Changing Engines Mid-Flight

The paradox of an extraordinarily profitable company

On 28 October 2026, Alphabet reported impressive quarterly results: $127.6 billion in revenue, $40.6 billion in operating income and 103% year-on-year growth in Google Cloud revenue. source

Yet its shares fell by nearly 10% in after-hours trading. source

Why would such a profitable company unsettle investors?

Beyond several results falling short of market expectations, one question stands out: how much will the next phase of growth cost?

One figure illustrates the scale of the challenge: $51.4 billion in capital expenditure across Alphabet as a whole in a single quarter, the vast majority directed towards technical infrastructure supporting, among other things, AI.

The company generated $42.3 billion in operating cash flow during the quarter. However, capital expenditure exceeded that amount, resulting in negative quarterly free cash flow of $9.1 billion.

That figure requires context.

According to management, cash generation was also affected by inventory build-up associated with TPU systems and the payment of approximately $5.2 billion relating to a European Commission fine originally imposed in 2018. source

And over the trailing twelve months, Alphabet still generated approximately $19.7 billion in positive free cash flow.

The challenge is therefore not an inability to generate cash. It is the growing intensity of the investments required to transform the business.

Alphabet remains extraordinarily profitable at the operating level. But financing the next technological generation requires enormous resources long before the returns on those investments can be fully demonstrated.

This is where the real leadership question begins.

The dilemma: protect the present or finance the future?

For more than twenty years, Google has refined one of the most effective business models in digital history.

Its search engine attracts users. Advertising converts their attention and search intent into revenue. That revenue funds an extensive ecosystem of products, technologies and infrastructure.

Artificial intelligence introduces a profound shift.

We no longer expect a search engine merely to point us towards information. Increasingly, we expect an assistant to understand our intentions, compare alternatives, prepare decisions and perform tasks.

Value may therefore shift from accessing information to solving the underlying problem.

This does not necessarily mean the disappearance of advertising. New forms of monetisation may emerge. But their economics still need to be demonstrated at scale.

Unlike a start-up building its business model from scratch, Alphabet must transform a business that continues to perform remarkably well.

Google Search revenue still grew by 12% year on year in the third quarter.

That is precisely what makes the dilemma so difficult: there is no obvious signal telling management that the historical model must be abandoned.

Wait too long, and competitors may capture the next source of value.

Move too aggressively, and the company may undermine the profitability that finances innovation.

Leadership therefore means accepting a period in which two models coexist, sometimes compete with each other and require substantial resources.

The new engine is already running. But what is it really delivering?

Alphabet is not starting from zero.

Google Cloud generated $30.7 billion in revenue and $11.1 billion in operating income in the third quarter of 2026.

According to Sundar Pichai, the Gemini application has reached one billion users. source

Alphabet also benefits from an exceptional distribution advantage through Android, Chrome, Gmail, Maps, YouTube and Workspace.

These are encouraging results.

But they do not establish that the overall return on Alphabet's AI investments is already satisfactory.

Google Cloud includes established businesses alongside AI-related activities. Some AI research and development costs are also accounted for at group level.

We therefore need to distinguish three fundamentally different concepts.

Adoption ≠ Value Creation ≠ Profitability.

Adoption tells us that people are using a product.

Value creation tells us that something is genuinely improving for them.

Profitability tells us that the company can deliver and capture that value on economically sustainable terms.

And even when an activity becomes profitable, another question remains: does its return justify the capital invested?

Confusing these dimensions leads to poor investment decisions.

A product can achieve spectacular adoption while remaining economically fragile. Conversely, a relatively invisible activity may already be highly profitable.

User numbers alone are not an adequate answer to the question of return on investment.

What business leaders can learn from Alphabet

Alphabet's dilemma is not exclusive to technology giants.

A bank may need to rethink its distribution model. A retailer may see the relationship between customers and sales channels change. A professional services firm may discover that clients increasingly want to buy outcomes rather than hours of work.

Some transformations are incremental. They improve processes, productivity or service quality.

Others are more fundamental. They challenge how the company creates value and generates revenue.

In both cases, leaders must make decisions without knowing precisely what the future model will cost, how long it will take to develop or what its economics will look like.

But not every organisation has the same resources to confront this uncertainty.

Alphabet can finance several strategic bets simultaneously. Most companies cannot. For them, the discipline is not to invest more, but to learn fast enough to know where to invest.

A small or mid-sized business often has to make choices earlier, with less information and a narrower margin for error.

The temptation is to demand a complete budget, a detailed business case and a projected return on investment before anything begins.

But how can we accurately calculate the return of a business model we have not yet tested?

The right answer is not necessarily to finance the entire transformation. It may be to finance the reduction of uncertainty first.

Consider a professional services company exploring the use of AI to prepare commercial proposals.

Before funding a company-wide rollout, it could test the new process on twenty proposals.

The team would measure time saved, proposal quality, necessary corrections and actual adoption.

But it would also need to answer a more important economic question: can those improvements be converted into business benefits?

Will the organisation produce more proposals, improve its conversion rate or reduce costs?

If employees save two hours per proposal but the released time is never redirected towards value-creating work, productivity has improved without necessarily improving profitability.

A productivity gain is not automatically an economic gain.

The experiment must therefore provide evidence for an investment decision, not merely demonstrate that the technology works.

This approach starts with a discovery phase: understanding existing processes, current costs, customer expectations and the assumptions on which success depends.

Even a simple prototype can then test those assumptions against reality, identify risks and produce a more credible estimate of the investment required.

At this stage, the objective is not to prove at all costs that the project will be profitable.

It is to gather enough evidence to decide whether to proceed, adjust the approach, investigate further or stop.

We are not trying to replace uncertainty with artificially precise forecasts. We are progressively building the evidence needed to make better-informed decisions.

Four questions before accelerating

To govern this kind of transformation, an executive leadership team should be able to answer four questions.

1. What must we preserve?

Which customers, revenues, margins or competencies in our existing business are essential to financing the future?

And which elements of that business might become obstacles to its transformation?

2. Which assumptions must we test?

What must be true for the new model to succeed?

Which assumptions are both the most uncertain and the most consequential?

And what experiments could confirm or disprove them?

3. How will we measure value?

What is our starting point?

Which indicators will demonstrate genuine improvements for customers, employees and the business?

What will be the full cost of delivering those improvements?

And how will we distinguish theoretical productivity gains from economic benefits actually realised?

4. When will we accelerate, adjust or stop?

Which results will justify additional investment?

Which signals should cause us to reconsider the project?

And at what point might preserving the historical model become riskier than transforming it?

These questions do not eliminate uncertainty.

They make it possible to govern it.

The real challenge: knowing when to transfer power

Alphabet's transformation is not simply a race to build AI models or data centres.

It raises a much older question that almost every business eventually encounters.

How do we build our future using the resources of our present without becoming prisoners of what made us successful?

The danger is not only moving too slowly. It is also investing too quickly without understanding what genuinely creates value.

A leader's role is not to predict the future with certainty. It is to establish the conditions for making sound decisions even when the available information remains incomplete.

Alphabet must change engines mid-flight.

It has the resources to attempt the manoeuvre.

But having enough fuel is not sufficient.

The leadership challenge is to know which engine to preserve, which one to develop, how much to invest and under what conditions to transfer power from one to the other.

The real risk is not keeping the old engine running. It is never defining the conditions under which power should be transferred to the next.