Optimism surrounds artificial intelligence numbers

Optimism surrounds artificial intelligence numbers

As I delve a little deeper into the aggregate artificial intelligence (AI) capital expenditure (capex) numbers and the revenue and profits subsequently required to recoup them, one thing stands out: a world of AI forecasts that are universally optimistic.

That’s not usual in any hype-inspired boom, but these numbers are extraordinary.

Table 1. Gartner’s September 2025 AI spending forecast

Global spending on AI could hit US$1.5 trillion in 2025 alone, with infrastructure sucking up US$500-650 billion, according to Gartner and Deloitte. Those consultants are secretly cheering from the sidelines, of course. McKinsey and PwC have also joined the party, unsurprisingly discovering that AI will add US$13-20 trillion to annual global Gross Domestic Product (GDP) through productivity surges, innovation, and demand multipliers by 2030.

They also say innovation spillovers, new markets, and policy fixes, such as a Universal Basic Income (UBI), will prop up demand ensuring a return for AI’s mega-spenders.

It’s an alluring vision: AI does the grunt work, humanity thrives, and economies soar.

It’s all so smooth, seamless and…obvious!

Transitions are never seamless

The optimistic estimates arise even as 400-800 million jobs (up to 18 per cent of the global working-age population) are expected to be displaced by AI-inspired automation.

Don’t you find it peculiar that even in the most pessimistic employment scenarios – where hundreds of millions of jobseekers are sidelined – these models still churn out optimistic GDP growth?

It’s essential for investors to consider, as the inflated GDP numbers are part of the narrative encouraging companies you invest in to pour billions, nay, trillions, into the AI dream (or is it a sinkhole?).

As Charlie Munger and Warren Buffett often quipped:

Don’t ask a barber whether you need a haircut.

What self-respecting business consultant is going to tell you not to invest, not to spend, not to out-commit your archrival?

To every man with a hammer, the problem is always a nail.

Could AI be nothing more than the latest trend, the newest best hammer that everyone should be using to fix every ailment?

An alternate reality

McKinsey’s own task-based simulations and the Organisation for Economic Co-operation and Development (OECD)’s computable general equilibrium frameworks acknowledge ‘transitional drags’ like inequality and sluggish reskilling. Yet, they insist on net positives (of course!), often relying on assumptions of seamless labour shifts and wage premiums for the AI-savvy who keep their jobs.

PwC justifies its optimism through its Jobs Barometer, which points to historical tech transitions creating more roles than they destroyed. The issue, however, is that current data shows AI-exposed jobs growing unevenly, with low-skill workers bearing the brunt. And what of automation’s negative pressure on wage growth. GDP can only go up in a world of falling wage growth if more volume offsets the stalled price increases. But fewer jobs mean fewer people need, want or can afford more stuff.

The optimistic AI narrative glosses over a flaw: If AI displaces hundreds of millions, who exactly is buying all this innovative “stuff” to fuel the revenue that underpins GDP uplift? Apparently, fewer people will have jobs, but those who do will be able to spend more than enough to compensate for those who can no longer buy anything because they are out of work.

Productivity

The bullish proponents of AI cite ‘productivity efficiencies’ gained through AI adoption.  Efficiencies, however, are usually gleaned from lower costs rather than magical boosts to revenue. To boost revenue, ultimately, consumers – individual people – need to buy more things collectively.  For companies to sell more (and thus drive GDP growth), there must be buyers with disposable income.

Rising unemployment threatens the narrative. Displaced workers, stripped of income, slash spending on everything from gadgets to groceries, generating a collapse in demand, where even cheaper AI-enhanced products gather dust on shelves.

The Institute of Labour Economics (Institut zur Zukunft der Arbeit- IZA), a global research network based in Bonn, Germany, warns that a declining labour share of GDP (down 10-20 per cent in some models) exacerbates inequality, as low-income groups can’t reskill quickly enough, thereby capping consumption.

While B2B sales and global exports might temporarily cushion tech giants like Microsoft or NVIDIA, one needs to remember that their downstream customers rely on consumers with jobs. And consumer-facing firms could be in trouble much sooner if aggregate purchasing power declines. Therefore, Morgan Stanley’s trillion-dollar GenAI revenue dreams by 2028, which rely on enterprise adoption, ignore how widespread joblessness could ripple into B2B slowdowns via reduced business investment.

A voice in the wilderness

Meanwhile, Daron Acemoglu is a MIT Institute Professor and a Nobel laureate in economics, renowned for his work on institutions, growth, and the societal impacts of technology.

Acemoglu’s views challenge the prevailing narrative that AI will deliver massive productivity booms, job creation, and universal prosperity.

Instead, he argues AI’s current trajectory prioritises automation over human augmentation, risking inequality, disempowerment, and modest economic gains at best.

Acemoglu also argues AI has limited transformative potential, and that generative AI, like large language models (LLMs) (e.g., ChatGPT), excel at easy-to-learn tasks but struggle with complex, context-dependent ones. This, he claims will limit broad adoption. That certainly seems to be what MIT’s recent widely-reported study uncovered.

In his 2024 NBER paper “The Simple Macroeconomics of AI”, Acemoglu estimates AI will profitably automate only about five per cent of U.S. labour tasks over the next decade, yielding a “nontrivial but modest” GDP boost of roughly one per cent and a productivity increase of 0.7 per cent.

This contrasts starkly with bullish projections, such as Goldman Sachs’ seven per cent global GDP rise or McKinsey’s US$17-25 trillion annual addition.

Elsewhere, Daron Acemoglu warns that AI’s automation focus risks a “two-tier society” with vast inequalities, particularly for low-educated women in service roles, by centralising power, stifling learning, and undermining democracy. He highlights the adjustment costs resulting from the mismatch between big-tech and small-business needs, debunks the myths of inevitable Artificial General Intelligence (AGI) or societal benefits due to insufficient evidence, and critiques AI’s massive environmental toll and potential for “pervasive manipulation” under deregulation.

For investors, the more immediate concern may be the speculative capital expenditure on a technology that an insufficient number of people are currently purchasing to justify the investment. And having missed out on an early investment in Amazon, Acemoglu is one of the few voices challenging the blind optimism in the forecasts of consultants like McKinsey, where even high job displacement scenarios (400-800 million jobs lost) assume GDP uplifts.

One scenario is AI doesn’t create enough new tasks or markets to sustain demand, potentially leading to revenue shortfalls for companies amid falling labour income shares.

Alternatively, Acemoglu could be right, and AI may not be as transformative, resulting in fewer job losses. Either way, it points to an AI overbuild that will probably end in tears.

With trillions poured into AI infrastructure ‘on spec’, the industry needs rapid recoupment for profitability. And that’s far from guaranteed. Low AI adoption results in poor return on investment (ROIs), as does high adoption, followed by an unemployment-fueled revenue drought. 

Trillions are riding the AI wave, but all waves crash when they meet the reality of the beach.

Read part one here: The artificial intelligence gold rush & bubbles past.

MORE BY RogerINVEST WITH MONTGOMERY

Roger Montgomery is the Founder and Chairman of Montgomery Investment Management. Roger has over three decades of experience in funds management and related activities, including equities analysis, equity and derivatives strategy, trading and stockbroking. Prior to establishing Montgomery, Roger held positions at Ord Minnett Jardine Fleming, BT (Australia) Limited and Merrill Lynch.

He is also author of best-selling investment guide-book for the stock market, Value.able – how to value the best stocks and buy them for less than they are worth.

Roger appears regularly on television and radio, and in the press, including ABC radio and TV, The Australian and Ausbiz. View upcoming media appearances. 

This post was contributed by a representative of Montgomery Investment Management Pty Limited (AFSL No. 354564). The principal purpose of this post is to provide factual information and not provide financial product advice. Additionally, the information provided is not intended to provide any recommendation or opinion about any financial product. Any commentary and statements of opinion however may contain general advice only that is prepared without taking into account your personal objectives, financial circumstances or needs. Because of this, before acting on any of the information provided, you should always consider its appropriateness in light of your personal objectives, financial circumstances and needs and should consider seeking independent advice from a financial advisor if necessary before making any decisions. This post specifically excludes personal advice.