The tech giants leading the rush into generative AI are in rude financial health. But strong as their latest earnings performance may look, not much can be directly attributed to generative AI.
Nor is it entirely clear how they will ultimately justify the massive increase in capital spending to build and deliver the technology.
After Alphabet last week, that was the message again as stellar results from Microsoft and Meta lifted tech stocks across the board on Thursday.
At Meta, AI is certainly having an impact, bringing improvements in advertising efficiency and better personalisation of content. But this reflects an earlier investment in a different form of AI, not the generative kind.
READ MORE
Talking to Wall Street, chief executive Mark Zuckerberg was clear that his sights are on the horizon: Meta doesn’t expect generative AI to have much impact on revenue for at least the next couple of years.
That makes the $100 billion (€87.7 billion) or so it is planning to plough into new facilities and equipment in 2026 – close to 50 per cent of its entire expected revenue for the year – a staggering shot in the dark.
Microsoft’s booming spending on AI is on more solid ground. The $30 billion it expects to spend on capex in the current quarter is more than it spent in the whole of 2023. A burgeoning backlog of business and shortage of data centres means much of this new capacity is already spoken for.
But the centrepiece of its latest results – a 34 per cent surge in its Azure cloud platform – owed much to other factors, such as a general move by companies to the cloud, rather than to generative AI. While many customers are piloting the new form of AI in their businesses, few have put it into full-scale use.
If the biggest tech companies are ploughing deeper into generative AI regardless, it is due to two beliefs. One is that the technology is about to reach a level where it is truly useful, and though they may not have all the answers yet about how it will ultimately be used, they are best placed to figure it out.
[ Don’t be fooled: AI is a long way from being able to think for itselfOpens in new window ]
Zuckerberg has described this as the coming of superintelligence, echoing the quasi-mystical belief in some corners of Silicon Valley that AI is about to accelerate past some critical threshold and leave humans in the rear-view mirror.
There is no objective measure for comparing overall machine and human intelligence. But while at risk of inflating the hype, his comments point to a wider belief in the technology reaching a new level of usefulness.
Zuckerberg wrote last week of trying to build a “personal superintelligence” for all Meta users that will help people to “achieve your goals”. The implication: Meta is joining the race to create all-purpose, personalised AI agents to run your life, in competition with companies such as Google and OpenAI.
The vagueness leaves many questions about what form these services will take, and how they will be monetised.
The general message from the tech giants is that this doesn’t matter: they have the audience, they’re creating the infrastructure, they’ll eventually figure it out.
Moving fast, however, will be critical. Companies like OpenAI are making the technology generally available, empowering a new generation of tech start-ups with their own ideas for disruptive AI services.
A second belief is that big tech companies alone have the financial wherewithal to build and deliver generative AI and, at this stage in the game, none of them want to risk getting left behind by underinvesting.
For now, they have been able to take the financial strain. Microsoft’s capital spending has jumped 130 per cent in the past two years to $66.5 billion. But its free cash flow still advanced 20 per cent, and it didn’t need to interrupt the $40 billion or so it pays out each year in dividends and stock buy-backs.
[ McKinsey bars China practice from generative AI work amid geopolitical tensionsOpens in new window ]
With 2026 shaping up to be another big year of investment, the spending will start to weigh more on profits. Wall Street analysts had been anticipating an earnings slowdown for Big Tech. Burgeoning depreciation charges will add to the pressure: Meta and Microsoft cautioned that their latest wave of investment is heavily concentrated in equipment such as servers – short-lived assets that hit the income statement more quickly.
The DeepSeek shock this year was a warning that smaller, cheaper AI models, made widely available, could pose serious competition. Big Tech may be basking in Wall Street’s approval this week, but the urgency to show a payback from generative AI spending will only intensify. – Copyright The Financial Times Limited 2025
