Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
A year ago, the generative AI mania sweeping through Silicon Valley and Wall Street faced a serious reality check.
In a widely quoted note, Goldman Sachs’ head of equity research, Jim Covello, questioned whether the companies planning to pour $1bn into building generative AI would ever see a return on the money. A partner at venture capital firm Sequoia, meanwhile, estimated that tech companies needed to generate $600bn in extra revenue to justify their extra capital spending in 2024 alone — around six times more than they were likely to produce.
The warnings helped to trigger the first real test of investment sentiment since the launch of ChatGPT electrified the industry. Revenue from the end customers who were meant to benefit from this new technology was negligible. Where were generative AI’s “killer apps”? It led to a summer of angst for tech investors.
A year on, the leading AI stocks have just been through another volatile swing. Shaking off worries that began with the low-cost AI models created by China’s DeepSeek, Nvidia rebounded to hit a new record high this week, a gain of about $1.5tn in stock market value from its April low. Microsoft has also registered a $1tn market cap bounce in less than three months.
What’s remarkable, though, is how little has changed in generative AI’s broader revenue outlook since the warnings of a year ago. The hope (and hype) is as powerful as ever, but it is still hard to see where the returns will come from to justify the huge capital spending on AI, at least in the short term.
On the cost side, the effects of AI mania are all too apparent. The four tech companies leading the charge — Alphabet, Amazon, Meta and Microsoft — increased their capital spending by nearly two-thirds, or $95bn, in 2024. As this year got under way, they were planning to boost capex by another $75bn.
Some content could not load. Check your internet connection or browser settings.
A further escalation through the rest of the decade is baked into expectations. Bank of America Securities predicts that for the tech industry as a whole, spending on data centres will jump from $333bn last year to about $1tn in 2030. By the end of the period, 83 per cent of the money will go into AI-related investments.
On the revenue side of the equation, meanwhile, some of the AI leaders are starting to notch up big percentage increases in business — but the extra revenue is counted in the tens of billions rather than the hundreds.
Early this year, Microsoft said its annualised revenue rate from AI had climbed 175 per cent to reach $13bn. That is still only about 5 per cent of the total revenue it is expected to produce this year. OpenAI’s revenue run-rate from subscriptions, its main source of income, just topped $10bn, doubling from the end of last year. The rates of increase are notable, but the absolute figures still pale in comparison to the capex.
There have also been signs of an explosion in chatbot use since a year ago, turning OpenAI almost overnight into an unlikely consumer tech company. But while large numbers of people now use AI chatbots, the business remains small. Only about 3 per cent are paying for the AI service they use, producing annual revenue of about $12bn, according to a survey of 5,000 American adults by Menlo Ventures.
Meanwhile, the business world is still searching for uses of generative AI that would justify serious spending. The technology has at least fuelled a new generation of high-growth software start-ups, particularly in coding, the first area of “knowledge work” to be seriously disrupted. But it has yet to cause an inflection in the revenue growth of the biggest software companies, which are best positioned to bring AI to the business world in the form of a new generation of AI-powered apps.
The first wave of AI co-pilots and assistants did little to change working life. The hope has now shifted to agents — tools capable of automating individual tasks, or even entire work processes. According to McKinsey, agents promise to create serious business value by automating complex and important operations. But the consultants also warn this will require a rethink of entire business processes.
Persuading its customers to move beyond the many GenAI pilot projects that litter the corporate world will be a heavy lift for the tech industry. That does not mean that, in the longer run, generative AI has no chance of bringing about the kind of transformation in working life that its boosters claim. But, for now, the chasm between capital spending and revenue has shown little sign of narrowing.