Tech Giants take on $350 billion in debt; the next bubble?

JP Morgan estimates AI giants will take on an extra $200bn in debt before the end of this year. This would account for 20% of all debt in U.S. markets this year - but just how sustainable is this?

AI to today

Whether AI is an asset bubble or not is a highly debated topic among economists and financial professionals. Ultimately, the balance sheets suggest this is an asset bubble; not one like the Dot-com bubble but a new bubble brewed by the fiscal irresponsibility of AI executives.  Despite ChatGPT being one of the most popular platforms, its operator, OpenAI, is not public. In terms of publicly traded stocks, the stocks holding the ‘most AI’ are Nvidia, Broadcom and Taiwan Semiconductor Manufacturing Co, says US News. On average, AI equities can make returns of over 60% a month in some cases and even outperform the S&P 500 by 100% in certain periods.  This huge ROI not only helps to explain why investors are so driven towards AI equities but also nods towards it being an asset bubble. While the data alone does not provide insight into whether a bubble exists or not, it certainly does demonstrate the increased popularity of these stocks, which may align with historic bubbles. Take Nvidia as an example, as seen below:

Nvidia is a big player in the AI market in terms of firms that have had an IPO. There has been a huge increase in both share price and demand since 2024, with a 5-year growth rate of 1058%.  A good historical example of a technological bubble is the dot-com bubble in 1995-2000. During this time, Nvidia was also an emerging stock, and its price then can be shown below:

The bubble can be seen around 2001 on the graph - following a large price rise since the IPO, Nvidia shows a similar trend today. Only after this was there a large crash, plateauing in 2002. While past performance does not guarantee another 2001, the structural parallels between Nvidia’s current trajectory and the Dot-com bubble are simply impossible to ignore. To see the full effect of this potential bubble, it's pivotal to look beyond stock market performance and dive into the balance sheets - it's these figures that pose the most concern.

The anatomy of an AI bubble

Historically, a bubble has been caused by a revolutionary new technology, the most famous being the Dot-com bubble in the 1990s or the rail boom in the 1840s. AI, just like the rest, is a huge development in technology that is hugely inexpensive and accessible to consumers - even more so than the introduction of the internet or rail, given the context of the time. Extreme stock valuations can be evidenced by the Nvidia example above; furthermore, the venture capital sector is eating up AI startups, with over 50% of global venture capital spending being in the AI industry. This shows a flood into the AI scene by investors, which highlights a deeper network of speculation towards AI potential, thus driving up the share price - articulating the first sign of a bubble.

It's this spending and investment that people should be concerned about. In terms of concentration ratios within the industry, the higher the concentration, the more volatility the market poses, and the more direct investment into a small number of firms. This is true with the dominance of the ‘AI Big 10’ (including Amazon, Microsoft, Nvidia, Alphabet) holding about 40% of the S&P 500 market value. This overwhelming concentration adds a high level of risk to the balance sheets; this leaves the stock market highly exposed if tech giants were to fall. Furthermore, the debt-to-revenue ratio for firms is beginning to serve as a warning sign. For example, five of the biggest data centre builders - Alphabet, Amazon, Meta, Microsoft, and Oracle have collectively added $350 billion to their debt obligations over the past five years.

Despite this huge debt in Q1 2026, OpenAI lost $7bn, meaning it cost the company $2.22 to generate $1.0 in revenue. Of course, OpenAI is not publicly traded; however, this still demonstrates how firm debt is being expanded massively in accordance with large investments in the stock market. By almost every historical metric- valuation, concentration and rising leverage - AI is behaving like a brewing asset bubble.

But what if?

This is obviously a highly debated topic, and there is evidence to suggest this is not a bubble. The most prominent point is profit. During the Dot-com bubble, firms with very low revenue saw their valuations double overnight. However, the ‘AI top 10’ are among the most profitable firms in the world, with Alphabet the most profitable. This essentially argues that these firms have enough retained profit to reinvest sustainably. However, this is just not true for all firms; for example, Oracle, a large firm responsible for AI cloud infrastructure, is currently spending 1.47x more than its operating cash flow. This is a highly concerning figure alongside its debt-to-equity ratio of 333%. This just accentuates the point that this is structurally volatile when the cloud responsible for big AI is so low on cash and relies on huge debts to fund its leases for data centres that are not even built yet. In addition, AI maintains P/E ratios of 30x-40x; on the contrary, during the Dot-com bubble were sometimes as high as 100x. The real threat to the development of the AI bubble is not the speculation of rising share prices; it's the underlying capital expenditure. Companies are investing such immense amounts of capital into building data centres without even being able to make a profit - in fact, their losses are growing; for example, in 2025 OpenAI lost £5bn, almost an eight-fold increase over 2024.

Ultimately, AI is showing the right signs, the right debts, the right equity price rises. AI is going to drive gargantuan structural change in the global macro and microeconomy - it really is here to stay. However, the financial scaffolding around this innovation is simply not sustainable.  People won’t understand AI is a bubble if they keep suggesting it will be caused by consumer-led investing like 2001. This is instead a bubble that will be derived as a consequence of fiscal irresponsibility from AI bosses. AI most certainly is a bubble, a slow-burning one and the fuse is already lit.