every dollar of accounting profit becomes an actual dollar.
The result is that Apple, despite being the slowest growing of the three, now converts more cash than either of them.
That is the bull case stated plainly, and it is a real one. It is also the bear case, because the reason Apple’s cash conversion looks so good is that it is not building the infrastructure its competitors believe the next decade runs on.
Where Apple’s growth has actually come from
Strip out the narrative and the last four years look like this.
Apple’s products revenue went from $297.4bn in FY2021 to $307.0bn in FY2025. That is 3.2% across four years, a decline in real terms. iPhone went from $192.0bn to $209.6bn.
Services went from $68.4bn to $109.2bn, up 59.6%, and now account for 26.2% of revenue against 18.7% four years ago.
Because services carry much higher margins than hardware, that mix shift dragged the whole company’s gross margin from 41.78% to 46.91%. A 5.13 percentage point expansion at this scale is remarkable, and it happened without Apple selling meaningfully more devices.
A large share of the earnings growth is arithmetic
Between FY2021 and FY2025 Apple’s net income rose 18.3%, from $94.7bn to $112.0bn.
Diluted earnings per share over the same period rose 33.0%, from $5.61 to $7.46.
The difference is the share count, which fell 11.0% from 16.865bn to 15.005bn, driven by $438.6bn of buybacks across those five fiscal years.
Apply FY2025’s net income to FY2021’s share count and EPS would have been $6.64 rather than $7.46. So of the $1.85 of EPS growth, $1.03 came from the business and $0.82 came from there being fewer shares. Roughly 44% of Apple’s earnings-per-share growth over four years came from the buyback.
Buybacks are a legitimate use of cash, and Apple’s balance sheet supports them easily: net debt is $21.95bn against a business earning more than $110bn a year. The point is only that 33% EPS growth and 33% business growth are not the same thing. If you hold Apple through a fund or a pension, the same distinction is worth applying to your own returns, and simple investing calculators will separate compounding from contribution in the same way.
What the price assumes
At $332.27, Apple trades at 38.1 times earnings and 35.8 times free cash flow.
Over the past four fiscal years Apple’s revenue compounded at 3.3% a year and its free cash flow at 1.5%. Consensus estimates compiled by Visible Alpha and S&P Global put FY2028 revenue near $557bn, which from the FY2025 base of $416.2bn implies roughly 10% a year.
So analysts and the market broadly agree with each other, and both disagree with the last four years of filings. The case for the current multiple rests on a re-acceleration that has not yet appeared in the accounts.
It is worth being precise about what would have to drive it. It cannot be capital expenditure, because Apple is not spending it. It would have to come from services continuing to compound at double digits, from AI features lifting the device replacement cycle, or from margin expansion continuing past 47%.
A discounted cash flow built on Apple’s own figures rather than on forecasts makes the gap concrete. Projecting the last four years of free cash flow growth forward, with an 8% discount rate and a 25% margin of safety, produces scenario values of $122.05, $172.37 and $234.07 per share, all below the market price.
Those numbers are deliberately backward looking and should be read that way. Swap the growth input for the 10% analysts expect and the output changes substantially. That is the honest state of the argument: Apple is cheap if the forecasts are right and expensive if the last four years are the better guide.
The question underneath
Apple has the strongest cash conversion in big technology because it is not spending on the thing its competitors are spending on. If AI infrastructure turns out to be the decisive investment of the decade, that discipline will look like a strategic error that flattered the cash flow statement in the meantime.
If it turns out to be an expensive arms race, Apple will have sat it out while returning $438.6bn to shareholders.
The filings cannot tell you which. They can tell you that the current share price has already picked a side.
Figures are taken from Apple’s, Microsoft’s and Alphabet’s annual filings with the SEC. Apple data covers fiscal years 2021 to 2025. Written by Philip van den Berge, Founder & CEO of Intrinsiqq. This is not investment advice.
