Love it or hate it, AI is everywhere. Markets are rallying, companies are investing astronomical amounts into the latest tech, and as a result, financial media can’t go a day without debating whether we’re experiencing the next big tech revolution or the next disastrous bubble. With all the latest noise coming from the AI space, it’s easy to get lost in some of the jargon that gets thrown around.
So in this edition of Bullish Beginnings, we’re bringing the latest instalment of our fluent finance series. This week we’re talking all things AI, breaking down some key phrases and terminology that have been popping up in recent weeks. From Bubbles and Capex to Circular Deals and Depreciation Expense, this blog provides the clear-cut definitions and explanations you need to navigate the latest headlines.
Capex – The AI Arms Race
Capital expenditure, or Capex, refers to the money that companies spend to acquire, upgrade or maintain their physical assets, which could include buildings, machinery and technology. Unlike day-to-day operating expenses, capex is all about investing for the future, on a company’s balance sheet it’s considered a long-term expense which is depreciated over time.
In the AI world, capex has become one of the most talked about metrics in finance. Companies are pouring eye-watering sums into the acquisition and development of new AI technology, such as data centres, advanced computing chips, energy infrastructure, and hardware for training and implementing AI. This extreme level of capex is increasing too, with AI capex being $200 billion in 2024, nearing $400 billion in 2025, and estimated to reach $600 billion in 2027. This investment is being predominantly driven by the Magnificent-7 (comprised of Alphabet, Microsoft, Meta, Nvidia, Amazon, Tesla, and Apple) and is being amplified by these companies rushing to keep up with one another, ensuring they don’t get left behind in the AI arms race.
It’s in this arms race that the risk lies. High capex doesn’t guarantee high returns, and that’s why investors are so focused on it. If companies aren’t able to convert their extreme levels of investment into revenue, they could be stuck with billions of dollars of sunk costs, and nothing meaningful to show for it.
Bubble – When Prices Get Out of Hand
In financial markets, a bubble refers to a situation where asset prices go above and beyond a value that can be justified by traditional fundamentals and valuation methods. Driven by behavioural finance and investor psychology, bubbles are caused by a range of factors, from investor greed to herd mentality. Bubbles typically grow and develop through phases.
Initial Investor Displacement – The initial phase where investors become attracted to something new, this can come in the form of new investment opportunities, innovations, or technology.
Market Boom – Prices begin to rise slowly, however, assets quickly gain momentum as more and more investors pile into the new, desirable assets. Fear of missing out begins to take place, and investors ‘herd’ in the same assets.
Euphoria – Here investor rationality truly falls away. More and more investors pile in, and prices skyrocket, sending valuations to extreme levels. At this stage investors often use new or unique valuations and arguments to justify prices and their investments.
The Bubble Burst – At this stage, a single, or multiple, events can occur which signal assets may be overvalued, swinging investor confidence. Such events could be a key company missing earnings predictions, large quantities of investors taking profit, or a change in the economic landscape. At this stage the bubble is no longer growing, prices may stay at the same level for some time, but warning signs become more prominent.
Panic – Investor sentiment fully reverses. In this phase, assets that have had their prices inflated often fail to, or produce lower than expected returns, revealing their overvaluation. Panic selling begins, and asset prices plunge. This creates a lethal spiral as selling drives down prices, causing further selling and further price drops.
Not all bubbles burst dramatically, some simply deflate as investors slowly pull back their expectations and adjust their portfolios accordingly, leading to prices drifting back towards reality.
Throughout history, famous asset bubbles have had harsh impacts on economies. The 1980s Japanese Economic Bubble came from large levels of stimulus driving up spending in the stock and real estate market, increasing asset prices. The bubble’s imminent burst in 1991 led to a long period of stagnation and deflation in Japan, now known as the Lost Decade.
The Dot-com Bubble was another example of over speculation. In the 1990s, the internet was the latest technology driving excitement among investors. As a result, huge sums of speculative investments were thrown into new internet tech companies as they listed on the stock exchange. The NASDAQ, where the majority of these new firms were listed, shot up from 750 to 5000 in only a decade. Shortly after this, investors caught on to the unsustainable business models and lack of cash flows many of these companies demonstrated. As investors pulled out, the NASDAQ plummeted nearly 80%, and the U.S. was flung into a recession.
The word ‘bubble’ is being thrown around in the current AI-driven market due to its striking resemblance to the Dot-com boom of the early 2000s. Share prices of major AI-exposed companies have surged at a pace far beyond their actual earnings growth, and investors continue to pile in out of fear of missing out on extraordinary future returns. Analysts are now highlighting classic warning signs: skyrocketing capex, revenue projections built on unproven technology, and a market increasingly dependent on a small cluster of mega-cap firms. Whether AI is the next global tech breakthrough or a blow-up, the early signs show a market behaving eerily like past bubbles.
Circular Deals – Money on Repeat
Circular deals refer to situations where companies engage in transactions with one another in ways that can artificially inflate activity, demand, and revenue. Rather than reflecting market growth, these arrangements create the appearance of momentum and demand by essentially cycling money, services, or commitments from one firm to another, all keeping within the same closed loop of companies. These circular deal networks aren’t unique to the AI surge, however, they often appear during times of new innovations and are another early warning sign of a potential asset bubble.
In today’s market, the rise of circular deals is becoming increasingly obvious. Major players such as Microsoft, OpenAI, Nvidia, and AMD are all interconnected through a series of investments, spending agreements, repurchase deals, and revenue-share structures. The image below demonstrates how interconnected and overlapping these deals have become. Frankly, it would take a whole blog post in itself to explain the details behind each relationship, however, the illustration clearly demonstrates how capital is being recycled.
These circular deals aren’t intentionally deceptive or fraudulent, however, they do blur the lines between what is genuine organic demand for new tech and what is essentially synthetic growth being manufactured within the ecosystem itself. Given the scrutiny big tech and AI are already under for bubble-like behaviour, these deal loops make it harder to judge the sustainability of AI growth and make it all the easier to pin current valuations and revenues on a massive bubble.
Depreciation Expense – The Hidden Cost of Innovation
Depreciation expense might not be the most exciting concept in finance, but it has become a staple of market headlines in recent weeks. Depreciation refers to how the cost of a long-term asset is spread over the period it remains useful. When firms spend billions on data centres, chips, and AI hardware, the cost isn’t recorded upfront, it’s allocated gradually across the few years those assets are expected to deliver value.
In the AI boom, this matters more than ever. The massive capex arms race between big tech companies today directly translates into soaring depreciation expenses in the years ahead. As capex compounds year after year, so too does depreciation, steadily eating into profit growth. And because AI hardware is rapidly evolving and becomes outdated quickly, these assets are depreciated over shorter lifespans, meaning the annual expense is even larger.
There is also growing speculation that some big tech firms may be using unrealistically long depreciation schedules to soften the short-term hit to earnings. If these timelines prove too optimistic, companies could be forced into sudden write-downs, recognising the remaining cost of outdated assets all at once. A shock like that could reveal cracks in the economics of the AI boom and make investors question how sustainable the industry’s growth really is.
Wrapping Up
The AI boom has brought excitement, innovation, and no shortage of confusion. Hopefully, with these definitions and explanations you can more confidently navigate the headlines. Capex, depreciation expense, circular deals and bubbles may all seem like separate concepts at first glance, however, taken together they paint a more complete picture of the current landscape. Massive investment is fuelling depreciation expenses, while circular deal flows are inflating the massive growth the AI industry has seen to new levels. All these factors point to the big question, is this all just one big bubble?
What do you think? Is the AI surge a disaster in the making, or is this new tech truly revolutionary and worth it’s lofty price tag?
