As euphoria over artificial intelligence drives the bull market, investors are asking these questions

Maybe artificial intelligence really will change the world. Maybe the AI euphoria that has driven the stock market higher will turn out to be fully justified by a gusher of profits in years to come.

Maybe. But it’s worth pondering all the things we still don’t know about the force that is driving so much of the current bull market.

For starters, there is the question of who is going to develop AI. Investors had assumed that a handful of U.S.-based companies would dominate the field. It now seems that may not be true.

DeepSeek, a Chinese upstart, burst onto the scene last month with an impressively smart chatbot that it claims to have developed for considerably less money than comparable U.S. products. While the specifics on cost are still unclear, DeepSeek’s success rattles the comfortable assumption that a few well-heeled U.S. pioneers will be able to build an effective fence around the AI world.

The implications for investors are profound. Over the past couple of years, billions of dollars have poured into a tiny band of private U.S.-based companies – OpenAI, Anthropic and Inflection AI, among them – that appear to have the lead in the AI race.

Meanwhile, giddy optimism around the sector has helped drive several stocks higher. Most notably, the Magnificent Seven stocks – Alphabet Inc. GOOGL-Q, Amazon.com Inc. AMZN-Q, Apple Inc. APPL-Q, Meta Platforms Inc. META-Q, Microsoft Corp. MSFT-Q, Nvidia Corp. NVDA-Q and Tesla Inc. TSLA-Q – have surged on the premise that they will all benefit hugely from AI.

That premise is now under threat. If DeepSeek and other upstarts can crowd into the market with models developed on relatively meagre budgets and with relatively few specialized chips, the Magnificent Seven may find their deep pockets and early lead don’t guarantee them any lasting advantage in the AI game. If so, then their lofty stock prices are in peril – which would be bad news for the U.S. stock market in general since the Magnificent Seven stocks make up about a third of the S&P 500 index’s market value.

While we’re waiting to see how the upstart threat turns out, investors might want to ask four questions about AI.

First, where is the killer app? ChatGPT and similar chatbots are nice-to-have but not must-have tools. They don’t generate cash for users, save huge amounts of money or spawn new industries – at least, not yet. Jim Covello, head of global equity research at Goldman Sachs, estimates that U.S. companies will pour a trillion dollars into AI development over the next few years. Yet, there is no obvious trillion-dollar problem that AI solves. This is rather odd.

Second, how smart will AI get? AI chatbots offer an impression of expertise, but they’re often “unoriginality machines,” in the words of British economist Dan Davies.

Where AI models tend to shine is either in areas that don’t demand original thought or in areas where they can brute-force a problem by running repeated attempts at an answer and seeing which one works best.

Granted, this can yield impressive results. Mr. Davies notes that AlphaGo, an AI program designed to play the classic board game Go, beat human masters by playing in a style that seemed utterly foreign to them. The key to its success was AlphaGo’s ability to play against itself, over and over, at blurry-fast computer speeds. That myriad of repetitions turned up winning strategies that human players, even champions, had quite simply missed.

Still, Mr. Davies is skeptical that AI models will develop a human-like generalized intelligence anytime soon. He also doubts that AI will immediately transform work. “I think that AI, like every other information technology, will end up creating complexity as well as processing it, that the robots will get in each other’s way just like we do, and that we are going to systematically overestimate the benefits of the technology during the initial phase,” Mr. Davies writes.

This raises a third question: Who will derive the benefits from AI? Investors appear to assume it will be the creators of the apps and the companies that find the most cunning ways to employ the technology. But that’s not a sure thing because it’s not clear how widely used AI will be.

Cost is likely to be a constraining factor because AI, in its current incarnation, gobbles up obscene amounts of power. “Even in low-compute mode, a single prompt on ChatGPT’s o3 model costs $20 to perform,” writes Dan Rasmussen of Verdad Research. He notes that Daron Acemoglu, the MIT economist, has estimated that only a quarter of AI-exposed tasks will be cost effective to automate within the next 10 years.

So that brings us to a fourth question: How much of an effect could AI have on economic growth? Perhaps less than you think. If you examine a long-term graph of labour productivity in the United States, it’s difficult to discern any clear link between technological innovation and productivity growth.

Productivity growth was robust after the Second World War, then it hit a soft patch in the late 1970s and 1980s despite the explosive growth of personal computers and office software. Productivity growth quickened for a decade between 1995 and 2005 – perhaps because of the advent of the internet? Or was that more about falling interest rates? – then it slowed again.

You can invent many stories to explain these ups and downs. It’s clear, though, that there are no cases over the past few decades where a breakthrough technology suddenly and permanently accelerated productivity growth.

Investors may want to keep that in mind when they’re betting on a glorious AI-powered future.