To The Moon

For the past two years, the AI investment story has largely been about growth. Technology companies have been racing to build data centres, secure advanced chips, expand cloud capacity and develop increasingly powerful AI models. Every major spending announcement has reinforced the belief that AI could become one of the biggest sources of future corporate earnings.

But the market is beginning to look beyond the size of these investments.

As earnings reports reveal higher capital expenditure and pressure on free cash flow, investors are facing a more important question: Can the companies spending billions on AI eventually generate enough additional cash to justify the cost?

This is where the distinction between revenue growth, accounting profit and free cash flow becomes particularly important.

A company can report strong revenue growth and still experience weaker cash generation if it is spending heavily on infrastructure. Data centres require significant upfront investment in buildings, electricity, cooling systems, networking equipment and specialised processors. These costs can arrive well before the additional revenue is fully realised.

The challenge is not simply whether AI demand exists. It is whether that demand can produce attractive returns on the capital being invested.

Amazon: AWS growth is impressive, but cash flow deserves attention

Amazon’s latest results demonstrate both sides of the AI opportunity.

AWS revenue increased 39% to $48.1 billion, reflecting strong demand for cloud computing and AI infrastructure. This is an important development because Amazon is not relying solely on consumer-facing AI products. It is also providing the infrastructure that other businesses need to develop and operate AI applications.

However, its trailing 12-month free cash flow declined 35% to $42.2 billion.

That creates an interesting contrast. AWS is growing rapidly, but Amazon’s overall cash generation is facing pressure as the company continues investing heavily in infrastructure.

Investors need to distinguish between spending that supports future growth and spending that may take too long to generate an adequate return.

Amazon’s opportunity is substantial. If AWS can maintain strong growth, improve infrastructure utilisation and generate higher operating profits from AI workloads, today’s investment could become a major competitive advantage.

But strong cloud revenue growth alone is not enough. The market will want evidence that the additional capacity is translating into sustainable cash generation.

Microsoft: Can its cloud leadership support the investment cycle?

Microsoft has a different advantage. Its AI strategy is integrated across Azure, enterprise software, productivity applications and its wider business ecosystem.

This gives the company several potential ways to monetise AI.

Azure can generate infrastructure revenue from customers running AI workloads. Copilot products can create additional subscription opportunities. AI features can also strengthen Microsoft’s existing enterprise relationships and make its software ecosystem more valuable.

This diversified monetisation strategy could be important because Microsoft does not need every AI investment to generate revenue through a single product.

However, the company still faces a major financial challenge. Expanding data-centre capacity requires significant capital, while demand for advanced computing continues to place pressure on the availability and cost of infrastructure.

The key question is whether Microsoft’s AI-related revenue can grow faster than the cost of building and operating the infrastructure behind it.

I would pay particular attention to Azure growth, operating margins, capital expenditure and free cash flow. If cloud demand remains strong while margins and cash generation hold up, Microsoft could demonstrate that its AI investments are producing meaningful economic returns.

Meta: Turning AI into better advertising economics

Meta’s approach is different from Amazon and Microsoft.

Rather than relying primarily on selling cloud infrastructure, Meta uses AI to improve its advertising business, content recommendations, engagement and targeting.

This is an important distinction. Meta already has a large advertising platform and an established customer base. If AI improves the effectiveness of its advertising tools, the company may be able to generate more revenue from its existing ecosystem.

Better recommendations can increase user engagement. More relevant advertising can improve conversion rates for advertisers. Improved efficiency could also strengthen margins over time.

The potential return is significant because Meta can apply AI across a business that already generates substantial revenue.

However, its infrastructure spending remains substantial. The company is investing in computing capacity to support its AI ambitions, while also pursuing longer-term developments that may not produce immediate financial returns.

For shareholders, the important question is whether AI can improve advertising revenue and profitability enough to compensate for the rising investment.

Meta’s success should therefore be measured not only by the sophistication of its AI models, but also by improvements in advertising performance, operating margins and free cash flow.

Why free cash flow matters more than ever

Free cash flow is the cash a company generates from its operations after capital expenditure. It provides an indication of how much cash remains available for debt reduction, acquisitions, dividends, share buybacks and other corporate priorities.

During a period of heavy investment, free cash flow can decline even when the underlying business is performing well. This is not necessarily a warning sign on its own.

The bigger concern is whether the decline is temporary and supported by a credible growth opportunity, or whether spending continues to increase without a corresponding improvement in financial returns.

For AI infrastructure providers, utilisation rates will be particularly important. Building more data-centre capacity only creates value if customers use it and are willing to pay prices that support attractive margins.

There is also a risk that today’s competitive spending eventually creates excess capacity. If several companies expand infrastructure at the same time, pricing pressure could emerge, particularly if demand does not grow as quickly as expected.

Another factor is depreciation. Data centres and computing equipment are expensive assets, but their useful economic lives may be relatively short as technology advances. Higher depreciation expenses could affect reported earnings even after the initial cash investment has been made.

This means investors need to consider both current cash expenditure and the longer-term cost of maintaining and replacing AI infrastructure.

Three financial signals I would watch

1. Free cash flow relative to capital expenditure

Are companies generating enough operating cash to fund a meaningful portion of their infrastructure investment? A widening gap deserves attention, particularly if management cannot explain when the spending should begin producing returns.

2. Incremental revenue and profit from AI

It is not enough for total revenue to increase. Investors need evidence that AI-related services are contributing additional revenue and operating profit. The strongest businesses should eventually demonstrate that each additional dollar invested is generating an attractive amount of incremental earnings.

3. Return on invested capital

This may be the most important long-term measure. If companies continue committing billions to AI infrastructure, the returns generated from that capital should eventually exceed the cost of financing and maintaining it. Otherwise, revenue growth may look impressive without creating equivalent shareholder value.

My view: The next AI winners may not be the biggest spenders

I remain positive about the long-term potential of AI, but I think the market is entering a more demanding phase.

The first stage rewarded companies that could secure chips, build capacity and attract AI customers. The next stage may favour companies that can turn that infrastructure into recurring revenue, stronger margins and sustainable free cash flow.

Amazon has impressive cloud growth, Microsoft has a broad enterprise monetisation strategy, and Meta has the opportunity to use AI to improve an already established advertising business.

Each has a different route to financial returns, and there is no guarantee that the company spending the most will ultimately create the most shareholder value.

I would rather see evidence of improving cash generation and returns on investment than simply another announcement of increased AI capital expenditure.

The AI opportunity is real, but valuations ultimately depend on financial results. At some point, the billions invested in AI infrastructure must show up in the cash flow statement, not just in management presentations and future growth forecasts.

What Do You Think, Tigers?

Among Amazon, Microsoft and Meta, which company has the clearest path to turning AI investment into sustainable free cash flow?

Would you favour Amazon’s cloud infrastructure growth, Microsoft’s enterprise AI ecosystem, or Meta’s ability to use AI to improve advertising returns?

And how long are you prepared to wait for today’s AI spending to translate into measurable shareholder returns?

For discussion purposes only, not financial advice.