AI spending races ahead of revenue, warns academyEX

academyEX has published a business technology briefing that questions the economics behind the global artificial intelligence race, arguing that the gap between AI spending and revenue is widening.

Founder and Chair Frances Valintine said investor expectations have moved ahead of commercial returns, with large technology groups committing vast sums to infrastructure, models and computing capacity while profitability remains uncertain.

Microsoft, Alphabet, Meta and Amazon are expected to spend about US$725 billion on AI this year, compared with projected revenue of US$80 billion to US$150 billion.

"So there's a huge amount of loss being made across these major companies," Valintine said.

The briefing also highlights the concentration of venture funding in the sector. In the first quarter, 79% of global venture capital went into AI businesses, with more than half of that investment directed to four companies.

Valintine said that level of concentration creates dependence across suppliers of computing infrastructure, model developers and the companies financing them.

"It feels very incestuous, with multiple layers of dependency across compute, infrastructure and model development," she said.

Two models

A central argument in the briefing is that the United States and China are pursuing different economic models for AI. Valintine said the US has largely positioned AI as a premium product controlled by a small number of large technology companies, while China is moving towards a lower-cost model that treats AI and computing more like a utility.

She said those approaches could shape how businesses assess cost, access, privacy and data sovereignty when choosing AI tools and platforms.

"It is really now about two different competing models, AI being treated as a luxury product in the US, versus infrastructure-like models emerging elsewhere," she said.

The briefing links those diverging approaches to wider questions about valuation in public markets. Current pricing, Valintine said, assumes productivity gains that have not yet been realised.

"Financial markets have priced this as though future productivity gains have already been achieved," she said.

Market exposure

The issue extends beyond technology specialists because of AI's weight in broader equity markets. Valintine argued that a correction in the sector would not be confined to a few companies, given how much of the S&P 500 is tied directly or indirectly to AI demand.

"Right now, a significant proportion of the S&P 500 is directly or indirectly tied to AI. If we saw a major correction, the impact would be felt across the global economy," she said.

Her comments reflect a broader debate over whether the current surge in spending on data centres, chips, software and model development can produce returns fast enough to justify market valuations. Concern has grown as companies race to secure computing resources and expand AI services while many customers are still testing where the technology delivers measurable value.

Valintine said the underlying buildout remains at an early stage, leaving a gap between investor expectations and the time needed to bring infrastructure online and generate revenue from it.

Costs rising

The briefing also focuses on the economics of usage. Valintine said the unit cost of AI use may be falling, but total spending is still rising as organisations deploy more systems and begin using autonomous agents and more complex workflows.

"Tokens are significantly cheaper, but the bills are exploding," she said.

She said the increase is tied to the growing use of AI agents, which consume more computing resources than simpler one-off queries.

"It's the agents that are chewing through these tokens," she said.

academyEX described the quarterly briefing as part of its work to help New Zealand business leaders understand the commercial and economic forces shaping technology adoption and the future of work. The school, founded by Valintine, works with professionals and organisations on postgraduate education and leadership development.

Despite the warning on valuations and spending, Valintine said businesses should stay focused on adopting AI in a disciplined way rather than dismissing the technology because of market excess.

"AI will continue to reshape every industry. The challenge for organisations isn't deciding whether AI matters, it's understanding how to invest wisely, adopt it responsibly and prepare for the economic shifts that will come with it," she said.