Wall Street wants cash, not another spending promise

For much of the past two years, investors have been happy to reward technology companies for spending heavily on artificial intelligence. The bigger the investment plans, the greater the excitement. This week offered a reminder that Wall Street's patience is not unlimited.

The week began on an optimistic note. Anthropic's filing for a US stock market listing helped lift the SPX500 by 0.6% on Monday, while the NAS100 gained 1.05%. By Thursday, however, investors were asking rather different questions. Alphabet's results sent its shares tumbling 12.6%, the NAS100 lost 1.8%, and Amazon subsequently fell around 6.5% in after-hours trading.

What makes this change in sentiment particularly interesting is that the underlying businesses are hardly struggling.

Alphabet reported a 25% increase in third-quarter revenue to $127.6 billion, while Google Cloud revenue more than doubled to $30.7 billion. Those are impressive numbers by almost any standard. The problem was the cost of achieving them. Capital expenditure reached $51.4 billion, exceeding operating cash flow of $42.3 billion and leaving quarterly free cash flow at negative $9.1 billion.

Amazon told a similar story. Its cloud business grew revenue by 39% to $46 billion, while group operating income reached $28.1 billion. Yet free cash flow over the preceding twelve months deteriorated to negative $42.4 billion, compared with a positive $14.8 billion a year earlier.

Nobody is seriously questioning whether demand for AI infrastructure exists. The figures make that abundantly clear. What investors increasingly want to know is how much of that demand will translate into cash for shareholders once the enormous infrastructure bills have been paid.

Microsoft provided a useful contrast. Quarterly revenue increased 19% to $92.3 billion, Azure grew 48%, and operating cash flow of $58 billion comfortably covered $39.4 billion in property and equipment purchases.

There is an important distinction emerging here. Wall Street still likes growth, but it is becoming considerably more selective about what companies must spend to achieve it.

A strong economy is complicating matters

The earnings disappointment did not unfold in isolation. Economic developments during the week gave investors another reason to question how easily the AI investment boom can sustain its current pace.

The Federal Reserve left interest rates unchanged at 3.75% to 4.00% on Wednesday, with two policymakers favouring a further quarter-point increase. The decision reflected an economy that remains surprisingly resilient, but where inflation is still proving difficult to contain.

Thursday's GDP figures reinforced that picture. The US economy expanded at an annualised rate of 3.4% in the third quarter, up from 2.2% previously. Investment in information-processing equipment contributed to that growth, highlighting just how important AI-related spending has become to the wider economy.

The inflation figures were less reassuring. September's personal consumption expenditures price index rose 3.6% from a year earlier, while core inflation reached 3.1%. Both remain uncomfortably above the Fed's 2% target.

Energy prices are hardly helping. UKOil settled at $115.49 a barrel on Thursday, despite some optimism surrounding renewed Middle East peace negotiations. For businesses already committing billions to data centres, expensive energy adds another layer of pressure to operating costs.

There were some interesting divergences across financial markets. The US 10-year Treasury yield slipped towards 4.94% as investors reassessed the outlook, while European equities managed to advance. The STOXX 600 gained 0.77% after the European Central Bank kept its deposit rate at 2.5%.

The broader concern is that AI investment is helping keep economic growth relatively strong, potentially giving central banks less reason to cut interest rates. That creates an awkward situation for technology companies committing extraordinary sums to projects whose returns may take years to materialise.

In other words, the same spending boom supporting economic activity could also help keep financing conditions tighter than investors would prefer.

From building capacity to earning a return

None of this means the AI investment story is coming to an end. Far from it. Cloud demand remains strong, computing requirements continue to grow, and the largest technology companies still have highly profitable businesses capable of funding substantial expansion.

But there is a difference between an exciting technological opportunity and an attractive investment at any price.

Building data centres is enormously expensive. Keeping them running brings further costs, from electricity and maintenance to hardware replacement. As these facilities become operational, depreciation expenses rise, and companies must demonstrate that the revenue generated is sufficient to justify the capital committed.

That is where the market's attention is likely to shift over the next 6 to 24 months.

Investors will be watching free cash flow, capital expenditure, cloud margins, borrowing requirements and, perhaps most importantly, returns on invested capital. Revenue growth will still matter, but increasingly it will need to be accompanied by convincing evidence of financial returns.

There are implications well beyond the technology sector. A slowdown in infrastructure spending would be felt by semiconductor manufacturers, construction companies, electricity suppliers and businesses across the industrial supply chain. Equally, if AI adoption delivers meaningful productivity improvements and stronger monetisation, today's investment commitments may ultimately prove justified.

One possibility investors should not overlook is that the rewards will be distributed very unevenly. Suppliers of scarce computing infrastructure can earn revenue almost immediately. The businesses purchasing that capacity, however, may spend years discovering whether their investment was worthwhile.

That distinction could become increasingly important when deciding which companies deserve premium valuations.

This week's earnings did not prove that the AI boom is unsustainable. They did, however, expose a growing reluctance among investors to accept ambitious spending plans as evidence of future success.

For much of this investment cycle, the question has been how quickly companies can build the infrastructure needed to power artificial intelligence. The next question is considerably harder. Can they earn an attractive return on everything they have built?

The AI revolution may well deliver on its technological promise. Whether it delivers equally impressive returns to shareholders is another matter entirely.

Sources & References

• sec.gov
• ir.aboutamazon.com
• microsoft.com
• federalreserve.gov
• bea.gov
• bea.gov
• reuters.com
• reuters.com