When confidence does all the work

Foord’s M.I.C.E. framework looks at four forces that support markets or hold them back: Money, Interest rates, Confidence and Earnings. All four matter, but they do not always matter equally. In today’s artificial-intelligence boom, Portfolio Manager RASHAAD TAYOB writes that the most dominant factor may be the C, even if the E is the most important.

Confidence is not a soft factor. Economist John Maynard Keynes (see Did You Know?) used the term ‘animal spirits’ to describe the instinctive optimism that pushes people to invest when the future cannot be known. Markets need optimism. Without it, companies would not build factories, hire people or fund new technology. But confidence can also become circular: rising stock prices validate the story, the story attracts more capital, and the new capital pushes prices higher still.

The AI boom is now in its fourth year, and this confidence cycle has been the main driver of global equity returns. The clearest winners have been the suppliers of AI infrastructure. Nvidia and other chipmakers have seen sales multiply as customers race to secure graphics processors, memory and related equipment. Profit margins have expanded to levels more commonly associated with software than hardware. Even suppliers in cyclical and historically commoditised areas — such as memory — are earning record profits, because demand has overwhelmed supply.

The harder question is whether the users of that infrastructure can earn the returns implied by today’s spending. AI providers are committing enormous sums to data centres, chips, power and talent. While revenues are rising quickly, capital expenditure is rising faster. In many cases, the result is deeply negative cash flow. That is not unusual in the early phase of a technology cycle. It does mean, however, that the boom still depends on investors believing that future revenues will justify present spending.

There are already signs of strain beneath the excitement. Leadership in the AI models race is proving less stable than markets once assumed. Some companies — like Facebook parent Meta — that spent tens of billions of dollars have fallen behind. Others — such as first-mover OpenAI — have seen growth slowing down. Open-source Chinese models are available at a fraction of the cost of leading Western models, raising an uncomfortable question about pricing power. If the best models become cheaper and less scarce, the economics of the AI providers may prove less attractive than the story suggests.

The speculative mania is most evident with the recent listing of Elon Musk’s rockets-to-robots business, SpaceX. Its listing combined three powerful market themes: Elon Musk, artificial intelligence, and the moonshot idea of space. The combination was irresistible to many retail investors. SpaceX raised a record amount of capital and briefly traded at a valuation that placed it alongside the world’s largest technology companies, despite still being loss-making and generating far less revenue than they do. Much of the value investors assigned to the company reflected not rockets or satellites, but the promise of its AI business. Yet its AI model, Grok, still trails the leading models in performance and user adoption. The share price has already surrendered much of its first surge. At the time of writing, SpaceX’s market valuation was back at an eye-popping $2 trillion.

This is how animal spirits in markets work — they lift prices, but also change the terms on which investors provide capital. When confidence is high, companies can raise money, spend aggressively and point to that spending as proof of future dominance. A virtuous cycle forms. The danger is that it can also run in reverse. If revenues disappoint, margins normalise or a market darling stock falls sharply, investors may become less willing to fund the next round of AI investment.

History offers a useful warning: railways, electrification and the internet all produced infrastructure that changed the world; they also produced many poor investment returns. Too much capital chased the same opportunity. Capacity was overbuilt, prices fell and early industry leaders disappeared.

AI may prove as important as its strongest advocates believe. But investors still face two familiar traps: paying trillion-dollar prices for companies with uncertain future cash flows, and assuming that today’s extraordinary hardware margins are permanent. Confidence has built the AI boom and may yet build something lasting. But confidence is not earnings, and animal spirits are not cash flow.