Does the birth of Agentic AI cause the death of the boom and bust cycle?
Throughout history, there’s been a tidal quality to the flow of capital because there is a tidal quality to the human condition. Every new technology, after welding on its inventors, attracts its devotees who, believing – often correctly – the tech will change the world, become the pied pipers for many more investors. If they’re successful, they generate a groundswell of financial, government, and social support that converge to drive their vision and their wealth, often to heights that natural laws cannot support.
That intersection of innovation and capital is a phenomenon that has long fascinated.
Today, another chapter in the story of that financial and technology tide is being written.
The question is whether artificial intelligence (AI) is a typical bubble in which investor behaviour drives overvaluation. A bubble where the associated hype’s ability to radically lower the cost of capital also inspires a fear of missing out (FOMO). A bubble in which investors stop demanding immediate, safe returns and instead pour billions into speculative ventures that would usually be laughed at. A bubble where even governments succumb to the exuberance, granting easy conditions and even easier funding.
As the companies at the centre of the hysteria take advantage of the cheap capital to scale the technology in a winner-takes-all race for market share, can they generate a sufficient return on capital to justify the market’s valuations?
In her seminal work, Technological Revolutions and Financial Capital, Carlota Perez suggests every major leap in human capability – be it the expansion of the railroads, the proliferation of the telegraph, the advent of radio, or the birth of the internet – follows the same predictable, albeit volatile, trajectory we’ve described here over the years.
It begins with a period of intense excitement, in which companies, spurred by the prospect of a paradigm-shifting ‘next big thing’, engage in a frantic race to overbuild the necessary infrastructure. During this phase, capital expenditure invariably and aggressively outpaces revenue. The over-extension culminates in a spectacular financial crash, bringing prices for attractive assets down to palatable levels, which permit widespread adoption.
Perez’s point is that the bubble and the era of technology’s transformative utility aren’t separate, competing realities; they’re successive stages in a singular and repeating historical process. The bubble is the necessary, if expensive and financially disastrous, mother of the infrastructure required for the consequent ‘golden age’ to transpire.
The scale of AI investment has today reached a level that can only be described as staggering. The sheer volume of capital being diverted into the AI ecosystem is difficult to overstate. The largest institutional players and technology giants are currently funnelling approximately seven hundred billion U.S. dollars annually into the development of specialised semiconductors, massive data centres, and the unprecedented energy infrastructure required to power them.
As The Atlantic’s Derek Thompson noted recently, the spending represents an inflation-adjusted Manhattan Project every three to four weeks, or a full Apollo Space program every five months. This is an unprecedented concentration of private sector resources, a level of spending that has historically been the exclusive domain of sovereign states during times of existential crisis or national mobilisation.
I would add that not only is it unique in scale, but also that it’s almost entirely privately funded.
The hitherto prevailing sentiment that AI represents a classic speculative mania is, however, being challenged, even though the logic is grounded in history: spending is rising at a pace that current revenue models can’t possibly justify, creating a widening gap that typically ends in a sharp, painful correction.
It’s also worth keeping in mind, however, that a vocal contingent of the bears harbour a deep-seated scepticism toward artificial intelligence for philosophical, ethical, or social reasons. For them, the bubble narrative suits their argument but is a form of wishful thinking. They hope for a crash because they want AI to fail, to see the technology proved to be unimportant or transitory, like NFTs or the Metaverse (for which Meta just wrote off $US80 billion).
Figure 1. AI Meme
Source: Instagram
Yet, for the objective and more considered investor, it’s essential to recognise that even if any bubble were to burst, its significance this time would be immense and potentially structural (as opposed to cyclical).
A major correction in the AI sector by 2028 would likely prove more consequential to the global economy, global labour markets, and even the stability of the political landscape than the outcome of the next U.S. presidential election. Today’s magnitude of today’s ‘bubble’ reflects a profound reallocation of global resources.
And, irrespective of whether the capital is lost or the technology prevails, or both(!), the world that emerges on the other side will be fundamentally different from the one that preceded it.
For a long period, the sceptical view was bolstered by the fact that AI, while impressive, seemed limited to generative tasks – producing text, images, or code that still required significant human oversight to be useful. But this year, a fundamental shift is occurring, necessitating some circumspection around the bubble thesis.
The catalyst for this change was the emergence of agentic AI, not only from industry leaders like Anthropic and OpenAI, from one-man-and-a-garage startups like OpenClaw.
Until late last year, the primary interaction with artificial intelligence had been through conversational interfaces – large language models (LLMs) that were adept at dialogue but lacked a clear path to replacing the complexity of multi-stage human processes. The arrival of AI agents capable of working autonomously on intricate projects marks a transition from ‘chatting’ to ‘executing’, if not multi-stage reasoning, which some say these agents are demonstrating. Public demonstrations show agentic AI is now capable of writing complex software, launching entire websites from scratch, and navigating the vast, often impenetrable thickets of government databases to extract and synthesise data without constant human prompting.
What is this shift from generative AI to agentic AI represents a pivotal moment, where the massive infrastructure spend begins to meet a tangible, high-value utility?
The proof of value or utility often lies in eliminating laborious ‘clicking’. I’d love an agent to enter my different logins and passwords every time I’m prompted to perform that mundane and repetitive task that consumes millions of cumulative hours of a professional workforce’s time.
A concrete example of this was seen in early 2026, when experiments involving complex data sets like the American Time Use Survey demonstrated the technology’s new maturity. By asking the AI agent to play the role of a specialised analyst, users found the technology could navigate these data sets to uncover stories and insights that were previously buried under a mountain of manual exertion. This is no longer about the novelty of a machine that can write a poem or draw a picture of a flower; it’s about a functional tool that can perform the structural work of an analyst, an engineer, or an administrator.
The transition to agentic AI might – and I emphasise ‘might’ – suggest that the hundreds of billions spent on chips and power aren’t merely being thrown into a speculative void but are building the foundation for a new kind of industrial and enterprise-level cognition.
In the Perez model, this is the transition from the installation phase to the deployment phase. If the railroads were the overbuilt infrastructure of the 19th century that eventually enabled a century of global commerce, the massive data centres and chip clusters of today are the tracks being laid for an agent-driven economy.
We might just be moving past the era of mere speculation and into a period where the technology demonstrates life-altering utility – skipping the bust altogether.
I reckon the core question for us as investors is no longer just about the extent of the bubble, or the depth of any subsequent crash, but about the depth and durability of the age that follows. Getting that answer right means being prepared to take advantage of the any meltdown because the crash may not be the structural mess many had previously concluded (or hoped) would occur.
Markets may fluctuate, and capital may still be lost in the short term if a shakeout occurs, but the fundamental transformation of how work is executed is now visible on the not-too-distant horizon.
The undeniably astronomical costs of the AI infrastructure build-out might soon be met by a technology that’s finally learning how to work rather than just talk.
What if this is a true technological revolution without the typically associated financial fad? As these autonomous agents become easier to adopt and deploy (last weekend, I had a friend and his son come over to demonstrate OpenClaw) they will become more integrated into the fabric of business processes. Labour-intensive mouse clicking will begin to vanish, and it could be replaced by a level of efficiency that potentially justifies the massive capital outlays we are seeing today.
I still think the financial road will be a volatile one – not everyone will win and the process of adoption will still experience cycles because customers are cyclical (high petrol prices, for example, might mean people can’t afford a subscription to an AI agent). But the rapid emergence of functional, reasoning agents might suggest the golden age predicted by the Perez framework isn’t as distant a hope as previously concluded.
Perhaps more importantly, it’s reasonable to conclude that a lot of money will be made by those who are first to adopt the new opportunities offered by agentic AI.
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.
