Big Tech Is Generating Record Cash Flow. So Why Are Share Buybacks Disappearing?

Imagine owning a business that generates more cash than ever before.
Revenue is growing.
Profits are increasing.
Customers are spending more.
And management is telling you that the future has never looked more promising.
Sounds like a fantastic investment, doesn't it?
But what if I told you that despite generating record cash flow, the company has almost stopped returning money to its shareholders?
That's essentially what's happening across some of the world's largest technology companies.
According to Reuters, Microsoft, Alphabet, Amazon, Meta and Oracle generated a combined US$542.9 billion in operating cash flow during the first nine months of 2026.
That's a 29% increase from a year earlier.
Impressive.
But here's the number that caught my attention.
These five companies spent approximately US$556.9 billion on capital expenditure over the same period.
In other words, they spent more on infrastructure and equipment than they generated from their operations.
And their combined share repurchases?
Just US$13.6 billion.
An 83% decline from the previous year.
So while their businesses are generating more cash than ever, shareholders are receiving significantly less through buybacks.
Why?
Artificial intelligence.
But I think there's a much bigger investment lesson here than simply whether AI spending is getting out of hand.

First, Let's Understand Where the Money Is Going
For years, large technology companies enjoyed an incredibly attractive business model.
Take Alphabet.
Its search business doesn't need to build a new physical store every time someone performs a Google search.
Meta doesn't need to manufacture a physical product every time someone scrolls through Instagram.
Microsoft can sell another software subscription without constructing an entirely new office building.
Of course, these businesses still require servers, data centres, employees and infrastructure.
But historically, their economics allowed them to generate enormous amounts of cash relative to the physical assets required.
That cash could then be used in several ways.
Management could reinvest in the business.
Acquire other companies.
Pay dividends.
Or repurchase shares.
For many years, share buybacks were an important part of how technology companies returned capital to shareholders.
But AI is changing that equation.
Training and operating increasingly sophisticated AI models requires enormous computing resources.
Data centres.
Advanced chips.
Networking equipment.
Power infrastructure.
Cooling systems.
And these investments aren't cheap.
According to Reuters, the five technology companies' combined capital expenditure approximately doubled during the first nine months of 2026.
The result?
Despite record operating cash flow, their combined free cash flow moved from a surplus of US$143.5 billion a year earlier to a deficit of approximately US$14 billion.
That's a dramatic change in how much cash remains after investment spending.
Subscribe to our newsletter if you enjoy insights like this.
Operating Cash Flow Isn't the Same as Free Cash Flow
This is where I think investors need to be careful.
A company can report record operating cash flow without necessarily generating more cash available for shareholders.
Let's use a simple example.
Imagine Company A generates $10 billion in operating cash flow.
It spends $2 billion maintaining and expanding its business.
That leaves $8 billion in free cash flow.
Now imagine the following year.
Operating cash flow increases to $12 billion.
That's 20% growth.
Fantastic.
But capital expenditure increases to $10 billion.
Suddenly, free cash flow falls to just $2 billion.
So despite growing operating cash flow by 20%, the cash remaining after capital expenditure has declined by 75%.
That's why I pay so much attention to free cash flow when analysing businesses.
Revenue tells us how much a company sells. Earnings tell us how much profit it reports. Free cash flow helps us understand how much cash remains after the capital spending needed to run and grow the business.
And right now, the distinction is becoming increasingly important for Big Tech.

Why Share Buybacks Matter More Than Many Investors Realise
Some investors don't pay much attention to share repurchases.
After all, buybacks don't put cash directly into your bank account like dividends do.
But they can still have a meaningful effect on shareholder returns.
Imagine a company earns $10 billion annually.
It has one billion shares outstanding.
That means earnings per share are $10.
Now suppose management repurchases 10% of the outstanding shares.
The company still earns $10 billion.
But there are now only 900 million shares.
Earnings per share increase to approximately $11.11.
That's an 11.1% increase in earnings per share without any improvement in the underlying business.
Each remaining share represents a larger ownership interest.
Over time, this can become a powerful contributor to per-share earnings growth.
And for companies with durable competitive advantages, strong cash generation and sensible valuations, buybacks can create meaningful shareholder value.
But there's an important condition.
The company must repurchase its shares at an attractive price.
If management spends $10 billion buying shares that are significantly overvalued, it may destroy value rather than create it.
A share buyback is ultimately another capital-allocation decision.
And like every investment, the price paid matters.
Want a concise version of Charlie Munger’s 25 human misjudgements? Download here!
So Are Fewer Buybacks Necessarily Bad?
Not at all.
In fact, there are situations where reducing share repurchases is exactly what management should do.
Imagine a company has $10 billion available.
It can use that money to repurchase its shares.
Or it can invest in a new project expected to generate an attractive return over the next decade.
If the new project can consistently earn a 20% return on invested capital, while repurchasing shares offers a much lower prospective return, reinvesting in the business may be the better decision.
That's how companies compound value.
They retain capital and reinvest it into opportunities that generate returns above their cost of capital.
But here's the problem.
We don't yet know what returns these enormous AI investments will generate.
And that's what makes the current situation so interesting.
AI Spending Is Changing the Investment Proposition
Historically, investors in large technology companies could benefit from two important sources of growth.
First, the underlying business could generate more profit.
Second, share repurchases could reduce the number of shares outstanding, increasing earnings per share.
Now, the second contributor is becoming less significant for some companies.
According to Reuters, the five companies returned just 9.3% of their combined operating cash flow to shareholders through dividends and buybacks during the first nine months of 2026.
A year earlier, that figure was 27.4%.
Alphabet and Meta reportedly made no share repurchases during the period.
That doesn't mean their businesses have deteriorated.
But it does mean the way investors might earn their returns is changing.
Instead of receiving more capital through repurchases, shareholders are effectively allowing management to retain and reinvest that money.
They're betting that the new investments will eventually produce greater economic value.
And that's a very different proposition.
The Hidden Cost of AI Infrastructure
There's another issue I think deserves more attention.
Depreciation.
When a company builds a data centre or purchases expensive computing equipment, the entire expenditure generally doesn't appear immediately as an expense on its income statement.
Instead, much of the cost is capitalised and depreciated over its estimated useful life.
Imagine a company spends $10 billion on computing equipment expected to last five years.
Very simplistically, that could translate into approximately $2 billion in annual depreciation expense.
Now imagine the company spends another $10 billion the following year.
And another $10 billion after that.
Eventually, depreciation expenses accumulate.
This matters because today's capital expenditure can become tomorrow's earnings expense.
There's also the question of technological obsolescence.
AI chips are improving rapidly.
A computing system that is highly competitive today may become less efficient relative to newer equipment in a few years.
That doesn't automatically mean the existing equipment becomes worthless.
But it does mean investors should be careful when assuming that AI infrastructure will have the same useful economic life as traditional physical assets.
And even after a company completes its initial investment, it may need to spend more money replacing or upgrading equipment.
The first investment may not be the last.

Not All Capital Expenditure Is Created Equal
This is where I'd separate capital expenditure into two broad categories.
Maintenance capital expenditure and growth capital expenditure.
Maintenance capex is the spending required to keep a business operating at its existing level.
Growth capex is spending intended to expand the business's future earning capacity.
In reality, the distinction isn't always clear.
A new data centre might support existing workloads while also providing capacity for new AI services.
But the distinction matters.
Imagine a company generates $20 billion in operating cash flow.
It spends $5 billion maintaining existing infrastructure.
And another $10 billion expanding AI capacity.
If the expansion creates attractive future profits, shareholders may eventually benefit enormously.
But if demand disappoints, the company could be left with expensive infrastructure that earns poor returns.
Worse, some of that infrastructure may eventually require additional maintenance and replacement spending.
So I wouldn't simply conclude that higher capex is bad.
I'd ask what the expenditure is intended to achieve and whether the future returns justify the investment.
The Real Question Is Return on Invested Capital
This brings us to one of my favourite metrics when analysing businesses.
Return on invested capital, or ROIC.
A company creates economic value when it can consistently earn returns on invested capital above its cost of capital.
Suppose a company invests $100 billion into AI infrastructure.
If that investment eventually generates $20 billion in sustainable annual operating profit after tax, the economics could be very attractive.
But what if it generates only $5 billion?
The company has committed the same amount of capital.
The technology may still be impressive.
Customers may still love the products.
But the financial returns would look very different.
And if the return is below the company's cost of capital, shareholders may be worse off.
This is why I don't think the question should be whether AI is useful.
AI is already proving useful across many industries.
The more important question is:
How much economic profit can these companies generate from the capital they're deploying?
The Market May Be Underestimating Both the Opportunity and the Risk
There's a reason management teams are spending so aggressively.
They see an opportunity to build infrastructure that could support an entirely new generation of products and services.
AI assistants.
Enterprise automation.
Autonomous systems.
Cloud computing.
Advertising optimisation.
Software development.
And applications that may not even exist today.
The opportunity could be enormous.
But enormous opportunities often attract enormous investment.
And when everyone invests aggressively at the same time, returns can become less attractive.
Think about telecommunications networks during periods of rapid expansion.
Or industries that build too much production capacity when demand is booming.
A growing market doesn't automatically guarantee attractive returns for every participant.
Sometimes customers benefit more than shareholders because competition pushes prices down.
The same possibility exists in AI infrastructure.
If computing capacity eventually becomes abundant, pricing power may weaken.
And companies that invested at the highest costs could struggle to earn their expected returns.
What I'd Watch as an Investor
Rather than focusing solely on headline AI capital expenditure, I'd pay attention to several things.
First, operating cash-flow growth.
Are the existing businesses continuing to generate more cash?
Second, capital expenditure relative to operating cash flow.
How much of that cash is being reinvested?
Third, incremental revenue and operating profit.
Is the additional investment translating into measurable financial results?
Fourth, depreciation and future replacement requirements.
How much will maintaining the infrastructure cost?
Fifth, returns on invested capital.
Is management creating economic value, or merely expanding the asset base?
And finally, I'd pay attention to how much capital is still available for shareholders.
Because a business can become much larger without necessarily becoming more valuable on a per-share basis.

The Bigger Lesson: Growth Isn't Free
I think this is one of the most important lessons investors can take away from the current AI investment cycle.
For years, we've become accustomed to technology companies generating enormous amounts of cash while requiring relatively modest incremental physical investment.
That made their economics incredibly attractive.
But AI is changing the capital requirements of some of these businesses.
They're becoming more infrastructure-intensive.
And shareholders are being asked to accept lower near-term cash distributions in exchange for potentially higher future earnings.
That could turn out to be a brilliant decision.
Or it could become an expensive lesson in overinvestment.
We won't know immediately.
But I wouldn't judge the success of AI investment simply by how much money companies are spending.
Nor would I judge a technology company's financial strength solely by its operating cash flow.
I'd want to know how much capital the business needs, what returns that capital can generate and how much value ultimately accrues to each shareholder.
Because generating more cash is good. But knowing what to do with that cash is what ultimately creates shareholder value.
And right now, Big Tech is making one of the largest capital-allocation decisions in its history.
Whether shareholders benefit will depend on the returns those investments eventually produce.






Comments